Regulatory Impact Analysis of the
Final Clean Air Mercury Rule
-------
EPA-452/R-05-003
March 2005
Regualtory Impact Analysis of the
Final Clean Air Mercury Rule
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Strategies and Standards Division
Innovative Strategies and Economics Group (MD 339-01)
Research Triangle Park, N.C. 27711
§
CVJ
o:
a.
-------
CONTENTS
ACRONYMS AND ABBREVIATIONS xxii
SECTION 1 INTRODUCTION M.
1.0 Introduction 1-1
SECTION 2 IMPACT OF MERCURY ON HUMAN HEALTH, ECOSYSTEMS, AND
WILDLIFE 2J.
2.1 Introduction 2-1
2.2 Mercury Poisoning Episodes 2-1
2.3 Reference and Benchmark Doses 2-2
2.4 Neurologic Effects 2-6
2.5 Cardiovascular Impacts 2-7
2.6 Genotoxic Effects 2-7
2.7 Immunotoxic Effects 2-7
2.8 Other Human Toxicity Data 2-8
2.9 Ecological Effects 2-9
2.10 Conclusions 2-9
2.11 References 2-10
SECTION 3 ECOSYSTEM SCALE MODELING FOR MERCURY BENEFITS
ANALYSIS M
Executive Summary 3-1
3.1 Introduction — Rule Background 3-4
3.1.1 Use of Mercury Maps (MMaps) to Project Changes in Fish Tissue
Concentrations 3-5
3.1.2 Goal/Purpose of Ecosystem Case Studies 3-9
3.2 Recent Advances in Mercury Science 3-10
3.2.1 Mercury Cycle Chemistry 3-10
3.2.2 Mercury Processes in the Atmosphere 3-10
3.2.3 Mercury Processes in Soils 3-11
3.2.4 Mercury Processes in Water 3-12
3.2.5 Bioavailability of Inorganic Mercury to Methylating Microbes 3-12
3.2.6 Mercury Accumulation in the Food Web 3-14
3.2.7 Summary of Findings in the METAALICUS Study 3-14
3.2.8 Summary of Florida Everglades Study 3-15
3.3 Overview of Models Used in This Study 3-16
3.3.1 Atmospheric Models 3-16
3.3.2 Ecosystem Models 3-17
3.4 Overview of Case Studies 3-21
3.4.1 Ecosystem Characteristics 3-22
3.4.2 Baseline Atmospheric Deposition at Each Site 3-24
3.4.3 Atmospheric Loading Scenarios Investigated 3-24
3.4.4 Summary of Model Evaluation 3-25
3.4.5 Baseline Fish Mercury Concentrations 3-27
in
-------
3.4.6 Magnitude of Changes in Fish Tissue Residues 3-30
3.4.7 Summary of Observed Temporal Responses to Declines in Loading . 3-32
3.4.8 Effect of Land Uses Changes 3-34
3.4.9 Summary 3-36
3.5 National Scale Ecosystem Variability 3-37
3.5.1 United States Lakes Distribution 3-37
3.5.2 Summary 3-39
3.6 References 3-46
SECTION 4 PROFILE OF FISHING ACTIVITY IN THE U.S 4-1.
4.1 Industry Characterization 4-1
4.1.1 Introduction and Overview of the Fishing Industry 4-1
4.1.2 Commercial Fishing 4-2
4.1.3 Recreational Fishing 4-6
4.2 U.S. Production Statistics for Commercial and Recreational Fishing 4-8
4.2.1 Commercial Fishing 4-8
4.2.2 Recreational Fishing 4-15
4.3 U.S. Demand for Commercial and Recreational Fishing 4-19
4.3.1 Commercial Imports 4-19
4.3.2 U.S. Demand for Commercial Fishery Products 4-24
4.3.3 Recreational Fishing 4-26
4.3.4 Total U.S. Demand 4-35
4.4 Economic Value of Key Species 4-36
4.4.1 Finfish 4-36
4.4.2 Shellfish 4-37
4.4.3 Fish Products 4-39
4.5 Characterization of Fish Consuming Populations 4-39
4.5.1 Fish Consumption Pathways and Associated Fish-Consuming
Populations 4-40
4.5.2 Fish Consuming Populations 4-41
4.5.3 General Fish Consumption Rates for Key Fish Consuming
Populations 4-42
4.5.4 Discussion of Population and Fish Consumption Data in the Context of the
Mercury Benefits Analysis 4-45
4.6 Summary 4-47
4.6.1 Commercial Fish Production, Demand, and Consumption 4-47
4.6.2 Recreational Fishing Activity, and Consumption 4-48
4.6.3 Overall Conclusions 4-49
4.7 References 4-49
SECTION 5 MERCURY CONCENTRATIONS IN FISH 5J.
5.1 Methylmercury Concentrations in Saltwater Fish Species 5-1
5.2 Methylmercury in Freshwater Fish Species 5-2
5.2.1 Sources of Variability in Hg within the NLFA 5-4
5.2.2 National Lake Fish Tissue Study 5-6
5.3 Comparison of the Differences in the NLFA and the NLFTS 5-7
5.4 Combining the NLFA and NLFTS Data 5-9
iv
-------
5.5 Normalization of Hg Fish Tissue Concentration Data 5-10
5.6 Resulting Fish Tissue Concentrations 5-13
5.7 Summary 5-17
5.8 References 5-17
SECTION 6 PROFILE OF THE UTILITY SECTOR 6J.
6.1 Power-Sector Overview 6-1
6.1.1 Generation 6-1
6.1.2 Transmission 6-3
6.1.3 Distribution 6-3
6.2 Deregulation and Restructuring 6-3
6.3 Pollution and EPA Regulation of Emissions 6-4
6.4 Pollution Control Technologies 6-5
6.5 Regulation of the Power Sector 6-6
6.6 Cap and Trade 6-7
6.7 Clean Air Interstate Rule 6-8
SECTION 7 COST AND ENERGY IMPACTS M
7.1 Modeling Background 7-1
7.2 Projected Hg Emissions 7-3
7.3 Projected SO2 and NOX Emissions 7-5
7.4 Projected Costs 7-5
7.5 Projected Control Technology Retrofits 7-7
7.6 Projected Generation Mix 7-8
7.7 Projected Capacity Additions 7-9
7.8 Projected Coal Production for the Electric Power Sector 7-9
7.9 Projected Retail Electricity Prices 7-10
7.10 Projected Fuel Price Impacts 7-12
7.11 Social Cost Calculations 7-12
7.12 Limitations of Analysis 7-13
7.13 Significant Energy Impact 7-17
7.14 Sensitivity Analysis on Assumptions for Hg Control Costs 7-17
7.15 Sensitivity Analysis on Assumptions for Natural Gas Prices and Electricity
Growth 7-22
7.16 Small Entity Impacts 7-27
7.16.1. Identification of Small Entities 7-29
7.16.2 Overview of Analysis and Results 7-30
7.16.3 Summary of Small Entity Impacts 7-35
7.17 Unfunded Mandates Reform Act (UMRA) Analysis 7-36
7.17.1 Identification of Government-Owned Entities 7-37
7.17.2 Overview of Analysis and Results 7-37
7.17.3 Summary of Government Entity Impacts 7-42
7.18 List of IPM Runs in Support of CAMR 7-43
-------
SECTION 8 AIR QUALITY MODELING: CHANGES IN HG DEPOSITION TO U.S.
WATERBODIES 8-1
8.1 Emissions Inventories and Estimated Emissions Reductions 8-2
8.2 Model, Domain, Configuration, Inputs, Application 8-5
8.2.1 Air Quality Model 8-5
8.2.2 Modeling Domain 8-6
8.2.3 Time Periods Modeled for Mercury Deposition 8-6
8.2.4 Model Inputs 8-7
8.3 CMAQ Model Performance Evaluation 8-8
8.4 Mercury Deposition Results 8-9
8.5 Summary of Findings: HUC Level Deposition Analysis 8-14
8.6 References 8-17
SECTION 9 ANALYSIS OF THE DOSE-RESPONSE RELATIONSHIP BETWEEN
MATERNAL MERCURY BODY BURDEN AND CHILDHOOD IQ 9-1
9.1 Introduction 9-1
9.2 Epidemiological Studies of Mercury and Neurodevelopmental Effects 9-2
9.3 Statistical Analysis 9-4
9.4 Strengths and Limitations of the IQ Dose-Response Analysis 9-8
9.5 References 9-11
SECTION 10 EXPOSURE MODELLING AND BENEFIT METHODOLOGY WITH AN
APPLICATION TO A NO-THRESHOLD MODEL 10-1
10.1 Introduction 10-1
10.1.1 Summary 10-2
10.1.2 Modeling Overview 10-6
10.1.3 Monetized Benefits: Results in Brief 10-8
10.1.4 Key Steps 10-11
10.2 Estimation of Mercury Levels in Freshwater Fish 10-12
10.3 Estimation of Exposed Populations and Fishing Behaviors 10-18
10.3.1 Primary Data Sources on Fishing Activity in the United States .... 10-18
10.3.2 Population Centroid Approach 10-24
10.3.3 Angler Destination Approach 10-36
10.4 Estimation of Mercury Exposures, IQ Decrements, and Lost Future Earnings
10-42
10.4.1 Modeling Approach for Estimating Individual Exposures 10-43
10.4.2 Modeling Approach for Estimating IQ Effects and Lost Earnings .. 10-45
10.5 Model Results: Estimated Benefits of Utility Mercury Emission Controls . 10-47
10.5.1 Results for the Population Centroid Approach 10-53
10.5.2 Results for the Angler Destination Approach 10-67
10.5.3 Comparison of Results from Two Approaches 10-89
10.5.4 Sensitivity Analysis of Alternative Dose-Response Functions 10-96
10.5.5 Distribution of Per-Capita IQ Changes for the Exposed Population (in
support of distributional equity analysis) 10-97
10.6 Analysis of Potentially High-Risk Subpopulations 10-103
10.6.1 Mercury Ingestion Estimates for Individuals in the Upper Range of the
Fish Consumption Distribution 10-104
vi
-------
10.6.2 Mercury Ingestion Estimates for Individuals in Low Income, High Fish
Consumption Households 10-110
10.6.3 Mercury Ingestion Estimates for Two Selected Ethnic Populations
10-112
10.6.4 Adaptation of the Population Centroid Approach to Estimate Exposed
Hmong and Chippewa Population 10-119
10.6.5 Sensitivity Analysis Examining the Economic Benefit Equity Issue in the
Context of High Fish Consuming (subsistence) Populations Including
Native Americans 10-129
10.7 Discussion and Qualification of Results: Assumptions, Limitations, and
Uncertainties 10-134
10.7.1 Mercury Concentration Estimates 10-135
10.7.2 Exposed Population Estimates 10-137
10.7.3 Matching of Exposed Populations to Mercury Concentrations .... 10-138
10.7.4 Fish Consumption Estimates 10-140
10.7.5 Modelling and Valuation of IQ Related Effects 10-141
10.7.6 Unqualified Benefits 10-142
10.8 References 10-144
SECTION 11 BENEFITS OF MERCURY REDUCTION CONSIDERING ESTABLISHED
HEALTH-BASED BENCHMARKS AND OVERALL BENEFITS
CONCLUSIONS 11-1
11.1 Introduction 11-1
11.2 The Mercury - IQ-loss Paradigm 11-1
11.3 Quantifying IQ Benefits Associated with Mercury Emission Reductions ... 11-3
11.4 Data Element (3) -The Level of the Threshold 11-4
11.5 Date Element (4) - The Baseline Levels of Exposure for Consumers of
Recreational-Caught Fish 11-4
11.6 Overview of Benefits Methodology 11-5
11.7 Freshwater Fish Mercury Exposure 11-5
11.8 Deriving Baseline Mercury Exposures from All Sources of Mercury 11-8
11.9 Deriving Scaling Factors 11-12
11.10 Monetization and Scaling of IQ Benefits 11-13
11.11 Uncertainties 11-15
11.12 Conclusions 11-15
11.13 References 11-16
SECTION 12 CO-BENEFITS RESULTING FROM REDUCTIONS IN EMISSIONS
OF PM2.5 12-1
12.1 Introduction 12-1
12.2 Emissions Modeling 12-3
12.3 Air Quality Modeling and Population-Level Exposure Estimation 12-4
12.4 Modeling Changes in Health Endpoint (Mortality) Incidence 12-5
12.5 Valuation of Benefits 12-7
12.6 Presentation of Results 12-8
12.7 Discussion of Uncertainties 12-8
12.8 References 12-11
vii
-------
APPENDIX A-l MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR
EAGLE BUTTE (SOUTH DAKOTA) Al-1
Al. 1 Introduction Al-1
A 1.2 Site Characteristics and Model Parameterization Al-4
Al .3 SERAFM Simulations of Eagle Butte, Lee Dam Al-5
A1.4 WASP7 Simulations of Eagle Butte, Lee Dam A1-6
A1.5 BASS Model Simulations of Methylmercury and Fish Dynamics in Lee Dam,
Eagle Butte, SD—Response to Changes in Mercury Loading Al-13
APPENDIX A-2 MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR
PAWTUCKAWAY LAKE (NEW HAMPSHIRE) A2-1
A2.1 Introduction A2-1
A2.2 SERAFM Application A2-2
A2.3 WASP Model Calibration A2-3
A2.4 References A2-10
APPENDIX A-3 MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR
LAKE WACCAMAW (NORTH CAROLINA) A3-1
A3.1 Introduction A3-1
A3.2 Empirical Data from Lake Waccamaw A3-2
A3.3 SERAFM Application: Lake Waccamaw A3-4
A3.4 Lake Waccamaw WASP Model Calibration A3-6
A3.5 References A3-13
APPENDIX A-4 MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR THE
BRIER CREEK WATERSHED (LOCATED IN THE SAVANNAH RIVER BASIN,
GEORGIA) A4-1
A4.1 Background A4-1
A4.2 Mercury Deposition Network A4-3
A4.3 Watershed Hydrologic and Sediment Loading Model A4-4
A4.4 Water Quality Fate and Transport Model A4-5
A4.5 Model Results A4-6
A4.5.1 Water Quality Model A4-6
A4.6 Brier Creek Watershed Results A4-8
A4.6.1 Brier Creek Soil Mercury Calibration A4-9
A4.6.2 Mercury Loading Fluxes A4-10
A4.6.3 Future Projections A4-11
A4.6.4 Sensitivity of Temporal Response A4-13
A4.7 Brier Creek Water Body Results A4-16
A4.7.1 Phase 1: Long Term Buildup A4-16
A4.7.2 Phase 2: Response to 2002 Flows A4-17
A4.7.3 Phase 3: Future Attenuation A4-20
A4.7.4 Sensitivity of Time Response A4-22
APPENDIX A-5 MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR
LAKE BARCO (FLORIDA) A5-1
viii
-------
A5.1 Introduction A5-1
A5.2 Empirical Data from Lake Barco A5-1
A5.3 SERAFM Application: Lake Barco A5-1
A5.4 References A5-3
APPENDIX B QUALITATIVE ECOLOGICAL REVIEW OF MERCURY LITERATURE B-l
B.I Introduction B-l
B.2 Potential Exposure Media B-2
B.2.1 Mercury in Air B-2
B.2.2 Mercury in Water B-2
B.2.3 Mercury in Soil B-2
B.3 Bioaccumulation of Mercury B-3
B.4 Exposure and Toxic Effects in Wildlife B-4
B.4.1 Aquatic Plant Species B-4
B.4.2 Aquatic Invertebrate Species B-4
B.4.3 Fish and Amphibian Species B-6
B.4.4 Terrestrial Plant Species B-8
B.4.5 Terrestrial Invertebrate Species B-8
B.4.6 Avian Species B-9
B.4.7 Mammalian Species B-ll
B.5 Ecosystems Potentially Affected B-13
B.6 Conclusions B-13
B.7 References B-14
APPENDIX C CARDIOVASCULAR EFFECTS AND METHYLMERCURY C-l
C.I Introduction C-1
C.2 Acute Myocardial Infarctions and Major Cardiovascular Effects C-l
C.2.1 The Kuopio Ischemic Heart Disease Risk Factor Study (KIHD) Cohort
C-2
C.2.2 The European Multicenter Case Control Study on Antioxidants,
Myocardial Infarction and Cancer of the Breast (EURAMIC) Cohort
C-4
C.2.3 Mechanisms for Cardiovascular Impacts C-4
C.2.4 Other Studies Evaluating CVD and Mercury Levels C-5
C.3 Other Cardiovascular Effects C-7
C.4 Cardiovascular Health Benefits of Fish Consumption C-8
C.5 Conclusions C-9
C.6 References C-10
APPENDIX D NORMALIZATION OF MERCURY IN FISH TISSUE SAMPLES D-l
D.I Methods D-l
D.I.I National Descriptive Model of Mercury in Fish (NDMMF) D-l
D.2 General Examination of Model Performance D-2
D.2.1 NDMMF Estimated Values D-2
D.2.2 Accuracy of NDMMF Estimated Values D-3
D.2.3 Spatial Examination of Model Performance D-5
D.2.4 Predictive Examination of Model Performance (Withheld Data Set) . D-6
ix
-------
D.3 References D-9
APPENDIX E-1 ANALYSIS OF TRIP TRAVEL DISTANCE FOR FRESHWATER
ANGLERS El-1
El.l Data El-1
El.2 Analysis of Travel Distance Data El-2
El.3 Summary Results Applied in the Population Centroid Approach El-4
APPENDIX E-2 METHODOLOGY FOR ESTIMATING FRESHWATER FISHING
DAYS BY WATERSHED E2-1
E-2.1 Data E2-1
Tables
Table 3- 1. Comparison of SERAFM and IEM-2M Forecasted Mercury Concentrations Using
Parameter Values for Model Ecosystem Described in the Mercury Study Report to
Congress (RtC) and a 50% Reduction in Atmospheric Deposition 3-19
Table 3- 2. Summary of Ecosystem Characteristics Used To Parameterize Mercury Models 3-23
Table 3- 3. Baseline Atmospheric Deposition For Each Model Ecosystem 3-24
Table 3- 4. Forecasted Atmospheric Deposition Rates in Case Study Areas Using the CMAQ
and REMSAD Models 3-25
Table 3- 5. List of Model Frameworks Applied to Ecosystems 3-26
Table 3-6. Summary of Mercury Parameters Used in the SERAFM Model 3-27
Table 3- 7. Empirically Derived BAFs for Each of the Ecosystem Case Studies 3-27
Table 3- 8. MMaps and SERAFM Forecasted Fish Mercury Concentration at Steady State after
Removal of Coal Fired Utilities as a Component of Deposition Using the REMSAD and
CMAQ Models (Zero-out Scenario) 3-31
Table 3- 9. Sediment Response Times in Years to Reach 90% of Steady-state Concentrations
Following 50% Mercury Deposition Reductions 3-32
Table 3- 10. Fish Tissue Response Times in Years to Reach 90% of Steady-state Concentrations
Following 50% Mercury Deposition Reductions 3-33
Table 3-11. Frequency of Different Lake Sizes Across the United States 3-38
Table 4-1. Finfish Fishing Industry (NAICS code 114111) 4-4
Table 4-2. Shellfish Fishing Industry (NAICS code 114112) 4-4
Table 4-3. Value of Aquacultural Products Sold, 1998 4-5
Table 4-4. Number of Anglers and Fishing Licenses in the United States 4-7
Table 4-5. Annual Domestic Landings for Commercial Fishing (Finfish and Shellfish) .... 4-8
Table 4-6. Commercial Fish Landings by Region 4-10
Table 4-7. 2002 U.S. Commercial Fish Landings by End Use 4-10
Table 4-8. 2002 U.S. Commercial Landings by Month 4-11
Table 4-9. Finfish Species with the Highest 2002 U.S. Commercial Landings 4-11
Table 4-10. Shellfish Types with the Highest 2002 Commercial Landings 4-12
Table 4-11. Total 2002 U.S. Aquacultural Production 4-13
Table 4-12. 2002 Exports of Edible Finfish Products 4-14
Table 4-13. 2002 Exports of Edible Shellfish Products 4-15
Table 4-14. 2002 Recreational Marine Landings for Selected Finfish Types 4-16
Table 4-15. 2002 Recreational Marine Catch and Harvest by Region and Top Species
Group 4-18
-------
Table 4-16. 2002 Imports of Edible Fresh or Frozen Finfish Products 4-22
Table 4-17. 2002 Imports of Edible Canned and Cured Finfish Products 4-23
Table 4-18. 2002 Imports of Edible Fresh and Frozen Shellfish Products 4-24
Table 4-19. 2002 Imports of Edible Canned Shellfish Products 4-24
Table 4-20. 2002 U.S. Demand for Commercial Finfish (metric tons) 4-25
Table 4-21. 2002 U.S. Demand for Commercial Shellfish (metric tons) 4-25
Table 4-22. 2002 U.S. Consumption of Commercial Fishery Products 4-26
Table 4-23. Number of Anglers and Days of Fishing for 2001 4-27
Table 4-24. 2001 U.S. Recreational Freshwater Fishing: Targeted Species by Region 4-28
Table 4-25. Freshwater Fishing Bag and Size Limits for Selected States and Species 4-31
Table 4-26. Saltwater Fishing Bag and Size Limits for Selected States and Species 4-32
Table 4-27. Percent of Freshwater Anglers by Age Group 4-33
Table 4-28. Percent of Anglers by Sex and Age Group 4-33
Table 4-29. 2001 Demographic Summary, Angler Race (% Non-White) 4-34
Table 4-30. Incidence of Fishing Among White, African American, and Hispanic Recreational
Boaters 4-34
Table 4-31. Percent of U.S. Population Who Fished (By Household Income) 4-35
Table 4-32. 2002 Economic Value of Commercial Landings for Important Finfish Types .. 4-36
Table 4-33. Average 2001 Wholesale Prices for Several Fresh Finfish Types 4-37
Table 4-34. 2002 Economic Value of Commercial Landings of Several Shellfish Types ... 4-38
Table 4-35. Average 2001 Wholesale Prices for Several Fresh Shellfish Types 4-39
Table 4-36. 2002 Economic Value of Commercial Fishery Products 4-39
Table 4-37. Demographic (count) Data for Key Fish Consuming Populations in the U.S. .. 4-43
Table 4-38 Fish Consumption Rates for Key Fish Consuming Populations in the U.S 4-44
Table 4-39. Total Fish Consumption for Recreational Saltwater Anglers, Recreational
Freshwater Anglers and General U.S. Fish Consuming Population 4-46
Table 5-1. Concentrations of Mercury in Marine Life 5-2
Table 5-2. Number of Fish Tissue Samples / Watershed From the National Listing of Fish
Advisories 5-3
Table 5-3. Hg Fish Tissue Concentrations From Various Environments (ppm) 5-5
Table 5-4. Statistical Distribution of Normalized Hg Fish Tissue Concentrations 5-12
Table 5-5. Statistical Distribution of Non-Normalized Hg Fish Tissue Concentrations Shown in
Figure 5-5 5-13
Table 6-1. Existing Electricity Generating Capacity by Energy Source, 2002 6-1
Table 6-2. Total U.S. Electric Power Industry Retail Sales in 2003 (Billion kWh) 6-2
Table 6-3. Electricity Net Generation in 2003 6-2
Table 6-4. Emissions of SO2 and NOX in 2003 and Percentage of Emissions in the CAIR
Affected Region (tons) 6-8
Table 6-5. Current Electricity Net Generation and EPA Projections for 2010 and 2015 6-9
Table 7-1. CAMR Options Annual Emissions Caps (Tons) 7-1
Table 7-2. CAIR Annual Emissions Caps (Million Tons) 7-1
Table 7-3. Projected Emissions of Hg with the Base Casea (No Further Controls), with CAIR,
and with CAMR (Tons) 7-4
Table 7-4. Projected Speciated Emissions of Hg in 2020 with CAIR and CAMR (Tons) 7-4
Table 7-5. Projected Emissions of SO2 with the Base Case3 (No Further Controls), with CAIR,
and with CAMR (Million Tons) 7-5
XI
-------
Table 7-6. Projected Emissions of NOX with the Base Case" (No Further Controls), with CAIR,
and with CAMR (Million Tons) 7-5
Table 7-7. Annualized National Cost and Present Value Cost Incremental to CAIR ($1999) . 7-6
Table 7-8. Marginal Cost of Hg, S02, and NOX Reductions with CAMR ($1999) 7-6
Table 7-9. Pollution Controls by Technology with the Base Case (No Further Controls), with
CAIR, and with CAMR (GW) 7-7
Table 7-10. Generation Mix with the Base Case (No Further Controls), with CAIR, and with
CAMR (Thousand GWhs) 7-8
Table 7-11. Total Coal and Natural Oil/Gas-Fired Capacity by 2020 (GW) 7-9
Table 7-12. Coal Production for the Electric Power Sector with the Base Case (No Further
Controls), with CAIR , and with CAMR (Million Tons) 7-9
Table 7-13. Projected National Retail Electricity Prices with the Base Case (No Further
Controls) and with CAIR (Mills/kWh) ($1999) 7-10
Table 7-14. Retail Electricity Prices by NERC Region with the Base Case (No Further
Controls), with CAIR, and with CAMR (Mills/kWh) ($1999) 7-11
Table 7-15. Henry Hub Natural Gas Prices and Average Minemouth Coal Prices with the Base
Case (No Further Controls), with CAIR, and with CAMR (1999$/mmBtu) 7-12
Table 7-16. Projected Emissions of Hg with CAIR and CAMR (Tons) 7-18
Table 7-17. Projected Emissions of SO2 with CAIR and CAMR (Million Tons) 7-18
Table 7-18. Projected Emissions ofNOx with CAIR and CAMR (Million Tons) 7-18
Table 7-19. Annualized Cost and Present Value Cost Incremental to CAIR ($1999) 7-19
Table 7-20. Marginal Cost of Hg, SO2, and NOX Reductions with CAMR ($1999) 7-19
Table 7-21. Pollution Controls by Technology with CAIR and with CAMR (GW) 7-19
Table 7-22. Generation Mix with the Base Case (No Further Controls), with CAIR, and with
CAMR (Thousand GWhs) 7-20
Table 7-23. Total Coal and Natural Oil/Gas-Fired Capacity by 2020 (GW) 7-20
Table 7-24. Coal Production for the Electric Power Sector with the Base Case (No Further
Controls), with CAIR, and with CAMR (Million Tons) 7-21
Table 7-25. Retail Electricity Prices by NERC Region with the Base Case (No Further
Controls), with CAIR, and with CAMR (Mills/kWh) ($1999) 7-22
Table 7-26. Projected Emissions of Hg for CAIR and CAMR with EPA and EIA Assumptions
for Natural Gas Prices and Electric Growth (Tons) 7-23
Table 7-27. Projected Nationwide Emissions of SO2 and NOX under CAIR and CAMR with EPA
and EIA Assumptions for Natural Gas and Electric Growth (Million Tons) 7-23
Table 7-28. Annualized Cost and Present Value Cost Incremental to CAIR with EPA and EIA
Assumptions for Natural Gas Prices and Electric Growth (Billion $1999) 7-24
Table 7-29. Marginal Cost of SO2 and NOX Reductions under CAIR and CAMR with EPA and
EIA Assumptions for Natural Gas Prices and Electric Growth ($/ton, in $1999) 7-24
Table 7-30. Pollution Controls under CAIR with EPA and EIA Assumptions for Natural Gas
and Electricity Growth (GWs) 7-24
Table 7-31. Generation Mix under CAIR and CAMR with EPA and EIA Assumptions for
Natural Gas and Electric Growth (Thousand GWhs) 7-25
Table 7-32. Coal Production for the Electric Power Sector under CAIR and CAMR with EPA
and EIA Assumptions for Natural Gas and Electricity Growth (Million Tons) 7-26
Table 7-33. Retail Electricity Prices by NERC Region for the Base Case (No Further Controls),
CAIR, and CAMR with EPA and EIA Assumptions for Natural Gas and Electricity
Growth (Mills/kWh) ($1999) 7-27
xii
-------
Table 7-34. Potentially Regulated Categories and Entities3 7-28
Table 7-35. Projected Impact of CAMR on Small Entities 7-30
Table 7-36. Summary of Distribution of Economic Impacts of CAIR on Small Entities ... 7-34
Table 7-37. Incremental Annualized Costs under CAMR relative to CAIR, Summarized by
Ownership Group and Cost Category ($1,000,000) 7-35
Table 7-38. Summary of Potential Impacts on Government Entities under CAIR 7-37
Table 7-39. Distribution of Economic Impacts on Government Entities under CAMR .... 7-41
Table 7-40. Incremental Annualized Costs under CAMR Relative to CAIR Summarized by
Ownership Group and Cost Category ($1,000,000) 7-42
Table 7-41. Listing of Runs from the Integrated Planning Model Used in Analyses Done in
Support of the CAMR Final Rule Analyses 7-44
Table 8-1. Summary of Emissions Sources for 2001 and 2020 Mercury Emissions
Inventories 8-3
Table 8-2. Summary of Mercury Emissions by Species: 2001 and 2020 (with CAIR)
Baselines 8-3
Table 8-3. Summary of Changes in Mercury Emissions Associated with CAMR Control
Option 1: 2020 8-4
Table 8-4. Summary of Changes in Mercury Emissions Associated with CAMR Control
Option 2: 2020 8-4
Table 8-5. CMAQ Performance Statistics for Mercury Wet Deposition: 2001 8-9
Table 8-6. Summary Statistics of Total Mercury Depositions (ug/m2) by Modeling
Scenario 8-14
Table 8-7. Summary Statistics of Utility Attributable Deposition (ug/m2) by Modeling
Scenario 8-16
Table 9-1. Neurobehavioral Tests Administered at the 6-Year Evaluations in the
New Zealand Study 9-3
Table 9-2. Neurobehavioral Tests Administered at the 7-Year Evaluations in the Faroe Islands
Study 9-3
Table 9-3. Neurobehavioral Tests Administered at the 9-Year Evaluations in the Seychelles
Islands Study 9-4
Table 9-4. Relationship Between Maternal Mercury Body Burden and IQ in Three Studies: IQ
Decrement per ppm of Maternal Hair Mercury 9-6
Table 10-1 (a). Summary of Per Capita Changes in IQ Due to Mercury Exposure 10-4
Table 10-l(b). Impacts of Mercury on High Fish Consuming Groups 10-6
Table 10-l(c). Summary of Total Benefits Associated with Modelled Avoided IQ Decrements in
Prenatally Exposed Children Due to Reduced Mercury Exposure from Freshwater
Recreational Angling 10-10
Table 10-2. Summary Statistics for Estimated Fish Tissue Mercury Concentrations (ppm) by
State: 2001 Base Case3 10-16
Table 10-3. HUC-Level Distribution of Mercury Sampling Sites and Estimated Fish Tissue
Concentrations: 2001 Base Casea 10-17
Table 10-4. Summary of Fishing Activity Levels by State in 2001 from NSFHWR 10-20
Table 10-5. Overview of Key Attributes of the Population Centroid and Angler Destination
Models 10-23
Table 10-6. Block Group Demographic Characteristics by State (in 2000): Data Used in
Population Centroid Approach 10-30
xm
-------
Table 10-7. Estimated Annual Number of Prenatally Exposed Children for Selected Lag Periods
from 2001: Population Centroid Approach 10-31
Table 10-8. Estimated Annual Number of Prenatally Exposed Children for Selected Lag Periods
from 2020: Population Centroid Approach 10-33
Table 10-9. Average Estimated Mercury Concentrations (ppm) in Freshwater Fish by Distance
Interval from Block Group Centroids: Base Case 2001 10-35
Table 10-10. State-Level Summary of Exposed Population Estimates: Angler Destination
Approach 10-42
Table 10-11. Effects of Emission Control Scenarios—Percent Reduction in Estimated Fish
Tissue Mercury Concentrations from 2001 Base Case3 10-49
Table 10-12. Effects of Emission Control Scenarios—Percent Reduction in Estimated Fish
Tissue Mercury Concentrations from 2020 Base Case with CAIR3 10-51
Table 10-13. Estimated Distribution of Mercury Ingestion by Distance Traveled to Fish:
Population Centroid Approach—2001 Base Case 10-54
Table 10-14. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and
Foregone Earnings: Population Centroid Approach—2001 Base Case3 10-55
Table 10-15. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and
Foregone Earnings: Population Centroid Approach—2020 Base Case with CAIR3
10-57
Table 10-16. 2020 Base Case with CAIR: Modelled Avoided Losses Relative to 2001 Base
Case (Applied to 2020 Demographics)—Population Centroid Approach3'b 10-61
Table 10-17. 2001 Utility Mercury Emissions Zero Out: Modelled Avoided Losses Relative to
2001 Base Case—Population Centroid Approach3-11 10-63
Table 10-18. 2020 with CAIR Emissions Zero Out: Modelled Avoided Losses Relative to 2020
with CAIR Base Case—Population Centroid Approach3'b 10-65
Table 10-19. Estimated Benefits of 2020 CAMR Control Option 1: Relative to 2020 with
CAIR—Population Centroid Approach3'1" 10-68
Table 10-20. Estimated Benefits of 2020 CAMR Control Option 2: Relative to 2020 with
CAIR—Population Centroid Approach3'15 10-70
Table 10-21. Summary of Annual Benefit Estimates: Population Centroid Approach3 . .. 10-72
Table 10-22. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and
Foregone Earnings: Angler Destination Approach—2001 Base Case3 10-77
Table 10-23. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and
Foregone Earnings: Angler Destination Approach—2020 with CAIR3 10-81
Table 10-24. 2020 Base Case with CAIR: Modelled Avoided Losses Relative to 2001 Base
Case Applied to 2020 Demographics—Angler Destination Approach3'b 10-83
Table 10-25. 2001 Utility Mercury Emissions Zero Out: Modelled Avoided Losses Relative to
2001 Base Case—Angler Destination Approach3'b 10-85
Table 10-26. 2020 with CAIR Emissions Zero Out: Modelled Avoided Losses Relative to 2020
with CAIR Base Case—Angler Destination Approach3 b 10-87
Table 10-27. Estimated Benefits of 2020 With CAIR Control Option 1: Relative to 2020 with
CAIR—Angler Destination Approach3 b 10-90
Table 10-28. Estimated Benefits of 2020 With CAIR Control Option 2: Relative to 2020 with
CAIR—Angler Destination Approacha-b 10-92
Table 10-29. Summary of Annual Benefit Estimates: Angler Destination Approach 10-94
Table 10-30. Summary and Comparison of Annual Benefit Estimates: Population Centroid
Approach vs. Angler Destination Approach 10-95
xiv
-------
Table 10-31. Summary and Comparison of Annual Benefit Estimates Under Alternative IQ
Dose-Response Assumptions: Population and Angler Destination Approach 10-96
Table 10-32. Summary of Estimated Mercury Exposures for Consumption-Based Subsistence
Population, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2001 Base Case3 10-105
Table 10-33. Summary of Estimated Mercury Exposures for Consumption-Based Subsistence
Population, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2020 with CAIRa 10-107
Table 10-34. Summary of Annual Benefit Estimates for Consumption-Based Subsistence
Population: Population Centroid Approach 10-109
Table 10-35. Summary of Estimated Mercury Exposures for Income-Based Subsistence
Population, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2001 Base Case3 10-114
Table 10-36. Summary of Estimated Mercury Exposures for Income-Based Subsistence
Population, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2020 with CAIRa 10-116
Table 10-37. Summary of Annual Benefit Estimates for Income-Based Subsistence Population:
Population Centroid Approach 10-118
Table 10-38. Block Group Demographics for Hmong and Chippewa Females, Aged 15 to 44 (in
2001) 10-120
Table 10-39. Estimated Annual Number of Prenatally Exposed Children from Special
Populations for Selected Lag Periods: Population Centroid Approach 10-124
Table 10-40. Summary of Estimated Mercury Exposures for Special Populations in 2001, with
Associated IQ Decrements and Foregone Earnings: Population Centroid
Approach—Base Case 2001 10-125
Table 10-41. Summary of Estimated Mercury Exposures for Special Populations in 2020, with
Associated IQ Decrements and Foregone Earnings: Population Centroid
Approach—Base Case 2020 with CAIRa 10-126
Table 10-42. Summary of Annual Benefit Estimates for Hmong Special Population: Population
Centroid Approach 10-127
Table 10-43. Summary of Annual Benefit Estimates for Chippewa Special Population:
Population Centroid Approach 10-128
Table 10-44. Results of the Sensitivity Analysis Examining Distributional Equity for Native
American (subsistence) Populations 10-133
Table 10-45. Unquantified Health and Ecosystem Effects Associated with Exposure to Mercury
10-142
Table 11-1. Freshwater Fish IQ Loss and Hair Mercury from the 2020 Baseline 11-6
Table 11-2. Change in IQ Loss and Hair Mercury from the 2020 Zero Out (No Threshold)
Compared to the Baseline, and the Relative Probability of each Change Category .. 11-7
Table 11-3. Change in IQ Loss and Hair Mercury from the CAMR Option 1 (No Threshold)
Compared to the Baseline, and the Relative Probability of each Change Category .. 11-8
Table 11-4. Change in IQ Loss and Hair Mercury from the CAMR Option 2 (No Threshold)
Compared to the Baseline, and the Relative Probability of each Change Category .. 11-8
Table 11-5. Joint Distribution of Mercury Exposure from Freshwater Fish and Total Mercury
Exposure 11-10
Table 11-6. Scaling Factors 11-13
xv
-------
Table 11-7. IQ Benefits for CAMR Option 1 under Established Health-Based Benchmarks
11-14
Table 11-8. IQ Benefits for CAMR Option 2 under Established Health-Based Benchmarks
11-14
Table 12-1. PM2.5 Co-Benefits Associated with CAMR Regulatory Options 1 and 2
in 2020 12-8
Table A1-1. Observed Mercury Concentrations in Northern Pike from Lee Dam (DMA-80
results) Al-4
Table Al-2. Statistical summary of northern pike length normalized BAF (4 years) for Eagle
Butte (used in SERAFM) Al-5
Table Al-3. SERAFM Parameter Values for Eagle Butte Al-5
Table Al-4. Calibrated SERAFM Rate Constants for Eagle Butte Al-6
Table Al-5. SERAFM 50% Load Reduction Scenario for Eagle Butte Al-6
Table Al-6. SERAFM Zero-Out Scenario for Eagle Butte (Removal of Deposition attributed to
coal-fired utilities) in the CMAQ and REMSAD Models Al-6
Table A1-7. WASP Forecasted Mercury Concentrations in Eagle Butte Sediments in Response
to 50% Loading Reduction Scenario Al-12
Table A2-1. Summary of Yellow Perch Mercury Data from Pawtuckaway Lake A2-1
Table A2-2. Pawtuckaway Lake Parameter Values A2-2
Table A2-3. A Comparison of Measured and Baseline Steady State Values for Pawtuckaway
Lake A2-2
Table A2-4. Lake Pawtuckaway SERAFM Calibrated Rate Constants A2-3
Table A2-5. Time to Reach 90% Steady State After 50% Reduction in Atmospheric
Deposition A2-3
Table A2-6. SERAFM Model Forecasts with Zero-Out Scenario for Coal-Fired Power Plants
(Medium Response Time Scenario) A2-3
Table A2-7. Mercury Response Times for Lake Pawtuckaway, in years A2-8
Table A3-1. Observational Data from Lake Waccamaw A3-2
Table A3-2. Raw Fish Tissue Data Collected from Lake Waccamaw A3-3
Table A3-3. Annual Wet Deposition of Mercury at Waccamaw 1998-2000 A3-4
Table A3-4. Model Parameter Values A3-5
Table A3-5. Measured and Baseline Steady State Values for Lake Waccamaw A3-5
Table A3-6. SERAFM Calibrated Rate Constants for Lake Waccamaw A3-6
Table A3-7. SERAFM 50% Load Reduction Scenario for Lake Waccamaw A3-6
Table A3-8. SERAFM Zero-Out Scenario for Lake Waccamaw (Removal of Deposition
Attributed to Coal-fired Utilities) in the CMAQ and REMSAD Models A3-6
Table A3-9. WASP Response Time Estimates for Lake Waccamaw A3-11
Table A4-1. Average Mercury Deposition Hg Concentrations and Depositions Rates .... A4-4
Table A4-2. Specified and Calculated Reaction Rates and Coefficients A4-6
Table A4-3. Flows, Depths, Length and Volumes used in WASP Model A4-7
Table A4-4. Measured vs. Predicted for Sediment Components A4-7
Table A4-5. Predicted and Observed Mercury Concentrations under Annual Average Load and
Flow A4-8
Table A4-6. Soil Mercury Data in Local Region A4-8
Table A4-7. June 2003 Survey vs WASP Predictions for Mercury A4-18
Table A5-1. Lake Barco Parameter Values A5-1
Table A5-2. Measured and Baseline Steady State Values for Lake Barco A5-2
xvi
-------
Table A5-3. Lake Barco SERAFM Calibrated Rate Constants A5-2
Table A5-4. Time to Reach 90% Steady State After 50% Reduction in Atmospheric
Deposition A5-2
Table A5-5. SERAFM Model Forecasts with Zero-Out Scenario for Coal-Fired Power
Plants A5-3
Table D-l. Statistical Distribution of Residuals D-3
Table D-2. Differences between the Performance of Lake and River Samples Used as Inputs
into the NDMMF D-5
Table D-3. Statistical Distribution of Residuals from Withheld Data Set D-7
Table El-1. Reported Trip Travel Distance for Freshwater Anglers (miles) El-2
Table El-2. Demographic Characteristics of Freshwater Anglers3 El-3
Table El-3. Demographic Characteristics of Freshwater Anglers El-3
Table El-4. OLS Regression Results for Determinants of Reported Trip Travel
Distance (miles) El-5
Table El-5. Travel Distance Frequencies by Demographic Group (Percentage in each Distance
Category) El-6
Table E2-1. Frequency Distributions for HUC Level-of Use Indicators E2-4
Table E2-2. Variable Definitions and Descriptive Statistics E2-5
Table E2-3. Estimated Determinants of HUC Level-of-Use Indicators for Lake Trips: Negative
Binomial Regressions E2-6
Table E2-4. Estimated Determinants of HUC Level-of-Use Indicators for River Trips: Negative
Binomial Regressions E2-7
Table E2-5. Predicted Level-of-Use Indicators for HUCs in Study Area: Negative Binomial
Regression Model Predictions E2-8
xvn
-------
Figures
Figure 2-1. Probability Distribution Function of Blood Mercury Levels in US Women of
Childbearing Age (NHANES Data 1999-2002) 2-5
Figure 3- 1. BASS Predicted BAFs for Pike/Perch in Lee Dam, Eagle Butte 3-28
Figure 3- 2. Observed vs. Predicted Fish Mercury Concentrations in Model Ecosystems at
Steady State with No Change in Atmospheric Loading 3-29
Figure 3- 3. Temporal Response of Mercury Concentrations in Fish from Pawtuckaway Lake,
NH to a Decline in Mercury Loading 3-30
Figure 3- 4. Temporal Response of Mercury Concentrations in Fish from Lake Barco, FL to a
Decline in Mercury Loading 3-31
Figure 3- 5. Projected Ecosystem Response Times to Zero-out Deposition Scenario Using the
SERAFM Model 3-33
Figure 3- 6. Upper Brier Creek Loading Flux Attenuation 3-35
Figure 3- 7. Watershed Loading Flux Attenuation Considering Land-use Change 3-35
Figure 3- 8. Model Ecosystem Locations with CMAQ Grid Cells and Locations of Electricity
Generating Units (EGUs) 3-40
Figure 3- 9. Model Ecosystem Locations with Gradient in Sulfate Deposition Across the Eastern
US 3-41
Figure 3- 10. Model Ecosystem Locations with Percent Wetland Area Aggregated for
Each HUC 3-42
Figure 3-11. Model Ecosystem Locations with CMAQ 2001 Total Mercury Deposition . . 3-43
Figure 3-12. Model Ecosystem Locations with Measured Fish Tissue Concentrations .... 3-44
Figure 3- 13. Measured 1995-2001 Fish Hg Concentrations > 0.3 ppm 3-45
Figure 4-1. Sources of U.S. Fish Consumption 4-2
Figure 4-2. Location of U.S. Aquacultural Operations, 1998 4-6
Figure 4-3. Recreational Fishing Participation Rates by State 4-7
Figure 4-4. 2002 Commercial Landings by Distance from Shore 4-9
Figure 4-5. 2002 Recreational Marine Finfish Landings by Distance from Shore 4-17
Figure 4-6. U.S. Commercial Fish Imports by Area, 2002 4-20
Figure 4-7. U.S. Commercial Fish Imports by Country, 2002 4-21
Figure 4-8. 2001 Total Recreational Fishing Days 4-29
Figure 4-9. 2001 Recreational Fishing Days, State Residents 4-29
Figure 4-10. 2001 Recreational Fishing Days, Non-State Residents 4-30
Figure 4-11. 2002 Market Share of Commercial Finfish (% of Total Economic Value) 4-37
Figure 4-12. 2002 Market Share of Commercial Shellfish (% of Total Economic Value) .. 4-38
Figure 4-13. Fish Consumption Pathways 4-40
Figure 5-1. NLFA Sample Locations 5-3
Figure 5-2. Frequency Distribution of Average Watershed Fish Tissue Concentrations (ppm) 5-4
Figure 5-3. Frequency and Average Concentrations of Various NLFA Sample Methods (Cuts of
Fish) 5-4
Figure 5-4. Sample Locations from the NLFTS 5-6
Figure 5-5. Cumulative Distribution Functions (CDFs) for Normalized NLFTS and NLFA Lake
Data 5-8
Figure 5-6. Total Area/Sample in the Combined NLFA and NLFTS Data Set 5-9
Figure 5-7. Locations Where Normalized Fish Tissue Concentrations are Utilized, and Where
Non-Normalized Data are Utilized 5-12
Figure 5-8. Baseline Average Hg Fish Tissue Concentrations 5-13
xviii
-------
Figure 5-9. Statistical Distribution of Averages of Hg. Fish Tissue Concentrations in ppm . 5-14
Figure 5-10. Number of Unique Sampling Events Within Each HUC 5-15
Figure 5-11. Average Fish Tissue Concentrations By HUC 5-16
Figure 5-12. Frequency Distribution of HUC Averaged Concentrations (ppm) 5-16
Figure 6-1. Status of State Electricity Industry Restructuring Activities (as of February 2003)6-4
Figure 6-2. Emissions of Hg, SO2, and NOX from the Power Sector (2003) 6-5
Figure 7-1. Projected Mercury Emissions in 2020 by State 7-3
Figure 7-2. NERC Power Regions 7-10
Figure 8-1. CMAQ Modeling Domain 8-6
Figure 8-2. Base Case Total Mercury Deposition: 2001 8-10
Figure 8-3. Decrease in Total Mercury Deposition with Power Plant Zero-Out
Simulation: 2001 8-11
Figure 8-4. Change in Total Mercury Deposition for All Sources: 2020 (with CAIR)
Relative to 2001 8-12
Figure 8-5. Total Mercury Deposition: 2020 (with CAIR) 8-12
Figure 8-6. Change in Mercury Depositions from Power Plants Due to CAMR
Option 1: 2020 8-13
Figure 8-7. Change in Mercury Deposition from Power Plants Due to CAMR
Option 2: 2020 8-13
Figure 8-8. Cumulative Distribution of Total Mercury Deposition (ug/m2) Fat HUC-8 Level by
Modeling Scenario 8-15
Figure 8-9. Cumulative Distribution of Utility Attributable Mercury Deposition at HUC-8 Level
by Model Scenario 8-16
Figure 8-10. Cumulative Distribution of Percent Deposition (ug/m2) Attributable to Utilities at
HUC-8 Level by Modeling Scenario 8-17
Figure 9-1. 95% Confidence Intervals for Full Scale IQ from the New Zealand, Seychelles and
Faroes Studies 9-7
Figure 10-1. Locations of Lake Fish Tissue Mercury Sampling Sites Used in the Analysis 10-14
Figure 10-2. Locations of River Fish Tissue Mercury Sampling Sites Used in the Analysis
10-15
Figure 10-3. Flow Diagram for Population Centroid Approach 10-25
Figure 10-4. Population Centroid Approach: Linking Census Block Groups to Demographic
Data and Mercury Fish Tissue Samples 10-27
Figure 10-5. Flow Diagram for Angler Destination Approach 10-37
Figure 10-6. Estimated Distribution of Lake-Fishing Days Across HUCs in 2001 10-40
Figure 10-7. Estimated Distribution of River-Fishing Days Across HUCs in 2001 10-41
Figure 10-8. Spatial Distribution of Estimated Average Daily Maternal Mercury Ingestion
Rates: Angler Destination Approach—2001 Base Case 10-74
Figure 10-9. Spatial Distribution of Estimated IQ Decrements per HUC: Angler Destination
Approach—2001 Base Case 10-75
Figure 10-10. Spatial Distribution of Estimated Percent Reduction in IQ Losses: Improvement
with 2001 Utility Emissions Zero Out Scenario (Zero Lag) 10-76
Figure 10-11. Distribution of Modelled Avoided IQ Decrements (Benefits) due to Mercury
Emissions Reductions: 2001 Utility Emissions Zero-Out Relative to 2001 Base Case;
Population Centroid Approach; Variable Consumption Rate 10-99
xix
-------
Figure 10-12. Cumulative Distribution of Modelled Avoided IQ Decrements (Benefits) due to
Mercury Emissions Reductions: 2001 Utility Emissions Zero-Out Relative to 2001 Base
Case; Population Centroid Approach; Variable Consumption Rate 10-99
Figure 10-13. Distribution of Modelled Avoided IQ Decrements (Benefits) due to Mercury
Emissions Reductions: CAMR Control Option 1 Relative to 2020 Base Case with CAIR;
Population Centroid Approach; Variable Consumption Rate 10-100
Figure 10-14. Cumulative Distribution of Modelled Avoided IQ Decrements (Benefits) due to
Mercury Emissions Reductions: CAMR Control Option 1 Relative to 2020 Base Case
with CAIR; Population Centroid Approach; Variable Consumption Rate 10-100
Figure 10-15. Distribution of Modelled Avoided IQ Decrements (Benefits) due to Mercury
Emissions Reductions: CAMR Control Option 2 Relative to 2020 Base Case with CAIR;
Population Centroid Approach; Variable Consumption Rate 10-101
Figure 10-16. Cumulative Distribution of Modelled Avoided IQ Decrements (Benefits) due to
Mercury Emissions Reductions: CAMR Control Option 2 Relative to 2020 Base Case
with CAIR; Population Centroid Approach; Variable Consumption Rate 10-101
Figure 10-17. Distribution of Modelled Avoided IQ Decrements (Benefits) due to Mercury
Emissions Reductions: 2020 Utility Emissions Zero-Out Relative to 2020 Base Case
with CAIR; Population Centroid Approach; Variable Consumption Rate 10-102
Figure 10-18. Cumulative Distribution of Modelled Avoided IQ Decrements (Benefits) due to
Mercury Emissions Reductions: 2020 Utility Emissions Zero-Out Relative to 2020 Base
Case with CAIR; Population Centroid Approach; Variable Consumption Rate .. . 10-102
Figure 10-19. U.S. Census Tracts with Native American Populations 10-113
Figure Al-1. Location of Lee Dam (lower left quadrant) on La Plant SW quadrangle .... A1-3
Figure Al-2. WASP Water Column Solids Calibration Al-7
Figure Al-3. WASP Upper Sediment Solids Calibration Al-8
Figure Al-4. WASP Burial Rate Calibration Al-8
Figure Al-5. WASP Total Mercury Buildup in Water Al-9
Figure Al-6. WASP Methyl Mercury Buildup in Water Al-9
Figure Al-7. WASP Total Mercury Buildup in Sediment Al-10
Figure Al-8. WASP Methyl Mercury Buildup in Sediment Al-10
Figure Al-9. WASP Total Mercury Attenuation in Water Al-11
Figure Al-10. WASP Total Mercury Attenuation in Surface Sediment Al-11
Figure Al-11. WASP Attenuation Sensitivity in Water Al-12
Figure Al-12. WASP Attenuation Sensitivity in Water Al-13
Figure Al-13. Base Case Response of Northern Pike to Methylmercury Exposure Al-14
Figure Al-14. Base case response of yellow perch to methylmercury exposure (0.5ng/L) Al-14
Figure Al-15. Attenuation of Methylmercury in Northern Pike after Load Reduction . .. Al-15
Figure Al-16. Attenuation of Methylmercury in Yellow Perch after Load Reduction . . . Al-15
Figure A2-1. WASP Water Column Solids Calculation A2-4
Figure A2-2. WASP Solids Simulation for Surface Sediment A2-5
Figure A2-3. WASP Simulation of Burial Velocity A2-5
Figure A2-4. WASP Total Mercury Buildup in Water A2-6
Figure A2-5. WASP Methyl Mercury Buildup in Water A2-7
Figure A2-6. WASP Total Mercury Buildup in Sediment A2-7
Figure A2-7. WASP Methyl Mercury Buildup in Sediment A2-8
Figure A2-8. WASP Total Mercury Attenuation in Epilimnion A2-9
Figure A2-9. WASP Total Mercury Attenuation in Hypolimnion A2-9
xx
-------
Figure A2-10. WASP Total Mercury Attenuation in Surface Sediment A2-10
Figure A3-1. Southeastern North Carolina and Lake Waccamaw A3-1
Figure A3-2. WASP Water Column Solids Calculation A3-7
Figure A3-3. WASP Solids Simulation for Surface Sediment A3-8
Figure A3-4. WASP Simulation of Burial Velocity A3-8
Figure A3-5. WASP Total Mercury Buildup in Water A3-9
Figure A3-6. WASP Methyl Mercury Buildup in Water A3-10
Figure A3-7. WASP Total Mercury Buildup in Sediment A3-10
Figure A3-8. WASP Methyl Mercury Buildup in Sediment A3-11
Figure A3-9. WASP Total Mercury Attenuation in Epilimnion A3-12
Figure A3-10. WASP Total Mercury Attenuation in Surface Sediment A3-12
Figure A4-1. Brier Creek Watershed A4-1
Figure A4-2. Brier Creek Subwatersheds for Hg Loadings A4-2
Figure A4-3. Brier Creek Watershed Landuses A4-3
Figure A4-4. Mercury Deposition Network Sampling Locations A4-4
Figure A4-5. Brier Creek Soil Mercury Buildup A4-9
Figure A4-6. Brier Creek Loading Flux Buildup A4-10
Figure A4-7. Upper Brier Creek Soil Mercury Attenuation A4-11
Figure A4-8. Brier Creek Loading Flux Attenuation A4-12
Figure A4-9. Upper Brier Creek Soil Mercury Attenuation A4-13
Figure A4-10. Upper Brier Creek Loading Flux Attenuation A4-14
Figure A4-11. Watershed Loading Flux Attenuation considering Landuse Change A4-1S
Figure A4-12 Base Case Water Column Mercury Concentration for Brier Creek A4-16
Figure A4-13. Base Case Sediment Mercury Concentration for Brier Creek A4-17
Figure A4-14. Brier Creek Total Mercury Water Column Concentration A4-18
Figure A4-15. Brier Creek Methyl Mercury Water Column Concentration A4-19
Figure A4-16. Mercury Attenuation over Time in Water Column A4-20
Figure A4-17. Mercury Attenuation over Time in Sediments A4-21
Figure A4-18. Sensitivity Range for Upstream Waters A4-23
Figure A4-19. Sensitivity Range for Downstream Waters A4-24
Figure D-l. Box and Whisker Plots of the NLFWA Observed, NDMMF Estimated, and
Residuals Measurements in ppm D-3
Figure D-2. Scatterplot of Predicted vs. Observed Measurements D-4
Figure D-3. Scatterplot of Residual vs. Observed Measurements D-4
Figure D-4. Locations of Withheld Observations D-7
Figure D-5. Box and Whisker Plots of Observed, Predicted, and Residual (Error) Distributions
for the Withheld Data Set D-8
Figure D-6. Scatterplot of Predicted vs. Observed Measurements D-8
Figure D-7. Scatterplot of Residual vs. Observed Measurements D-9
Figure E2-1. U.S. Hydrologic Regions E2-2
xxi
-------
ACRONYMS AND ABBREVIATIONS
ACI
ACS
ADHD
ADI
ADP
AERMOD
AHRQ
AMI
ANL
ASA
atm
ATSDR
BAF
BASS
BC
BEA
BenMAP
BLS
BMD
BMDL
BMI
BMR
BOC
C-R
CAAA
CAIR
CAMR
CDF
CEEPR
cfs
CHSHg
CHD
CI
cm
CMAQ
CPS
CRDM
CRF
CRST
CSFII
CVD
D-MCM
DEP
DHHS
activated carbon injection
American Cancer Society
attention deficit hyperactivity disorder
acceptable daily intake
adenosine diphosphate
American Meteorological Society/EPA Regulatory Model
Agency for Healthcare Research and Quality
acute myocardial infarction
Argonne National Laboratory
American Sportfishing Association
atmosphere
Agency for Toxic Substance and Disease Registry
bioaccumulation factor
Bioaccumulation and Aquatic System Simulator
boundary conditions
Bureau of Economic Analysis
Benefits Mapping and Analysis Program
Bureau of Labor Statistics
benchmark dose
BMD lower statistical confidence limit
body mass index
benchmark response
Bureau of Census
concentration-response
Clean Air Act Amendments
Clean Air Interstate Rule
Clean Air Mercury Rule
Cumulative Distribution Function
Center for Energy and Environmental Policy Research
cubic feet per second
monomethyl mercury
coronary heart disease
confidence interval
centimeter
Community Multi-Scale Air Quality
Current Population Survey
Climatological Regional Dispersion Model
capital recovery factor
Cheyenne River Sioux Tribal
Continuing Survey of Food Intake by Individuals
cardiovascular disease
Dynamic Mercury Cycling Model
Department of Environmental Protection
Department of Health and Human Services
xxii
-------
DDT
DHA
DOC
DOE
DOI
DOM
DPA
E-MCM
ECG
EFH
EGRID
ECU
EIA
EIA
EKG
ELA
EMF
EMMA
E.G.
EPA
EPRI
ESP
EU
EURAMIC
EXAMS2
F
FAO
FDA
FERC
FF
FGD
FGETS
FL
ft
g
GDP
GIS
GRU
GW
GWh
H-PAC
HDL
Hg
Hg°
Hg(H)
HgC12
dichloro-diphenyl-trichloroethane
docosahexaenoic acid
dissolved organic carbon
Department of Energy
Department of the Interior
dissolved organic matter
docosapentaenoic acid
Everglades Mercury Cycling Model
electrocardiogram
Exposure Factors Handbook
Emissions & Generated Resource Integrated Database
electric generating unit
Economic Impact Analysis
Energy Information Administration
electrocardiogram
Experimental Lakes Area
emission modification factor
Environmental Monitoring and Measurement Advisor
Executive Order
Environmental Protection Agency
Electric Power Research Institute
electrostatic precipitator
European Union
The European Multicenter Case Control Study on Antioxidants,
Myocardial Infarction and Cancer of the Breast
Exposure Analysis Modeling System, Version 2
Fahrenheit
Food and Agriculture Organization
Food and Drug Administration
Federal Energy Regulatory Commission
fabric filters
flue gas desulfurization
Food and Grill Exchange of Toxic Substances
fork length
feet
gram
gross domestic product
Geographic Information System
Gainesville Regional Utilities
gigawatt
gigawatt hours
Hazard Prediction and Assessment Capability
high-density lipoprotein
mercury
elemental mercury
inorganic divalent mercury
mercuric chloride
xxin
-------
HgP
HgT
hrs
HUC
HYSPLIT
ICR
IEM-2M
IGCC
IMT
in
IOM
IOU
IQ
IPM
IRIS
ISC3
K
kg
KIHD
km
kWh
Ib
Ibs
L
LC50
LC omega-3 PUFA
LDL
m
M
mm
MACT
MAS/MILS
MCM
MDN
ME
MeHg
METAALICUS
mg
MI
MIT
MM5
MMAPS
mmBtu
mol
MRFSS
MRL
particulate mercury
total mercury
hours
hydrologic unit code
Hybrid Single Particle Lagrangian Integrated Trajectory
Information Collection Request
Indirect Exposure Model, Version 2
Integrated Gasification Combined Cycle
intima-media thickness
inch
Institute of Medicine
investor-owned utility
intelligence quotient
Integrated Planning Model
Integrated Risk Information System
Industrial Source Complex
Kelvin
kilogram
Kuopio Ischemic Heart Disease
kilometer
kilowatt hour
pound
pounds
liter
lethal concentration for 50% percent of the population
long chain omega-3 polyunsaturated fatty acids
low-density lipoprotein
meter
molar mass
millimeter
Maximum Achievable Control Technology
Mineral Availability System/Mineral Industry Location System
Mercury Cycling Model
Mercury Deposition Network
Midwest/Northeast
methylmercury
Mercury Experiment to Assess Atmospheric Loading in Canada and the
United States
milligram
myocardial infarction
Massachusetts Institute of Technology
Mesoscale Model
Mercury Maps
million British thermal units
mole
Marine Recreational Fishing Statistical Survey
minimal risk level
xxiv
-------
MW
MWC
MWh
MWI
NAAQS
NAICS
NAS
NASS
NDMMF
NEI
NERC
NES
NFTS
ng
NHD
NHANES
NH3
NLCD
NLFA
NLSY
NLFTS
NMFS
NOAA
NOAEC
NOAEL
NODA
NOx
NPR
NPV
NRC
NRS
NSFHWR
NSPS
NSR
NSRE
NW
O&M
OM
OMB
OR
ORD
P-PUFA
PAC
PCB
megawatt
municipal waste combustor
megawatt hour
medical waste incinerator
National Ambient Air Quality Standards
North American Industry Classification System
National Academy of Sciences
National Agriculture Statistics Service
National Descriptive Model of Mercury in Fish
National Emissions Inventory
North American Electric Reliability Council
Neurobehavioral Evaluation System
National Fish Tissue Survey
nanogram
National Hydrography Database
National Health and Nutrition Examination Survey
ammonia
National Land Cover Data
National Listing of Fish and Wildlife Advisories
National Longitudinal Study of Youth
National Lake Fish Tissue Survey
National Marine Fisheries Service
National Oceanic and Atmospheric Administration
No Observable Adverse Effects Concentration
No Observed Adverse Effect Level
Notice of Data Availability
nitrogen oxides
Notice of Proposed Rulemaking
net present value
National Research Council
National Recreation Survey
National Survey of Fishing, Hunting and Wildlife-Associated Recreation
New Source Performance Standards
New Source Review
National Survey on Recreation and the Environment
Northwest
operation and maintenance
organic matter
Office of Management and Budget
odds ratio
Office of Research and Development
plasma polyunsaturated fatty acids
powder activated carbon
polychlorinated biphenyl
xxv
-------
PCS
PH
PM
POTW
ppb
ppm
RARE
RELMAP
REMI
REMSAD
RFA
RfD
RFF
ROM
RIA
RNA
RQ
RR
RtC
RUSLE
SAB-HES
SCR
SD
SE
sec
SIC
SIP
SMR
SNCR
SO2
sqkm
SR-MATRIX
SRB
st dev
SW
SWAT
TDI
TL
TMDL
TOLD-SL
TRIM
TSD
UF
Permit Compliance System
potential of hydrogen
participate matter
Publicly Owned Treatment Works
parts per billion
parts per million
Regional Applied Research Effort
Regional Lagrangian Model of Air Pollution
Regional Economic Models, Inc.
Regulatory Modeling System for Aerosols and Deposition
Regulatory Flexibility Act
Reference Dose
Resources for the Future
Reactive Gaseous Mercury
Regulatory Impact Analysis
ribonucleic acid
risk quotient
relative risk
Report to Congress
Revised Universal Soil Loss Equation
Science Advisory Board Health Effects Subgroup
selective catalytic reduction
standard deviation
Southeast
second
standard industrial classification
state implementation plan
standard mortality rate
selective non-catalytic reduction
sulfur dioxide
square kilometer
Source Receptor Matrix
sulfate reducing bacteria
standard deviation
Southwest
Soil and Water Assessment Tool
tolerable daily intake
total length
Total Maximum Daily Load
Test of Language Development - Spoken Language
Technology, Retrofit and Upgrading Model
Technical Support Document
uncertainty factor
microgram
xxvi
-------
uM
um
UMRA
U.S.
USAGE
U.S.C.
USDA
USEPA
USFWS
USGS
UV-B
VMI
VSL
WASP
wcs
WHO
wise
WISC-III
WISC-R
WRAML
WTP
yr
micromole
micrometer
Unfunded Mandates Reform Act
United States
United States Army Corps of Engineers
United States Code
United States Department of Agriculture
United States Environmental Protection Agency
United States Fish and Wildlife Service
United States Geological Survey
ultraviolet light, type B
Visual-Motor Integration
Value of Statistical Life
Water Quality Analysis Simulation Program
Watershed Characterization System
World Health Organization
Wechsler Intelligence Scales for Children
Wechsler Intelligence Scales for Children administered in the Seychelles
Islands
Wechsler Intelligence Scales for Children administered in New Zealand
and the Faroe Islands
Wide-Range Assessment of Memory Learning
willingness to pay
year
xxvi i
-------
EPA-452/R-05-003
March 2005
Regulatory Impact Analysis of the Clean Air Mercury Rule
Final Report
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Strategies and Standards Division
Innovative Strategies and Economics Group (MD 339-01)
Research Triangle Park, N.C. 27711
-------
SECTION 1
INTRODUCTION
1.0 Introduction
This report provides an analysis of the benefits and costs of the final Clean Air Mercury
Rule (CAMR). In Section 2, we discuss the potential health effects of mercury. Section 3
provides a detailed discussion of mercury in the environment, including how mercury deposited
to water bodies transforms into methylmercury in fish tissue. This section also provides an
assessment of the response time for systems after a change in mercury deposition. Because fish
consumption is the primary pathway for exposure to methylmercury, Section 4 provides a profile
of fishing activity in the United States. Section 5 presents information on concentrations of
mercury in fish. Because this regulation requires control on coal-fired power plants, Section 6
provides a profile of the power sector in the United States, while Section 7 describes the
emissions, control requirements, control options considered for CAMR, and the regulatory costs
of the final CAMR. In addition, Section 7 also provides an assessment of impacts on small
businesses and government entities. Section 8 describes the resulting change in mercury
deposition from air quality modeling of the CAMR regulatory options. Section 9 presents a
derivation of a dose-response function that relates mercury consumption in women of
childbearing with changes in IQ seen in children that were exposed prenatally. IQ is used as a
surrogate for the neurobehavioral endpoints that EPA relied upon for setting the methylmercury
reference dose (RfD). Chapter 10 presents exposure modeling and benefit methodologies
applied to a no-threshold model (i.e., a model that assumes no threshold in effects at low doses
of mercury exposure). Chapter 11 presents the final benefit analysis numbers of CAMR giving
consideration to established health benchmarks (i.e., consideration of potential thresholds on
effects at low doses of mercury exposure). Finally, Chapter 12 presents a benefit analysis of
reductions in PM as a result of controls applied for mercury. Table 1-1 below summarizes the
benefits, costs, and net benefits of the CAMR.
1-1
-------
SECTION 2 IMPACT OF MERCURY ON HUMAN HEALTH, ECOSYSTEMS, AND
WILDLIFE 2-1
2.1 Introduction 2-1
2.2 Mercury Poisoning Episodes 2-1
2.3 Reference and Benchmark Doses 2-2
2.4 Neurologic Effects 2-6
2.5 Cardiovascular Impacts 2-7
2.6 Genotoxic Effects 2-7
2.7 Immunotoxic Effects 2-7
2.8 Other Human Toxicity Data 2-8
2.9 Ecological Effects 2-9
2.10 Conclusions 2-9
2.11 References 2-10
Figures
Figure 2-1. Probability Distribution Function of Blood Mercury Levels in US Women of
Childbearing Age (NHANES Data 1999-2002) 2-5
-------
SECTION 2
IMPACT OF MERCURY ON HUMAN HEALTH, ECOSYSTEMS, AND WILDLIFE
2.1 Introduction
This section discusses the potential human health and ecological effects due to exposure
to methylmercury. The material in this section is based upon the National Research Council
(NRC) of the National Academies of Science report titled "Toxicological Effects of
Methylmercury," which provides a thorough review of the effects on mercury on human health
(NRC 2000), augmented by other related and more recent publications regarding the effects of
methylmercury exposure. Many of the peer-reviewed articles cited in this section are
publications originally cited in the NRC report (all secondary citations are clearly noted in the
reference section).
The section starts with a short account of mercury poisoning episodes in Japan and Iraq
(Section 2.2), which provide much of the basis for research on human health at very high
exposure levels. Next, the reference dose and benchmark dose for methylmercury exposure are
discussed in Section 2.3 to provide context for the ensuing descriptions of specific health effects
(Sections 2.4 - 2.8). The section concludes with a discussion of ecological effects (Section 2.9)
and conclusions (Section 2.10).
2.2 Mercury Poisoning Episodes
Instances of methylmercury poisoning have made it clear that adults, children, and
developing fetuses are at risk from ingestion exposure to methylmercury. These episodes
resulted in exposures well above those observed in any US subpopulations, however, they
provided early motivation for risk management of mercury. Two of these high-dose mercury
poisoning occurred in Japan and Iraq. In Japan, industrial by-products containing organic
mercury were discharged into Minamata Bay between 1953 and 1960, contaminating fish and
resulting in methylmercury poisoning of the local population via consumption offish. The
central nervous system was the primary target; symptoms of exposure included paresthesia (a
burning or prickling sensation in the skin), ataxia (failure of muscle control), sensory
disturbances (e.g., impaired vision, hearing, and smell), tremors, difficulty in walking,
irritability, and others, including death (NRC 2000). Children of exposed women displayed a
higher incidence of symptoms than did exposed adults. Some victims were born with a
condition resembling cerebral palsy, with severe disturbances of nervous function, and affected
offspring were very late in reaching developmental milestones (EPA 1997, UNEP 2002).
Maternal hair mercury concentrations in this population ranged from 3.8 to 133 ppm
(mean of 41 ppm). There is significant uncertainty associated with these exposure estimates,
primarily because measurements of methylmercury exposure were not taken until several years
after the poisoning episode had begun and identification of cases was incomplete; however, it is
clear that the exposures were quite high.
2-1
-------
Another series of acute mercury poisonings occurred in Iraq in the late 1950s and 1972
following consumption of bread made with seed grain treated with fungicides containing
alkylmercury compounds, affecting thousands of people in total. Symptoms in the exposed
population were similar to those observed in Minamata, with severely affected individuals
exhibiting paresthesia, ataxia, blurred vision, slurred speech, hearing difficulties, blindness,
deafness, and death. Toxicity was observed in many adults and children who had consumed this
bread over a three-month period, but the population that showed greatest sensitivity were
offspring of women who had eaten contaminated bread during pregnancy. Maximum maternal
hair mercury levels during pregnancy for mothers of affected children ranged to over 600 ppm,
with some effects in children (e.g., delayed ability to walk) possibly associated with maternal
hair mercury levels less than 100 ppm.
In both Iraq and Japan, the effects in offspring prenatally exposed to methylmercury were
more serious, and in some cases seen at lower doses, than in adults (EPA 1997, NRC 2000).
These instances of methylmercury poisoning have made it clear that adults, children, and
developing fetuses are at risk from ingestion exposure to methylmercury. In both episodes,
mothers with few or no symptoms of nervous system damage gave birth to infants with severe
disabilities, and it became clear that the developing nervous system of the fetus is more
vulnerable to methylmercury than the adult nervous system. Even though these episodes
resulted in exposures well above those observed in any US subpopulations, they provided early
motivation for risk management of mercury, and the U.S. FDA first proposed an administrative
guideline for mercury levels in fish and shellfish in 1969 in response to the poisonings in Japan
(EPA 1997). In the years since these episodes, much research has been undertaken to more fully
understand the effects associated with high-dose methylmercury poisoning as well as more
common lower-dose exposures, and these data have been used by EPA and others in mitigating
potential human health effects.
2.3 Reference and Benchmark Doses
EPA has set a health-based ingestion rate for chronic oral exposure to methylmercury,
termed an oral Reference Dose (RfD). The RfD is an estimate (with uncertainty spanning
perhaps an order of magnitude) of a daily exposure to the human population (including sensitive
subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime
(EPA 2002). EPA believes that exposures at or below the RfD are unlikely to be associated with
appreciable risk of deleterious effects. It is important to note, however, that the RfD does not
define an exposure level corresponding to zero risk; mercury exposure near or below the RfD
could pose a very low level of risk which EPA deems to be non-appreciable. It is also important
to note that the RfD does not define a bright line, above which individuals are at risk of adverse
effect.
In 1995, EPA set an oral RfD for methylmercury at 0.0001 mg/kg-day based on a study
of the Iraqi poisoning episode (Marsh et al. 1987). Subsequent research from large
epidemiological studies in the Seychelles, Faroe Islands, and New Zealand added substantially to
the body of knowledge on neurological effects from methylmercury exposure. Per
Congressional direction via the House Appropriations Report for Fiscal Year 1999, the NRC was
contracted by EPA to examine these data and, if appropriate, make recommendations for
2-2
-------
deriving a revised RfD. NRC's analysis concluded that the Iraqi study should no longer be
considered the critical study for the derivation of the RfD and also provided specific
recommendations to EPA regarding methylmercury based on analyses of the three large
epidemiological studies (NRC 2000). EPA's current assessment of the methylmercury RfD,
revised in 2001, relied on the quantitative analyses performed by the NRC (EPA 2002).
In their analysis, NRC examined in detail the epidemiological data from the Seychelles,
the Faroe Islands, and New Zealand, as well as other toxicological data on methylmercury. In
determining a recommended point of departure (i.e., the specific dose on which health criteria
should be based), NRC recommended a benchmark dose approach which applies mathematical
models to the available data to identify the point of departure. The BMD is the exposure level at
which a particular level of response (i.e., the benchmark response, or BMR) for some outcome of
concern is predicted to occur. In their assessment of the epidemiological data, NRC proposed
that the Faroe Islands cohort was the most appropriate study for defining an RfD, and
specifically selected children's performance on the Boston Naming Test (a neurobehavioral test)
as the key endpoint. They recommended a BMR of 0.05 (i.e., the level at which would result in a
doubling in the number of children with a response at the 5th percentile of the population).1 On
the basis of this study cohort and that test, NRC identified a BMD of 85 ppb in cord blood. The
NRC also estimated the 95% lower confidence limit for the BMD (i.e., the BMDL) for this
endpoint to be 58 ppb. The BMDL is a conservative estimate which is used as a point of
departure in risk assessment. Although this BMDL was specifically recommended by NRC as
appropriate for deriving the RfD, NRC also conducted BMD analyses on other endpoints in the
Faroe cohort and several endpoints in the other two populations, as well as an integrative
analysis of data from all three studies (NRC 2000).
In updating the RfD, EPA considered BMD analyses completed by NRC involving
endpoints of neuropsychological development from the Faroe Islands cohort (including results
for the Boston Naming Test), the New Zealand cohort, and the NRC's integrative analysis of all
three studies. The BMDLs for these endpoints, measured as concentrations of mercury in
umbilical cord blood, were considered. For the purposes of calculating the RfD, EPA converted
these BMDLs to maternal daily dietary intake in mg/kg-day using a one-compartment model.2
The BMDLs for these analyses (measured in terms of mercury in cord blood) were all observed
to be within a relatively close range, and the calculated RfDs converge at about 0.0001 mg/kg-
day. Specifically, BMDLs for a number of neurological endpoints based on tests that gauge a
child's ability to learn and process information (i.e., Boston Naming Test, Continuous
Performance Test, California Verbal Learning Test, McCarthy Perceived Performance, and
McCarthy Motor Test) were calculated by NRC to range from about 25 to 100 ppb mercury in
cord blood. These exposures were converted to dietary exposures of about 0.0005 mg/kg-
1 As noted by NRC in reference to data from the Seychelles, Faroe Islands, and New Zealand, "because those data
are epidemiological, and exposure is measured on a continuous scale, there is no generally accepted procedure for
determining a dose at which no adverse effects occur." The NRC chose a 5% response level in the BMD analysis
for test results in the lower 5% of the distribution.
2 The one-compartment toxicokinetic model employed by EPA is described by NRC (2000); it represents all
maternal body compartments as a single pool with a relatively small set of parameters, and assumes steady-state
conditions in the maternal system. Methylmercury dose levels were measured as concentrations in umbilical cord
blood (analysts have assumed that methylmercury concentration in cord blood is roughly equal to that in maternal
blood).
2-3
-------
bw/day to 0.0019 mg/kg-day, with most dietary exposures estimated to be about 0.001 mg/kg-
bw/day. The integrative BMDL (taking into account data from all three studies) was calculated
by NRC to be 32 ppb mercury in cord blood, or an exposure of about 0.6 ug/kg-day. All of these
results were considered in defining the RfD; as stated in the IRIS summary for methylmercury:
"Rather than choose a single measure for the RfD critical endpoint, EPA based this RfD
for this assessment on several scores from the Faroes measures, with supporting analyses from
the New Zealand study, and the integrative analysis of all three studies." (EPA 2002)
EPA used the various BMDLs and then applied an uncertainty factor of 10 to account for
interindividual toxicokinetic variability and pharmacodynamic variability and uncertainty. On
this basis, EPA defined the updated RfD of 0.0001 mg/kg-day in 2001. Although derived from a
more complete data set and with a somewhat different methodology, the current RfD is the same
as the previous (1995) RfD.
The levels at which these key neurological effects were observed - in the study
populations on which the updated RfD is based - provide a useful frame of reference for
considering other, non-neurological effects described below (see above for observed exposure
levels). It is important to note that although these populations were exposed to elevated levels of
methylmercury via fish or marine mammal consumption, the exposure levels of interest in these
studies are far below those associated with the Minamata and Iraqi poisoning episodes
mentioned previously. In addition, to put these exposure levels in perspective, it is useful to
consider typical mercury exposure levels in the U.S. measured in the National Health and
Nutrition Examination Survey (NHANES). This survey is conducted by the National Center for
Health Statistics via standardized interviews to provide continuous health data for the general
U.S. population, and it has included measurements of mercury in blood and hair as biomarkers of
mercury exposure. Based on NHANES data for blood collected for 1999-2002, the overall
distribution of blood mercury concentrations for women of child-bearing age (i.e., between 16
and 49 years of age) has been estimated for the U.S. population (see Figure 2-1). The RfD and
BMDL derived from the Faroe cohort effect level are included on this chart for reference.
Although all observed exposures are below the BMDL, and most of the exposures fall below the
RfD, about 6% of the population exposures were at or above the RfD (MMWR Vol. 53 / No.
43). The geometric mean blood mercury concentration in the NHANES data for 1999-2002 is
0.92 ppb, and the range of observed concentrations was from 0.07 to 38.90 ppb.3
3 The NHANES data summarized above suggests that exposures of women of child-bearing age in the U.S. exceed
the RfD.
2-4
-------
100%
90%
80%
70%
20
30
40
50 60
NRC BMDL
70
80 90
NRC BMD
Hg in Blood (ppb)
Note: Cumulative frequency (y-axis) refers to the fraction of the population exposed at or below a given blood
mercury level. EPA's RfD for methylmercury is 0.1 ug/kg-day, which is approximately equivalent to a concentration
of 5.8 ppb in blood.
Figure 2-1. Probability Distribution Function of Blood Mercury Levels in US Women of
Childbearing Age (NHANES Data 1999-2002)
NRC notes in their analysis that a biomarker conversion factor of about 5 ppb in blood
per 1 ppm in hair can be used to estimate the corresponding hair mercury values. Using this
approach, the RfD of 5.8 ppb in blood corresponds to a hair mercury concentration of about 1
ppm, and the BMDL and BMD are equivalent to about 12 ppm and 17 ppm, respectively.
Analyses of hair mercury for U.S. women of child-bearing age have been conducted using
NHANES data from 1999-2000. A geometric mean hair mercury concentration of 0.20 ppm was
reported for this population, and the geometric mean of the concentration of organic mercury
was 0.80 ppb in blood (Mahaffey et al. 2004). Among frequent fish consumers (i.e., study
participants who reported consuming fish three or more times in the previous 30 days),4
geometric mean hair mercury levels were three-fold higher compared with nonconsumers (viz.,
0.38 ppm vs. 0.11 ppm). Higher percentiles of exposure were also reported, with the 95th
percentile hair mercury levels corresponding to 1.73 ppm and 2.75 ppm for all women and
women frequently consuming fish, respectively (McDowell et al. 2004).
In general, the primary route by which the U.S. population is exposed to mercury is
through the consumption offish containing methylmercury. Exposure to methylmercury may
result in a variety of health effects. The various categories of health effects, and the evidence on
their significance, are described in the following pages.
2.4 Neurologic Effects
4 Fish consumption rates were collected by questionnaire at the time of the survey; no information was collected
regarding portion size or preparation methods.
2-5
-------
In their review of the literature, NRC found neurodevelopmental effects to be the most
sensitive endpoints and appropriate for establishing an RfD (NRC, 2000). Three large-scale
epidemiological studies have examined the effects of low dose prenatal mercury exposure and
neurodevelopmental outcomes through the administration of numerous tests of cognitive
functioning. These studies were conducted in the Faroe Islands (Grandjean et al. 1997), New
Zealand (Kjellstrom et al. 1989, Crump et al. 1998), and the Seychelles Islands (Davidson et al.
1998, Myers et al. 2003). The NRC noted that deficiencies of the magnitude observed in those
studies were likely to be associated with difficulty with vocabulary, verbal learning, attention,
and motor functions (NRC 2000). The NRC also concluded that children exposed at the levels
reported in those studies are likely to struggle to keep up in class and may need special
education, or other remedial help with school. Studies involving animals found sensory effects
and support the conclusions reached by studies involving human subjects, with a similar range of
neurodevelopmental effects reported (NRC 2000). As noted by the NRC, the clinical
significance of some of the more subtle endpoints included in the human low-dose studies is
difficult to gauge due to the quantal nature of the effects observed (i.e., subjects either display
the abnormality or do not) and the rather low occurrence rate of these effects.
Little is known about the effects of low level chronic methylmercury exposure in children
that can be linked to exposures after birth. The difficulty in identifying a cohort exposed after
birth but not prenatally, or separating prenatal from postnatal effects, makes research on the topic
complicated. These challenges were present in the three large epidemiologic studies used to
derive the RfD, as in all three studies there was postnatal exposure as well.
Several studies have also examined the effects of chronic low-dose methylmercury
exposures on adult neurological and sensory functions (e.g., Lebel et al. 1996, Lebel et al. 1998,
Beuter and Edwards 1998). Research results suggest that elevated hair methylmercury
concentrations (i.e., up to 50 ppm, and possibly as low as 20 ppm, though the NOAEL was not
always be clearly estimated) in individuals are associated with visual deficits, including loss of
peripheral vision and chromatic and contrast sensitivity. These individuals also exhibited a loss
of manual dexterity, hand-eye coordination, and grip strength; difficulty performing complex
sequences of movement; and (at the higher doses) tremors, although expression of some effects
was sex-specific. Although additional data would be needed to quantify a dose-response
relationship for these effects, it is noteworthy that the effects occurred at doses lower than the
Japanese and Iranian poisoning episodes, via consumption of mercury-laden fish in riverine
Brazilian communities (where extensive mercury contamination has resulted from small-scale
gold mining activities begun in the 1980s); however these doses are above the EPA's RfD
equivalent level for hair mercury. In regard to the Lebel et al. (1998) study, NRC states that "the
mercury exposure of the cohort is presumed to have resulted from fish-consumption patterns that
are stable and thus relevant to estimating the risk associated with chronic, low-dose
methylmercury exposure" (NRC 2000). NRC noted, however, "that the possibility cannot be
excluded that the neurobehavioral deficits of the adult subjects were due to increased prenatal,
rather than ongoing, MeHg exposure." More recent studies in the Brazilian communities
provide some evidence that the adverse neurobehavioral effects may in fact result from postnatal
exposures (e.g., Yokoo et al. 2003); however, additional longitudinal study of these and other
populations is required to resolve questions regarding exposure timing and fully characterize the
potential neurological impacts of methylmercury exposure in adults.
2-6
-------
2.5 Cardiovascular Impacts
While important, the weight of evidence for cardiovascular effects is not as strong as it is
for childhood neurological effects and the state of the science is still being evaluated. However,
in some recent epidemiological studies in men, methylmercury exposure is associated with a
higher risk of acute myocardial infarction, coronary heart disease and cardiovascular disease in
some populations (e.g. Salonen et al. 1995; Guallar et al.2002). Other recent studies have not
observed this association (e.g. Yoshizawa et al. 2002; Hallgren et al. 2001). The studies that
have observed an association suggest that the exposure to methylmercury may attenuate the
beneficial effects offish consumption. Studies investigating the relationship between
methylmercury exposure and cardiovascular impacts have reached different conclusions. The
findings to date and the plausible biologic mechanisms warrant additional research in this arena
(Stern 2005; Chan and Egeland 2004).
The potential for adverse cardiovascular effects due to consumption offish containing
methylmercury is of particular interest given the evidence for the protective cardiovascular effect
believed to occur from an increased dietary fish intake. Strong evidence indicates that
consumption offish, particularly fatty fish, has a cardio-protective effect (Wang et al. 2004;
2005 Dietary Guidelines Advisory Committee 2004; NRC 2000, Kris-Etherton et al. 2002).
Thus, consumption offish containing methylmercury is not necessarily detrimental even though
some evidence suggests that the cardiovascular system may be a target system for
methylmercury exposure.
A more robust discussion of multiple studies evaluating the association between
methylmercury exposure via fish consumption and acute myocardial infarction and other
cardiovascular effects as well as a description of the association between fish consumption and
cardioprotective effects is presented in Appendix B.
2.6 Genotoxic Effects
The NRC concluded that evidence that human exposure caused genetic damage is
inconclusive. However, in one recent study of adults living in the Tapajos River region in Brazil
Amorim et al. (2000) reported a direct relationship between methylmercury concentration in hair
and cytogenetic damage in lymphocytes, with polyploidal aberrations and chromatid breaks
observed at mercury hair levels around 7.25 ppm and 10 ppm, respectively. Long-term
methylmercury exposures in this population were believed to occur through consumption of
fish, suggesting that cytotoxic effects may result from dietary, chronic methylmercury exposures
similar to and above those seen in the Faroes and Seychelles populations.
2.7 Immunotoxic Effects
Although exposure to some forms of mercury can result in a decrease in immune activity
or an autoimmune response (ATSDR 1999), evidence for immunotoxic effects of methylmercury
is scarce (NRC 2000). However, a recent study offish-consuming communities in Amazonian
Brazil has identified a possible association between methylmercury exposure and immunotoxic
effects, although the authors noted that this may reflect interactions with infectious disease and
other factors (Silva et al. 2004). Exposures to these communities occurred via fish consumption
2-7
-------
(some community members were also exposed to inorganic mercury through gold mining
activities). The researchers assessed levels of specific antibodies that are markers of mercury-
induced autoimmunity. They found that both prevalence and levels of these antibodies were
higher in a population exposed to methylmercury via fish consumption compared to a reference
(unexposed) population. Median hair mercury concentration was 8 ppm in the more exposed
population (range 0.29-58.47 ppm) and 5.57 ppm in the less exposed reference population (range
1.19-16.96 ppm). The ranges of mercury hair concentrations reported in this study are within an
order of magnitude of the concentration corresponding to the methylmercury RfD. Overall, there
is a relatively small body of evidence from human studies that suggests exposure to
methylmercury can result in immunotoxic effects.
2.8 Other Human Toxicity Data
Based on limited human and animal data, methylmercury is classified as a "possible"
human carcinogen by the International Agency for Research on Cancer (IARC 1994) and in the
Integrated Risk Information System (IRIS) (EPA 2002). The existing evidence supporting the
possibility of carcinogenic effects in humans from low-dose chronic exposures is tenuous.
Multiple human epidemiological studies have found no significant association between mercury
exposure and overall cancer incidence, although a few studies have shown an association
between mercury exposure and specific types of cancer incidence (e.g., acute leukemia and liver
cancer; NRC 2000). The MSRC observed that "Methylmercury is not likely to be a human
carcinogen under conditions of exposure generally encountered in the environment" (p 6-16, Vol
V). This was based on observation that tumors were noted in one species only at doses causing
sever toxicity to the target organ. While some of the human and animal research suggests that a
link between methylmercury and cancer may plausibly exist, more research is needed.
There is also some evidence of reproductive and renal toxicity in humans from
methylmercury exposure. For example, a smaller than expected number of pregnancies were
observed among women exposed via contaminated wheat in the Iraqi poisoning episode of 1956
(Bakir et al. 1973); other victims of that same poisoning event exhibited signs of renal damage
(Jalili and Abbasi 1961); and an increased incidence of deaths due to kidney disease was
observed in women exposed in Minamata Bay via contaminated fish (Tamashiro et al. 1986).
Other data from animal studies suggest a link between methylmercury exposure and similar
reproductive and renal effects, as well as hematological toxicity (NRC 2000). Overall, human
data regarding reproductive, renal, and hematological toxicity from methylmercury are very
limited and are based on either studies of the two high-dose poisoning episodes in Iraq and Japan
or animal data, rather than epidemiological studies of chronic exposures at the levels of interest
in this analysis. Note that the U.S. EPA Mercury Study Report to Congress provides an
assessment of methylmercury cancer risk using the 1993 version of the Revised Cancer
Guidelines. For hazard identification, these are similar to the current EPA revisions.
2.9 Ecological Effects
Deposition of mercury to water bodies can also have an impact on ecosystems and
wildlife. While the benefit of further reducing mercury emissions cannot be quantified for
ecosystems at this time, we find it useful to qualitatively describe this benefit for context.
2-8
-------
Mercury contamination is present in all environmental media with aquatic systems
experiencing the greatest exposures due to bioaccumulation. Bioaccumulation refers to the net
uptake of a contaminant from all possible pathways and includes the accumulation that may
occur by direct exposure to contaminated media as well as uptake from food. Elimination of
methylmercury from fish is so slow that long-term reductions of mercury concentrations in fish
are often due to growth of the fish ("growth dilution"), whereas other mercury compounds are
eliminated relatively quickly. Piscivorous avian and mammalian wildlife are exposed to mercury
mainly through the consumption of contaminated fish and, as a result, accumulate mercury to
levels greater than those in their prey items (EPA 1997).
Numerous studies have generated field data on the levels of mercury in a variety of wild
species. Many of the data from these environmental studies are anecdotal in nature rather than
representative or statistically designed studies. The body of work examining the effects of these
exposures is growing but still incomplete given the complexities of the natural world. A large
portion of the adverse effect research conducted to date has been carried out in the laboratory
setting rather than in the wild; thus, conclusions about overarching ecosystem health and
population effects are difficult to make at this time. Nevertheless, numerous adverse effects
have been identified. Further reducing the presence of mercury in the environment may help to
alleviate the potential for adverse ecological health outcomes.
A full discussion of potential ecosystem effects updated since the 1997 Mercury Report
to Congress is provided in Appendix C.
2.10 Conclusions
In summary:
• Children who are exposed to low concentrations of methylmercury prenatally
may be at risk of poor performance on neurobehavioral tests, such as those
measuring attention, fine motor function, language skills, visual-spatial abilities
and verbal memory.
• Some recent epidemiological studies in men suggest that methylmercury is
associated with a higher risk of acute myocardial infarction, coronary heart
disease and cardiovascular disease in some populations. Other recent studies
have not observed this association. The studies that have observed an association
suggest that the exposure to methylmercury may attenuate the beneficial effects of
fish consumption. The findings to date and the plausible biologic mechanisms
warrant additional research in this arena (Stern 2005; Chan and Egeland 2004).
The exposure levels at which neurological effects have been observed may occur
via consumption offish (rather than high-dose poisoning episodes). Exposure
levels of concern for these effects are generally within two orders of magnitude of
typical exposures for women of child-bearing age based on NHANES data, and
within approximately an order of magnitude of the high end of the US exposure
distribution.
2-9
-------
• There is some recent evidence that exposures of methylmercury may result in
genotoxic or immunotoxic effects. Other research with less corroboration suggest
that reproductive, renal, and hematological impacts may be of concern. There are
insufficient human data to evaluate whether these effects are consistent with
levels in the U.S. population.
Plant and aquatic life, as well as fish, birds, and mammalian wildlife can be
affected by mercury exposure, however overarching conclusions about ecosystem
health and population effects are difficult to make at this time.. Ecological effects
are discussed in greater detail in Appendix C.
2.11 References
2005 Dietary Guidelines Advisory Committee, August, 2004. Report of the 2005 Dietary
Guidelines Advisory Committee.
http://www.health.gov/dietaryguidelines/dga2005/default.htm
http://www.health.gov/dietaryguidelines/dga2005/report/
Agency for Toxic Substances and Disease Registry (ATSDR). 1999. Toxicological Profile for
Mercury. U.S. Department of Health and Human Services, Public Health Service,
Atlanta, GA.
Amorim, M.I.M., D. Mergler, M.O. Bahia, H. Dubeau, D. Miranda, J. Lebel, R.R. Burbano, and
M. Lucotte. 2000. Cytogenetic damage related to low levels of methyl mercury
contamination in the Brazilian Amazon. An. Acad. Bras. Cienc. 72(4): 497-507.
Bakir, F., S.F. Damluji, L. Amin-Zaki, M. Murtadha, A. Khalidi, N.Y. al-Rawi, S. Tikriti, H.I.
Dhahir, T.W. Clarkson, J.C. Smith, and R.A. Doherty. 1973. Methylmercury poisoning in
Iraq. Science. 181(96):230-241 (as cited inNRC 2000).
Beuter, A., and R. Edwards. 1998. Tremor in Cree subjects exposed to methylmercury: a
preliminary study. Neurotoxicol. Teratol. 20(6):581-9.
Centers for Disease Control, Blood Mercury Levels in Young Children and Childbearing-Aged
Women -UnitedStates, 1999-2002, MMWR Morb Mortal Wkly Rep. 2004 Nov
5:53(43): 1018-1020. http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5343a5.htm
Chan, H.M. and G.M. Egeland. 2004. Fish Consumption, Mercury Exposure, and Heart
Disease. Nutrition Reviews. 62(2): 68-72.
Crump, K.S., T. Kjellstrom, A.M. Shipp, A. Silvers, and A. Stewart. 1998. Influence of prenatal
mercury exposure upon scholastic and psychological test performance: benchmark
analysis of a New Zealand cohort. Risk Anal. 18(6):701-713.
Davidson, P.W., G.J. Myers, C. Cox, C. Axtell, C. Shamlaye, J. Sloane-Reeves, E. Cernichiari,
L. Needham, A. Choi, Y. Wang, M. Berlin, and T.W. Clarkson. 1998. Effects of prenatal
and postnatal methylmercury exposure from fish consumption on neurodevelopment:
2-10
-------
outcomes at 66 months of age in the Seychelles Child Development Study. JAMA.
280(8):701-707.
Grandjean, P., K. Murata, E. Budtz-Jorgensen, and P. Weihe. 2004. Autonomic Activity in
Methylmercury Neurotoxicity: 14-Year Follow-Up of a Faroese Birth Cohort. J. Pediatr.
144:169-76.
Guallar, E., M.I. Sanz-Gallardo, P. van't Veer, P. Bode, A. Aro, J. Gomez-Aracena, J.D. Kark,
R.A. Riemersma, J.M. Martin-Moreno, and F.J. Kok; Heavy Metals and Myocardial
Infarction Study Group. 2002. Mercury, fish oils, and the risk of myocardial infarction.
N Engl J Med. 347(22): 1747-54.
Hallgren CG, Hallmans G, Jansson J-H, Marklund SL, Huhtasaari F, Schutz A, Stromberg U,
Vessby B, and Skerfving S. 2001. Markers of high fish intake are associated with
decreased risk of a first myocardial infarction. British Journal of Nutrition 86:397-404.
International Agency for Research on Cancer (I ARC). 1994. I ARC Monographs on the
Evaluation of Carcinogenic Risks to Humans and their Supplements: Beryllium,
Cadmium, Mercury, and Exposures in the Glass Manufacturing Industry. Vol. 58.
Jalili, H.A., and A.H. Abbasi. 1961. Poisoning by ethyl mercury toluene sulphonanilide. Br. J.
Indust. Med. 18(Oct.):303-308 (as cited in NRC 2000).
Kjellstrom, T., P. Kennedy, S. Wallis, A. Stewart, L. Friberg, B. Lind, P. Witherspoon, and C.
Mantell. 1989. Physical and mental development of children with prenatal exposure to
mercury from fish. Stage 2: Interviews and psychological tests at age 6. National Swedish
Environmental Protection Board Report No. 3642.
Kris-Etherton, P.M., W.S. Harris, and L.J. Appel. 2002. Fish consumption, fish oil, omega-3
fatty acids, and cardiovascular disease. Circulation. 106(21): 2747-2757.
Lebel, J., D. Mergler, M. Lucotte, M. Amorim, J. Dolbec, D. Miranda, G. Arantes, I. Rheault,
and P. Pichet. 1996. Evidence of early nervous system dysfunction in Amazonian
populations exposed to low-levels of methylmercury. Neurotoxicology. 17(1):157-167.
Lebel, J., D. Mergler, F. Branches, M. Lucotte, M. Amorim, F. Larribe, and J. Dolbec. 1998.
Neurotoxic effects of low-level methylmercury contamination in the Amazonian Basin.
Environ. Res. 79(l):20-32.
Mahaffey, K.R., R.P. Clickner, and C.C. Bodurow. 2004. Blood Organic Mercury and Dietary
Mercury Intake: National Health and Nutrition Examination Survey, 1999 and 2000.
Environ Health Perspect 112:562-570.
Marsh, D.O., T.W. Clarkson, C. Cox, et al. 1987. Fetal methylmercury poisoning: relationship
between concentration in single strands of maternal-hair and child effects. Arch. Neurol.
44:1017-1022. (as cited in EPA 2002 IRIS documentation.)
2-11
-------
McDowell, M.A., C.F. Dillon, J. Osterloh, P.M. Bolger, E. Pellizzari, R. Fernando, R. Monies de
Oca, S.E. Schober, T. Sinks, R.L. Jones, and K.R. Mahaffey. 2004. Hair mercury levels
in U.S. children and women of childbearing age: Reference range data from NHANES
1999-2000. Environmental Health Perspectives. 112(11):1165-1171.
Myers, G.J., P.W. Davidson, C. Cox, C.F. Shamlaye, D. Palumbo, E. Cernichiari, J. Sloane-
Reeves, G.E. Wilding, J. Kost, L.S. Huang, and T.W. Clarkson. 2003. Prenatal
methylmercury exposure from ocean fish consumption in the Seychelles child
development study. Lancet. 361(9370):1686-92.
National Research Council (NRC). 2000. Toxicological Effects of Methylmercury. Committee
on the Toxicological Effects of Methylmercury, Board on Environmental Studies and
Toxicology, Commission on Life Sciences, National Research Council. National
Academy Press, Washington, DC.
Salonen J.T., K. Seppanen, K. Nyyssonen, H. Korpela, J. Kauhanen, M. Kantola, J. Tuomilehto,
H. Esterbauer, F. Tatzber, and R. Salonen. 1995. Intake of mercury from fish, lipid
peroxidation, and the risk of myocardial infarction and coronary, cardiovascular, and any
death in eastern Finnish men. Circulation. 91:645-655.
Silva IA, J.F. Nyland, A. Gorman, A. Perisse, A.M. Ventura, B.C. Santos, J.M. de Souza, C.L.
Burek , N.R. Rose, and E.K. Silbergeld. 2004. Mercury exposure, malaria, and serum
antinuclear/antinucleolar antibodies in amazon populations in Brazil: a cross-sectional
study. Environ Health. 3(1):11.
Stern AH. 2005. A review of the studies of the cardiovascular health effects of methylmercury
with consideration of the suitability for risk assessment. Environmental Research
98(1):133-142.
Tamashiro, H., M. Arakaki, M. Futatsuka, and E.S. Lee. 1986. Methylmercury exposure and
mortality in southern Japan: A close look at causes of death. J. Epidemiol. Community
Health. 40(2):181-185 (as cited in NRC 2000).
United Nations Environmental Programme (UNEP). 2002. Global Mercury Assessment.
December. UNEP Chemicals, part of UNEP's Technology, Industry and Economics
Division.
U.S. Environmental Protection Agency (EPA). 1997. Mercury Study Report to Congress.
Volume V: Health Effects of Mercury and Mercury Compounds. EPA-452/R-97-007.
U.S. EPA Office of Air Quality Planning and Standards, and Office of Research and
Development.
U.S. Environmental Protection Agency (EPA). 2002 (date of most recent revision of on-line
materials; website accessed January 2005). Integrated Risk Information System (IRIS).
Methylmercury. U.S. EPA Office of Research and Development, National Center for
Environmental Assessment. Oral RfD and inhalation RfC assessments last revised
2-12
-------
7/27/2001. . Carcinogenicity assessment last revised 5/1/1995. Available online at
http://www.epa.gov/iris/subst/0073.htm
Wang C, Chung M, Lichtenstein A, Balk E, Kupelnick B, DeVine D, Lawrence A, Lau J. 2004.
Effects of Omega-3 Fatty Acids on Cardiovascular Disease. Summary, Evidence
Report/Technology Assessment No. 94. (Prepared by the Tufts-New England Medical
Center Evidence-based Practice Center, Boston, MA.) AHRQ Publication No. 04-E009-
1. Rockville,MD: Agency for Healthcare Research and Quality. March 2004.Agency for
Healthcare Research and Quality (AHRQ), DHHS March, 2004. Omega-3 Fatty Acids
Effects on Cardiovascular Disease, http://www.ahrq.gov/clinic/epcindex.htmSdietsup
Yokoo, E.M., J.G. Valente, L. Grattan, S.L. Schmidt, I. Platt, E.K. Silbergeld. 2003. Low level
methylmercury exposure affects neuropsychological function in adults. Environ. Health.
2003 Jun 04;2(1):8.
Yoshizawa, K., E.B. Rimm, S. Morris, V.L. Spate, C-C. Hsieh, D. Spiegelman, M.J. Stampfer,
and W.C. Willett. 2002. Mercury and the risk of coronary heart disease in men. N Engl J
Med. 347:1755-1760.
2-13
-------
SECTION 3 ECOSYSTEM SCALE MODELING FOR MERCURY BENEFITS ANALYSIS-1
Executive Summary 3-1
3.1 Introduction ~ Rule Background 3-4
3.1.1 Use of Mercury Maps (MMaps) to Project Changes in Fish Tissue
Concentrations 3-5
3.1.2 Goal/Purpose of Ecosystem Case Studies 3-9
3.2 Recent Advances in Mercury Science 3-10
3.2.1 Mercury Cycle Chemistry 3-10
3.2.2 Mercury Processes in the Atmosphere 3-10
3.2.3 Mercury Processes in Soils 3-11
3.2.4 Mercury Processes in Water 3-12
3.2.5 Unavailability of Inorganic Mercury to Methylating Microbes 3-12
3.2.6 Mercury Accumulation in the Food Web 3-14
3.2.7 Summary of Findings in the METAALICUS Study 3-14
3.2.8 Summary of Florida Everglades Study 3-15
3.3 Overview of Models Used in This Study 3-16
3.3.1 Atmospheric Models 3-16
3.3.2 Ecosystem Models 3-17
3.4 Overview of Case Studies 3-21
3.4.1 Ecosystem Characteristics 3-22
3.4.2 Baseline Atmospheric Deposition at Each Site 3-24
3.4.3 Atmospheric Loading Scenarios Investigated 3-24
3.4.4 Summary of Model Evaluation 3-25
3.4.5 Baseline Fish Mercury Concentrations 3-27
3.4.6 Magnitude of Changes in Fish Tissue Residues 3-30
3.4.7 Summary of Observed Temporal Responses to Declines in Loading . 3-32
3.4.8 Effect of Land Uses Changes 3-34
3.4.9 Summary 3-36
3.5 National Scale Ecosystem Variability 3-37
3.5.1 United States Lakes Distribution 3-37
3.5.2 Summary 3-39
3.6 References 3-46
-------
Tables
Table 3- 1. Comparison of SERAFM and IEM-2M Forecasted Mercury Concentrations Using
Parameter Values for Model Ecosystem Described in the Mercury Study Report to
Congress (RtC) and a 50% Reduction in Atmospheric Deposition 3-19
Table 3-2. Summary of Ecosystem Characteristics Used To Parameterize Mercury Models 3-23
Table 3- 3. Baseline Atmospheric Deposition For Each Model Ecosystem 3-24
Table 3- 4. Forecasted Atmospheric Deposition Rates in Case Study Areas Using the CMAQ
and REMSAD Models 3-25
Table 3- 5. List of Model Frameworks Applied to Ecosystems 3-26
Table 3- 6. Summary of Mercury Parameters Used in the SERAFM Model 3-27
Table 3- 7. Empirically Derived BAFs for Each of the Ecosystem Case Studies 3-27
Table 3-8. MMaps and SERAFM Forecasted Fish Mercury Concentration at Steady State after
Removal of Coal Fired Utilities as a Component of Deposition Using the REMSAD and
CMAQ Models (Zero-out Scenario) 3-31
Table 3- 9. Sediment Response Times in Years to Reach 90% of Steady-state Concentrations
Following 50% Mercury Deposition Reductions 3-32
Table 3- 10. Fish Tissue Response Times in Years to Reach 90% of Steady-state Concentrations
Following 50% Mercury Deposition Reductions 3-33
Table 3-11. Frequency of Different Lake Sizes Across the United States 3-38
Figures
Figure 3- 1. BASS Predicted BAFs for Pike/Perch in Lee Dam, Eagle Butte 3-28
Figure 3- 2. Observed vs. Predicted Fish Mercury Concentrations in Model Ecosystems at
Steady State with No Change in Atmospheric Loading 3-29
Figure 3- 3. Temporal Response of Mercury Concentrations in Fish from Pawtuckaway Lake,
NH to a Decline in Mercury Loading 3-30
Figure 3- 4. Temporal Response of Mercury Concentrations in Fish from Lake Barco, FL to a
Decline in Mercury Loading 3-31
Figure 3- 5. Projected Ecosystem Response Times to Zero-out Deposition Scenario Using the
SERAFM Model 3-33
Figure 3- 6. Upper Brier Creek Loading Flux Attenuation 3-35
Figure 3- 7. Watershed Loading Flux Attenuation Considering Land-use Change 3-35
Figure 3- 8. Model Ecosystem Locations with CMAQ Grid Cells and Locations of Electricity
Generating Units (EGUs) 3-40
Figure 3- 9. Model Ecosystem Locations with Gradient in Sulfate Deposition Across the Eastern
US 3-41
Figure 3- 10. Model Ecosystem Locations with Percent Wetland Area Aggregated for Each
HUC 3-42
Figure 3-11. Model Ecosystem Locations with CMAQ 2001 Total Mercury Deposition .. 3-43
Figure 3- 12. Model Ecosystem Locations with Measured Fish Tissue Concentrations .... 3-44
Figure 3-13. Measured 1995-2001 Fish Hg Concentrations > 0.3 ppm 3-45
-------
SECTION 3
ECOSYSTEM SCALE MODELING FOR MERCURY BENEFITS ANALYSIS
Executive Summary
In the United States, humans are exposed to methylmercury (MeHg) mainly by
consuming fish that contain MeHg. Aquatic ecosystems respond to changes in mercury
deposition in a highly variable manner as a function of differences in their chemical, biological
and physical properties. Depending on the characteristics of a given ecosystem, methylating
microbes convert a small but variable fraction of the inorganic mercury in the sediments and
water derived from human activities and natural sources into MeHg. Methylmercury is the only
form of mercury that biomagnifies in the food web. Concentrations of MeHg in fish are
generally on the order of a million times the MeHg concentration in water. In addition to
mercury deposition, key factors affecting MeHg production and accumulation in fish include the
amount and forms of sulfur and carbon species present in a given waterbody. Thus, two
adjoining water bodies receiving the same deposition can have significantly different fish
mercury concentrations.
For the Utility Mercury Reduction Benefits Assessment, EPA applied the Mercury Maps
(MMaps) model to estimate changes in freshwater fish mercury concentrations resulting from
changes in mercury deposition after regulation of mercury emissions from U.S. coal-fired power
plants. MMaps, a simplified form of the IEM-2M model applied in EPA's 1997 Mercury Study
Report to Congress, is a static model that assumes a proportional relationship between declines
in atmospheric mercury deposition and concentrations in fish at steady state. This means, for
example, that a 50% decrease in mercury deposition rates is projected to lead to a 50% decrease
in mercury concentrations in fish. MMaps does not consider the dynamics of relevant ecosystem
specific factors that can affect the methylation and bioaccumulation in fish in different water
bodies over time, nor does it consider the inputs of non-air sources to the watershed. In all cases,
the MMaps model does not address the lag time of different ecosystems to reach steady state
(i.e., when fish mercury concentrations reflect changes in atmospheric deposition). In addition,
applying the MMaps model assumes that atmospheric deposition is the principle source of
mercury to the waterbodies being investigated and environmental factors that affect MeHg
production and accumulation in organisms will remain constant, allowing each ecosystem to
reach steady state. While MMaps has several limitations, EPA knows of no alternative tool for
performing a national-scale assessment of such changes.
The objectives of this chapter are to: (a) provide information on the response times of
different ecosystems to declines in mercury deposition, and (b) characterize some of the key
sources of uncertainty around the proportional relationship used by the MMaps model. To do
this, EPA applied dynamic, ecosystem scale waterbody, watershed, and bioaccumulation models
to five freshwater systems spanning a range of types across the United States. We used model
results to investigate the magnitude and timing of changes in fish mercury concentrations
associated with changes in atmospheric mercury deposition after regulation of coal-fired power
plants. While important advances have been made in recent years to enhance scientific
understanding of the behavior of mercury in the environment, our ability to effectively model the
range in response times for different systems is constrained by our limited knowledge of how
3-1
-------
methylation and bioaccumulation occur in various ecosystems. Because of these uncertainties,
no modeling framework can be considered a priori predictive of ecosystem responses at this
time. In recognition of the above, we calibrated models applied in the ecosystem case studies to
monitoring data from each of the specific ecosystems studied. In addition, when choosing the
locations of these case studies it was essential to rely on well-studied ecosystems where there
were sufficient empirical data available to parameterize ecosystem scale models.
Sites investigated can be characterized as follows:
(1) Small, southern seepage lake with negligible watershed (Lake Barco, FL);
(2) Watershed dominated, coastal plain river (Brier Creek, GA);
(3) Large, shallow, well-mixed southern lake (Lake Waccamaw, NC);
(4) Medium sized, stratified seepage lake with a moderate sized drainage basin in the
Northeast (Pawtuckaway Lake, NH); and
(5) Shallow, well-mixed farm pond in the Midwest (Eagle Butte, SD).
When considering other ecosystem variables that may affect MeHg production (e.g.,
sulfate deposition, percent wetland coverage, and organic carbon), these case studies represent
ecosystem types of moderate methylation potential across the United States. Fish tissue
concentrations and atmospheric deposition rates measured in these regions also do not represent
the extremes observed on a national scale. Therefore, while these ecosystem case studies cover
the bulk of the distributions of the key environmental characteristics that will affect MeHg
production, they may miss the tails of the distributions for some characteristics.
For each of the above system types, we characterized a range of response times by
varying key parameters in the modeling scenarios known to drive the temporal response. Case
studies of individual ecosystems show that the time necessary for aquatic systems to reach a new
steady state after a reduction in mercury deposition rates can be as short as 5 years or as long as
50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades. Forecasted response times to changes in mercury inputs were
longest for Brier Creek, a system strongly influenced by the watershed mercury loading, and
Pawtuckaway Lake, a stratified cold water lake that had significant watershed mercury inputs.
Shallow, well-mixed systems like Lake Waccamaw and Lee Dam, which receive most of their
mercury inputs from the atmosphere are projected to respond to changes in atmospheric
deposition in less than a decade, although watershed loading dynamics could introduce
significant lag times in reaching the full response. These findings are consistent with those
observed in the Florida Everglades, which can be characterized as a shallow, well-mixed,
dynamic system and accordingly shows a measurable response within a decade. Results from
Brier Creek, the watershed dominated system, qualitatively concur with findings from the
METALLICUS study showing a much longer response time for mercury deposited in the
watershed than direct atmospheric deposition to the surface of a waterbody.
Overall, we conclude that the most likely appropriate response times for freshwater
ecosystems to be considered in the national scale assessment range between five and 30 years,
while recognizing that some systems will likely take more than 50-100 years to reach steady
state. This assessment is based on the "medium" or moderate estimates across the several
system types considered in this study. Because our modeling scenarios include two extremes of
3-2
-------
rapidly and slowly responding system types (e.g., a watershed dominated system and warm,
shallow, well-mixed systems), we expect that the range in responsiveness of freshwater bodies in
the United States will be captured within the "fast, medium and slow" scenarios presented in the
report. One additional uncertainty in our calculations related to preliminary results from the
METAALICUS study showing newly deposited mercury is converted to methylmercury more
rapidly than legacy mercury. These results, if extrapolated to other freshwater sites, imply that
the response time of some freshwater ecosystems may be more rapid than predicted by our best
available models at this time.
To appreciate the importance of the adjustment time for the benefits analysis, consider
the following scenario: Suppose that (a) current benefits are proportional to the reduction in fish
tissue concentrations from baseline levels, (b) fish tissue concentrations decrease exponentially
to eventually reach 90% of the total reduction, and (c) the "adjustment time" is measured as the
time required to reach 90% of the eventual total reduction. Under these conditions, using
discount rates between 3% and 7% the net present values (NPV) of benefits if the adjustment
times were 5 years would be between 1.6 and 2.1 times the NPV if the adjustment time were 50
years, even if the same eventual reduction were reached.
To investigate some of the key sources of uncertainty around the proportional
relationship used by the MMaps model, we modeled the magnitudes of expected changes in fish
mercury concentrations locations after removal of utilities as an emissions source. At each of
the case study locations considered, removal of coal fired utilities as a source of mercury
reduced atmospheric deposition rates between four and fifteen percent. These values are
relatively small in magnitude compared to maximum values observed across the country where
the maximum model forecasted difference in deposition in this "zero-out" scenario exceeds 70%.
To place this in context, this rule is expected to reduced total mercury emissions in the U.S. by
up to 70% for the "cap and trade" alternative. At the locations considered in this study, EPA
estimates that the reductions in emissions associated with the cap and trade alternative will
eventually reduce loading rates by approximately 10% from current conditions (see Table 3-8
below). This decline in deposition can be contrasted with areas most highly affected by the
proposed regulation, where declines in deposition are more likely to be on the order of 50%
according to modeling projections. Accordingly, we modeled a 50% decline in atmospheric
loading in all of the case studies to investigate responses of similar ecosystem types to large
declines in mercury deposition.
Our analysis suggests that differences between results from the ecosystem scale
waterbody models and MMaps model across all sites are mainly a function of the initial fish
mercury concentrations at steady state. Because error bars around each species of piscivorous
fish investigated in the case studies from a single water body are large, we expect that on a
national scale, techniques used to normalize fish mercury data will be a major source of
uncertainty in the MMaps model. Accordingly, both the completeness of national fish tissue
monitoring data coverage and the statistical techniques used to normalize these data among
different trophic levels and ages will have a major impact on how well the MMaps approach
captures the true variability in fish mercury concentrations across the country. Ecosystem
models may be used to supplement limited fish tissue data with forecasted fish mercury
concentrations given other information on mercury concentrations and dynamics in a given
waterbody. Overall, it is clear that the magnitude of the uncertainty in the atmospheric and
3-3
-------
ecosystem models at this time is much greater than the signal derived from a change in loading
following removal of the coal fired utilities at the sites investigated in this report.
Another source of uncertainty in the MMaps forecasts are the atmospheric deposition
rates used to forecast changes in fish mercury concentrations. For each case study site,
deposition rates in the corresponding CMAQ and REMSAD grid cells were compared to
empirically derived loading rates. At the locations chosen for these case studies, site specific
data suggest somewhat higher deposition rates than the CMAQ and REMSAD models. This
would result in an overestimate of the relative change in atmospheric deposition and changes in
fish mercury concentration by the MMaps model. These findings reinforce the need for
additional data sets that can be used to test model-forecasted atmospheric mercury deposition
rates.
The effect of epistemic uncertainty (i.e., lack of knowledge) about key mercury process
variables, such as the functional form of equations used to quantify methylation rate constants, is
a major contributor to overall uncertainty in the MMaps and ecosystem models that cannot be
quantified at this time. In addition, a preliminary assessment of the expected effect of land-use
changes on fish mercury concentrations for a watershed dominated system illustrates changes
like urbanization within a watershed can alter the magnitude and timing of fish mercury
concentrations. Accordingly, EPA's Office of Research and Development views this study as
part of an iterative modeling exercise.
3.1 Introduction — Rule Background
As described in the NODA (FR 69864-69877, December 1, 2004), EPA's revised
benefits analysis estimates the extent to which adverse human health effects will be reduced as a
result of reducing mercury (Hg) emissions from coal-fired power plants. In the United States,
humans are exposed to methylmercury (MeHg) mainly by consuming fish that contain MeHg.
Accordingly, to estimate changes in human exposure EPA must analyze how changes in Hg
deposition from U.S. coal-fired power plants translate into changes in MeHg concentrations in
fish. This proposed rule is expected to reduce total mercury emissions in the U.S. by up to 70%
using the "cap and trade" program. In the case studies described later in this section, EPA
estimates that the reductions in emissions associated with the cap and trade program will
eventually reduce loading rates to the locations considered in this study by approximately 10%
from current conditions (see Table 3-4 below). This decline in deposition can be contrasted with
areas most highly affected by the proposed regulation, where declines in deposition are more
likely to be on the order of 50% according to modeling projections. Quantifying the linkage
between different levels of Hg deposition and fish tissue MeHg concentration is an important
step in the benefits methodology and the focus of the material described in this chapter.
To effectively estimate fish MeHg concentrations in a given ecosystem, it is important to
understand that the behavior of Hg in aquatic ecosystems is a complex function of the chemistry,
biology, and physical dynamics of different ecosystems. The majority (95 to 97 percent) of the
Hg that enters lakes, rivers, and estuaries from direct atmospheric deposition is in the inorganic
form (Lin and Pehkonen, 1999). Microbes convert a small fraction of the pool of inorganic Hg
in the water and sediments of these ecosystems into the organic form of Hg (MeHg). MeHg is
the only form of Hg that biomagnifies in organisms (Bloom, 1992). Ecosystem-specific factors
3-4
-------
that affect both the bioavailability of inorganic Hg to methylating microbes (e.g., sulfide,
dissolved organic carbon) and the activity of the microbes themselves (e.g., temperature, organic
carbon, redox status) determine the rate of MeHg production and subsequent accumulation in
fish (Benoit et al., 2003). The extent of MeHg bioaccumulation is also affected by the number of
trophic levels in the food web (e.g., piscivorous fish populations) because MeHg biomagnifies as
large piscivorous fish eat smaller organisms (Watras and Bloom, 1992; Wren and MacCrimmon,
1986). These and other factors can result in considerable variability in fish MeHg levels among
ecosystems at the regional and local scale.
3.1.1 Use of Mercury Maps (MMaps) to Project Changes in Fish Tissue Concentrations
To analyze the relationship between Hg deposition and MeHg concentrations in fish in
freshwater aquatic ecosystems across the U.S. for the national scale benefits assessment, EPA
applied EPA's Office of Water's Mercury Maps (MMaps) approach. MMaps implements a
simplified form of the IEM-2M model applied in EPA's Mercury Study Report to Congress
(USEPA, 1997). By simplifying the assumptions inherent in the freshwater ecosystem models
that were described in the Report to Congress, the MMaps model showed that these models
converge at a steady-state solution for MeHg concentrations in fish that are proportional to
changes in Hg inputs from atmospheric deposition (e.g., over the long term fish concentrations
are expected to decline proportionally to declines in atmospheric loading to a waterbody). This
solution only applies to situations where air deposition is the only significant source of Hg to a
water body, and the physical, chemical, and biological characteristics of the ecosystem remain
constant over time. EPA recognizes that concentrations of MeHg in fish across all ecosystems
may not reach steady state and that ecosystem conditions affecting mercury dynamics are
unlikely to remain constant over time. EPA further recognizes that many water bodies,
particularly in areas of historic gold and Hg mining in western states, contain significant non-air
sources of Hg. Finally, EPA recognizes that MMaps does not provide for a calculation of the
time lag between a reduction in Hg deposition and a reduction in the MeHg concentrations in
fish. Despite these limitations, EPA is unaware of any other tool for performing a national-scale
assessment of the change in fish MeHg concentrations resulting from reductions in atmospheric
deposition of Hg.
MMaps has several limitations:
1. The MMaps approach is based on the assumption of a linear, steady-state
relationship between concentrations of methylmercury in fish and present day air
deposition mercury inputs. We expect that this condition will likely not be met in
many waterbodies because of recent changes in mercury inputs and other
environmental variables that affect mercury bioaccumulation. For example, the
US has recently reduced human-caused emissions while international emissions
have increased.
2. The requirement that environmental conditions remain constant over the time
required to reach steady state inherent in the MMaps methodology may not be
met, particularly in systems that respond slowly to changes in mercury inputs.
3-5
-------
3. Many water bodies, particularly in areas of historic gold and mercury mining in
western States, contain significant nonair sources of mercury. MMaps
methodology cannot be applied to these waterbodies.
4. Finally, MMaps does not provide for a calculation of the time lag between a
reduction in mercury deposition and a reduction in the methylmercury
concentrations in fish.
The following paragraphs provide additional details on the above limitations, as well as a
brief assessment of the degree to which conditions match those assumptions. The MMaps model
(US EPA, 2001) assumes that for long-term steady state conditions, reductions in fish tissue
concentrations are expected to track linearly with reductions in air deposition watershed loads.
The MMaps model represents a reduced form of the IEM-2M and MCM models used in the
Mercury Study Report to Congress (USEPA, 1997), as well as the subsequent Dynamic MCM
(D-MCM) model (Harris et al., 1996). That is, the equations of these mercury fate and transport
models are reduced to steady state and consolidated into a single equilibrium equation equating
the ratio of future/current air deposition rates to future/current fish tissue concentrations. At
certain sites, the MMaps model has been shown to produce results equivalent to those of these
complex models over the long term, under a specific set of conditions.
Though plainly stated, the steady state assumption is a compilation of a number of
individual conditions. For example, fish tissue data may not represent average, steady state
concentrations for two major reasons:
Fish tissue and deposition rate data for the base period are not at steady state. Where
deposition rates have recently changed, the watershed or waterbody may not have had
sufficient time to fully respond. The pool of mercury in different media could be
sufficiently large relative to release rates, and thus needs more time to achieve a new
equilibrium. This is more likely to occur in deeper lakes and lakes with large catchments
where turnover rates are longer and where the watershed provides significant inputs of
mercury.
• Fish tissue data do not represent average conditions (or conditions of interest for forecast
fish levels). Methylation and bioaccumulation are variable and dynamic processes. If fish
are sampled during a period of high or low methylation or bioaccumulation, they would
not be representative of the average, steady state or dynamic equilibrium conditions of
the waterbody. This effect is significantly more pronounced in small and juvenile fish.
Examples include tissue data collected during a drought or during conditions offish
starvation. Other examples include areas in which seasonal fluctuations in fish mercury
levels are significant, due for example from seasonal runoff of contaminated soils from
abandoned gold and mercury mines or areas geologically rich in mercury. In such a case,
MMaps predictions would be valid for similar, conditions (e.g. wet year/dry year, or
season) in the future, rather than typical or average conditions. Alternatively, sufficient
fish tissue would need to be collected to get an average concentration that represents a
baseline dynamic equilibrium.
3-6
-------
Other ecosystem conditions might cause projections from the MMaps approach to be
inaccurate for a particular ecosystem. Watershed and waterbody conditions can undergo
significant changes in capacity to transport, methylate, and bioaccumulate mercury. Examples of
this include regions where sulfate and/or acid deposition rates are changing (in turn affecting
methymercury production independently of total mercury loading), and where the trophic status
of a waterbody is changing. A number of other water quality parameters have been correlated
with increased fish tissue concentrations (e.g. low pH, high DOC, lower algal concentrations),
but these relationships are highly variable among different waterbodies. MMaps will be biased
when waterbody characteristics change between when fish were initially sampled, and the new
conditions of the waterbody.
As is stated above, the relationship between the change in mercury deposition from air to
the change in fish tissue concentration holds only when air deposition is the predominant source
of the mercury load to a waterbody. Due to this requirement in the model, the national
application of the MMaps approach screened out watersheds in which sources of mercury other
than air deposition were significant.1 Therefore, fish tissue concentrations are assumed to
remain unchanged if located in watersheds that contain potentially significant nonpoint sources
such as: historic mercury mining locations (as a surrogate for mercury bedrock deposits)
(MAS/MILS database for mercury mines (US EPA, 200Ib); significant producer gold mines
(USGS Database for Significant Deposits; Long, et al, 1998); or mercury cell chlor-alkali
facilities (USEPA - PCS database). Watersheds are also screened from the analysis where the
sum estimated mercury loads from other sources, e.g. Publicly Owned Treatment Works
(POTW) exceeds an arbitrary level of significance (e.g. 5% of total load).
To the degree to which the applicability of the above conditions is unknown, MMaps is a
screening level estimate of the changes in fish tissue as a result of changes in air deposition rates.
Where these specific conditions do apply, the results from MMaps will be equivalent. The above
criteria, for assessing the validity of the steady state assumption, were used to evaluate this
benefits analysis application (in the same order as above):
Changes in sulfate deposition and waterbody sulfate concentration and pH. This has been
shown to be the cause for 50% of the reduction in fish tissue mercury levels in a lake in
Minnesota. Significant changes in the presence of riparian wetlands in the Adirondack
park in New York, thought due to the resurgence in beaver population since the turn of
the century, has led to a dramatic increase of mercury export from watersheds to
waterbodies. Increased development and urbanization is associated with depressed
bioaccumulation rates. Thus in some areas, the predictions using this approach will be
somewhat inaccurate due to other confounding factors.
• While the base year for deposition modeling was 2001, the bulk of the fish tissue data
was collected in the early 1990's. We know that emissions in 2001 were 50% of that in
1990. Given a short time lag (5 years) the available fish tissue would reflect these higher
emissions from 1990. Thus, projected changes in fish tissue concentrations will be
applied to higher fish tissue concentrations, and thus be somewhat higher than is actually
1 An alternative approach, presented in US EPA, 2001, allows for taking into account other significant sources where
loads from these sources can be quantified and are not expected to change with time.
3-7
-------
expected. In this benefits context, this mismatch between air deposition and fish tissue
data results in an underestimate of benefits.
• It is unknown to which degree the NLFA and NFTS data reflect those that are commonly
fished. While state monitoring programs now generally focus on areas of high fishing
pressure, early monitoring programs (prior to 1995) used to develop fish advisories
focused their sampling efforts on areas near industrial outfalls and agricultural runoff,
looking primarily for organochlorine (e.g. PCBs, DDT, chlordane) contaminants. For this
reason, only data collected after 1999 were used in the RIA. Use of the Wente
covariance model removes bias that might be introduced by fish samples by correcting
for variability in mercury concentrations attributable to differences in trophic level,
length and age from those species most typically consumed by humans.
It should be noted that MMaps was designed to address an important, but very specific
issue - that of eventual response offish tissue to air deposition reductions. As such it responds to
a need to understand how mercury reductions, independent of other changes in the environment,
will impact fish contamination and human health. More complex models are required in cases
where more complete descriptions are needed. A dynamic model is essential for modeling
waterbody recovery during the period in which waterbody response lags reductions in mercury
loads. A dynamic model is also essential for understanding seasonal fluctuations, as well as year-
to-year fluctuations due to meteorological variability. Finally, a more complex model would be
essential for assessing the impact of other watershed and water quality changes (e.g. erosion,
wetlands coverage, and acid deposition) that might affect mercury bioaccumulation in fish.
These complex models are used to derive the MMaps approach, and are themselves based on a
number of assumptions. The science of mercury fate and transport in the environment is an
actively evolving area of research (e.g. see US EPA, 2003c). While these assumptions are
considered reasonable given the state of the science of environmental modeling and mercury in
the environment, the validity of assumptions inherent in both the MMaps approach and dynamic
ecosystem scale models will need to be reevaluated as the science of mercury fate and transport
evolves.
The MMaps methodology was peer reviewed by a set of national experts in the fate and
transport of mercury in watersheds. While two reviewers felt it could be used to predict future
fish tissue concentrations, a third cautioned it should not be considered a robust predictor until
scientific data can be generated to validate the approach. Reviewers systematically identified a
set of implicit assumptions that compose the steady state assumption in the MMaps approach.
They pointed out that due to evolving and complex nature of the science of mercury, some
features of the complex models are assumptions themselves, and thus cannot be wholly relied
upon as ultimate predictors of mercury fate and transport. The reviewers pointed out that there is
limited scientific information to directly verify this approach, and that some scientific data
appears to refute individual components of the overall steady state assumption. One reviewer did
perform a D-MCM and MMaps comparison, and found that, under these assumptions, MMaps
model did produce comparable steady-state results as the D-MCM model. There was
considerable discussion about how best to aggregate the data, to scale up to a deposition
reduction requirement, from fish-specific and waterbody specific information. The peer review
report has not been released because the document that it relates to has not yet been approved for
3-8
-------
release by EPA. However, the description of the approach, and the methodologies as applied in
this RIA, are largely consistent with the peer review recommendations.
The MMaps report (US EPA, 2001) presented a national-scale application of Mercury
Maps to determine the percent reductions in air deposition that would be needed in watersheds
across the country for average fish tissue concentrations to achieve the national methylmercury
criterion. In this national scale assessment, fish tissue concentrations were aggregated at the
scale of large watersheds, thus presenting average results for each watershed. The use of other
scales of aggregation, e.g., waterbody specific, is consistent with the Mercury Maps approach to
the degree to which different mercury loads can be discerned.
3.1.2 Goal/Purpose of Ecosystem Case Studies
To supplement the MMaps methodology, this report explores the range in temporal
responses of different ecosystems following reductions in atmospheric Hg emissions and some
of the sources of uncertainty around the proportional relationship used by the MMaps model. To
do this, we provide quantitative examples from five case studies of a range of freshwater
ecosystem types across the Eastern and Midwestern United States. For all of these systems, we
applied an updated dynamic version of the IEM-2M model originally used in the Mercury Study
Report to Congress (USEPA, 1997) to forecast the time lag offish mercury concentrations in
each ecosystem to different atmospheric mercury input scenarios. We present our results in the
context of recent scientific findings related to the temporal response of different ecosystems and
the factors affecting accumulation of mercury in fish.
Because different ecosystems exhibit dramatically different responses to changes in
mercury loading depending on their chemical and physical attributes, results from individual
case studies must be qualified by their representativeness of ecosystem variability across the
United States. Using georeferenced empirical databases that describe some of the watershed and
waterbody characteristics across United States, we describe a preliminary assessment of the
variability in some factors known to be important for MeHg formation and bioaccumulation in
fish. Although this analysis has not been completed, the concept is demonstrated by a
preliminary qualitative assessment of how much of the ecosystem variability in MeHg formation
and bioaccumulation has been captured by the modeling case studies. Developing broad
categories of ecosystem types based on their propensity for MeHg formation and bioaccumation
in fish and their frequency of occurrence is an iterative effort. By combining the frequency of
each category of ecosystem type with the magnitude and time lag in fish tissue reductions
modeled using dynamic, ecosystem scale models, such an analysis could ultimately provide an
alternate methodology for a national scale assessment of expected changes in fish MeHg
concentrations resulting from reductions in atmospheric mercury deposition.
EPA acknowledges that present modeling capabilities do not allow a priori predictions of
ecosystem responses due to considerable uncertainties in the science. These epistemic
uncertainties limit the predictive power of both the MMaps approach and the models applied in
this exercise (see Peer Review Comments, Appendix 8). Because of these limitations, modeling
scenarios employed in this study are first calibrated to real ecosystem and rely heavily on
empirical data to develop credible rate constants and flux terms for each of the case studies
investigated. EPA's Office of Research and Development views this modeling exercise as part a
3-9
-------
series of iterative modeling development phases that will eventually allow us to achieve our goal
of a priori modeled responses. Accordingly, this report highlights advances in our modeling
capabilities since the publication of the Mercury Study Report to Congress in 1997. EPA will
continue to develop the models described in this report by incorporating the latest scientific
knowledge on the factors that control the distribution and accumulation of mercury in freshwater
and coastal marine food web. This goal is consistent with EPA's Mercury Research Strategy
(U.S. EPA, 2000).
3.2 Recent Advances in Mercury Science
This modeling exercise is based on our understanding of how mercury cycles through
ecosystems and accumulates in fish. The set of physical, chemical, and biological processes
controlling mercury fate in watersheds and water bodies can be synthesized into a general
conceptual model that guides our model selection, refinement, and application. These processes
can be grouped into specific categories: mercury cycle chemistry; mercury processes in the
atmosphere, soils and water; bioavailability of mercury in water; and mercury accumulation in
the food web. The following is a narrative of our conceptual model, discussing the recent
scientific developments that have added to our understanding of mercury processes. This review
builds upon the work previously summarized in EPA's Mercury Report to Congress (USEPA,
1997). The end of this section concludes with the conceptual model summary.
3.2.1 Mercury Cycle Chemistry
Mercury occurs naturally in the environment as several different chemical species. The
majority of mercury in the atmosphere (95-97%) is present in a neutral, elemental state (Hg°)
(Lin and Pehkonen, 1999), while in water, sediments and soils the majority of mercury is found
in the oxidized, divalent state (Hg(II)) (Morel et al., 1998). A small fraction (percent) of this
pool of divalent mercury is transformed by microbes into methylmercury (CH3Hg(II)/ MeHg)
(Jackson, 1998). Methylmercury is retained in fish tissue and is the only form of mercury that
biomagnifies in aquatic food webs (Kidd et al., 1995). As a result, methylmercury
concentrations in higher trophic level organisms such as piscivorous fish, birds and wildlife are
often 104-106 times higher than aqueous methylmercury concentrations (Jackson, 1998).
Transformations among mercury species within and between environmental media result in a
complicated chemical cycle. Mercury emissions from both natural and anthropogenic sources
are predominantly as Hg(II) species and Hg° (Landis and Keeler, 2002; Seigneur et al., 2004).
Anthropogenic point sources of mercury consist of combustion (e.g., utility boilers, municipal
waste combustors, commercial/industrial boilers, medical waste incinerators) and manufacturing
sources (e.g., chlor-alkali, cement, pulp and paper manufacturing) (USEPA, 1997). Natural
sources of mercury arise from geothermic emissions such as crustal degassing in the deep ocean
and volcanoes as well as dissolution of mercury from geologic sources (Rasmussen, 1994).
3.2.2 Mercury Processes in the Atmosphere
The relative contributions of local, regional and long range sources of mercury to fish
mercury levels in a given water body are strongly affected by the speciation of natural and
anthropogenic emissions sources. Elemental mercury is oxidized in the atmosphere to form the
more soluble mercuric ion (Hg(II)) (Schroeder et al., 1989). Particulate and reactive gaseous
3-10
-------
phases of Hg(II) are the principle forms of mercury deposited onto terrestrial and aquatic
systems because they are more efficiently scavenged from the atmosphere through wet and dry
deposition than Hg° (Lindberg and Stratton, 1998). Because Hg(II) species or reactive gaseous
mercury (ROM) and particulate mercury (Hg(p)) in the atmosphere tend to be deposited more
locally than Hg°, differences in the species of mercury emitted affect whether it is deposited
locally or travels longer distances in the atmosphere (Landis et al., 2004). Recent research
indicates that certain meteorological conditions and atmospheric constituents can result in the
rapid oxidation of Hg° to ROM, potentially increasing the fraction of Hg°from anthropogenic
sources that is deposited more locally (Landis, Pers. Comm., 2004).
Atmospheric models use various mathematical frameworks to describe how meteorology
and atmospheric chemistry interact with different mercury species from a variety of sources to
determine mercury deposition (e.g., (Bullock and Brehme, 2002; Cohen et al., 2004). Modeling
the atmospheric fate and transport of mercury is outside of the scope of this project, although
outputs of several atmospheric models will be used in combination with available empirical data
to estimate atmospheric deposition of mercury to different water bodies under different
regulatory scenarios.
3.2.3 Mercury Processes in Soils
A portion of the mercury deposited in terrestrial systems is re-emitted to the atmosphere.
On soil surfaces, sunlight may reduce deposited Hg(II) to Hg°, which may then evade back to the
atmosphere (Carpi and Lindberg, 1997; Frescholtz and Gustin, 2004; Scholtz et al., 2003).
Significant amounts of mercury can be co-deposited to soil surfaces in throughfall and litterfall
of forested ecosystems (St. Louis et al., 2001), and exchange of gaseous Hg° by vegetation has
been observed (e.g., (Gustin et al., 2004).
Hg(II) has a strong affinity for organic compounds such that inorganic Hg in soils and
wetlands is predominantly bound to dissolved organic matter (Mierle and Ingram, 1991). MeHg
likewise forms stable complexes with solid and dissolved organic matter (Hintelmann and
Evans, 1997). These complexes can dominate MeHg speciation under aerobic conditions
(Karlsson and Skyllberg, 2003). Truly dissolved and dissolved organic carbon (DOC)-
complexed Hg(II) and MeHg are transported by percolation to shallow groundwater, and by
runoff to adjacent surface waters (Ravichandran, 2004). Sorbed Hg(II) and MeHg are
transported by erosion fluxes to depositional areas on the watershed and to adjacent surface
waters (e.g., (Hurley et al., 1998).
Concentrations of MeHg in soils are generally very low. In contrast, wetlands are areas
of enhanced MeHg production and account for a significant fraction of the external MeHg inputs
to surface waters that have watersheds with a large portion of wetland coverage (e.g., (St. Louis
et al., 2001). Accordingly, there is a positive relationship between MeHg yield and percent
wetland coverage (Hurley et al., 1995). Hydrology exerts an important control on the magnitude
and flux of MeHg in wetland ecosystems (Branfireun and Roulet, 2002), as well as the transport
of inorganic mercury deposited in a given watershed to surface waters (Babiarz et al., 2001).
It should also be noted that there are exceptions to the predominace of wetlands as an
external source of MeHg to surface waters. For example, preliminary simulations with a multi-
3-11
-------
cell version of EPRI's Dynamic Mercury Cycling Model (D-MCM®) for Lake Superior
generated a model result where direct atmospheric deposition of methylmercury was predicted to
be the largest single source of methylmercury to the waterbody (Harris et al., 2002). This was
not because methylmercury deposition rates were unusually high. It was instead because in-situ
production and terrestrial loading of MeHg were predicted to be low.
3.2.4 Mercury Processes in Water
In a water body, deposited Hg(II) is reduced to Hg° by ultraviolet and visible
wavelengths of sunlight as well as microbially mediated reduction pathways (Amyot et al., 2000;
Mason et al., 1995). In turn, Hg° is oxidized back to Hg(II), driven by sunlight as well as by
"dark" chemical or biochemical processes (Lalonde et al., 2001; Zhang and Lindberg, 2001).
Driven by wind and water currents, dissolved Hg° in the water column is volatilized, which can
be a significant removal mechanism for mercury in surface waters and a net source of mercury to
the atmosphere (Siciliano et al., 2002).
In the water column and sediments, Hg(II) partitions strongly to silts and biotic solids,
sorbs weakly to sands, and complexes strongly with dissolved and particulate organic material.
The abundance of various inorganic ligands (e.g., OH", Cl", S2", DOC) in freshwater and saltwater
ecosystems plays an important role in both oxidation and reduction of inorganic mercury as well
as its bioavailability to methylating microbes. For example, reduction of Hg(II) is hypothesized
to be a function of the predominance of Hg(OH)2, which is inversely correlated with pH (Mason
et al., 1995). Reduction of Hg(II) to Hg° and subsequent volatilization from the water column is
important because it effectively reduces the pool of inorganic mercury that could potentially
undergo conversion to MeHg.
Hg(II) and MeHg sorbed to solids settle out of the water column and accumulate on the
surface of the benthic sediment layer. Surficial sediments interact with the water column via
resuspension and bioturbation. The burial of sediments below the surficial zone can be a
significant removal mechanism for contaminants in surface sediments (e.g., (Gobas et al., 1998;
Gobas et al., 1995). The depth of the active sediment layer is a highly sensitive parameter for
predicting the temporal response of different ecosystems to changes in mercury loading in
environmental fate models. This is because the reservoir of Hg(II) potentially available for
conversion to MeHg in the sediments is a function of the depth and volume of the active
sediment layer. The.compartment conducive for methylation is similarly affected (Harris and
Hutchison, 2003; Sunderland et al., 2004). Physical characteristics of different ecosystem types
affect estuarine mixing and sediment resuspension, which also affect the production of MeHg in
the water and sediments (Rolfhus et al., 2003; Sunderland et al., 2004; Tseng et al., 2001).
3.2.5 Bioavailability of Inorganic Mercury to Methylating Microbes
The amount of bioavailable MeHg in water and sediments of aquatic systems is a
function of the relative rates of mercury methylation and demethylation. In the water, MeHg is
degraded by two microbial processes and sunlight (Barkay et al., 2003; Sellers et al., 1996).
Recent research has shown that demethylating Hg-resistant bacteria may adapt to systems that
are highly contaminated with total mercury, helping to explain the paradox of low MeHg and
fish Hg levels in these systems (Schaefer et al., 2004).
3-12
-------
Mass balances for a variety of lakes and coastal ecosystems show that in situ production
of MeHg is often one of the main sources of MeHg in the water and sediments (Benoit et al.,
1998; Bigham and Vandal, 1994; Gbundgo-Tugbawa and Driscoll, 1998; Gilmour et al., 1998;
Mason et al., 1999). Sulfate reducing bacteria (SRB) are thought to be the principle agents
responsible for the majority of MeHg production in aquatic systems (Beyers et al., 1999;
Compeau and Bartha, 1987; Gilmour and Henry, 1991). SRB thrive in the redoxocline, where
the maximum gradient between oxic and anoxic conditions exists (Hintelmann et al., 2000).
Thus, in addition to the presence of bioavailable Hg(II), MeHg production and accumulation in
aquatic systems is a function of the geochemical parameters that enhance or inhibit the activity
of methylating microbes, especially sulfur concentrations, redox potential (Eh) and the
composition and availability of organic carbon.
A number of factors affect the bioavailabilty of Hg(II). A strong inverse relationship
between complexation of Hg(II) by sulfides and MeHg production has been demonstrated in a
number of studies (Benoit et al., 1999; Benoit et al., 1999; Craig and Bartlett, 1978; Craig and
Moreton, 1986). Passive diffusion of dissolved, neutral inorganic mercury species is
hypothesized as one of the main modes of entry across the cell membranes of methylating
microbes (Benoit et al., 1999; Benoit et al., 2003; Benoit et al., 1999). Thus, the formation of
neutral, dissolved mercury species such as HgCl2, Hg(OH)2, HgClOH, and HgS°(aq.), which
depend on the availability of constituent ligands in the surface and interstitial waters, may
strongly influence the availability of inorganic mercury to SRB, although our understanding of
the forms of mercury that are bioavailable to methylating microbes is currently incomplete
(Benoit et al., 2001; Benoit et al., 1999; King et al., 2001). The availability of the pool of
inorganic mercury in the water and sediments for methylation is also dominated by the binding
of Hg(II) with dissolved organic matter (DOM) complexes (Ravichandran, 2004).
Changes in the bioavailability of inorganic mercury and the activity of methylating
microbes as a function of sulfur, carbon and ecosystem specific characteristics mean that
ecosystem changes and anthropogenic "stresses" that do not result in a direct increase in mercury
loading to the ecosystem but alter the rate of MeHg formation may also affect mercury levels in
organisms (e.g., (Grieb et al., 1990). Because mercury concentrations in fish can increase even
when there has been no change in the total amount of mercury deposited in the ecosystem,
environmental changes such as eutrophication, which may alter microbial activity and the
chemical dynamics of mercury within an ecosystem, must be considered together with emission
control strategies to effectively manage mercury accumulation in the food web.
Recent research indicates that the bioavailability or reactivity of newly deposited Hg(II)
may be greater than older "legacy" mercury in the system (Hintelmann et al., 2002). These
results suggest that lakes receiving the bulk of their mercury directly from deposition to the lake
surface (e.g., some seepage lakes) would see fish mercury concentrations respond more rapidly
to changes in atmospheric deposition than lakes receiving most of their mercury from watershed
runoff. The implications of these data are also that systems with a greater surface area to
watershed area ratio that receive most of their inputs directly from the atmosphere (e.g., seepage
lakes) may respond more rapidly to changes in emissions and deposition of mercury than those
receiving significant inputs of mercury from the catchment area.
3-13
-------
3.2,6 Mercury Accumulation In the Food Web
Dissolved Hg(II) and MeHg accumulate in aquatic vegetation, phytoplankton, and
benthic invertebrates. Unlike Hg(II), MeHg biomagnifies though each successive trophic level in
both benthic and pelagic food chains such that mercury in predatory, freshwater fish is found
almost exclusively as MeHg (Bloom, 1992; Watras et al., 1998). Thus, trophic position and
food-chain complexity plays an important role in MeHg bioaccumulation (Kidd et al., 1995).
The chemical and physical characteristics of different ecosystems affect MeHg uptake at
the base of the food chain, driving bioaccumulation at higher trophic levels. At the base of
pelagic freshwater food-webs, MeHg uptake by plankton is thought to be a combination of
passive diffusion and facilitated transport (Laporte et al., 2002; Watras et al., 1998). Uptake of
MeHg by plankton can be enhanced or inhibited by the presence of different ligands bound to
MeHg (Lawson and Mason, 1998). Similarly, the assimilation efficiency of MeHg at the base of
the food chain is also affected by the type of dissolved MeHg-complexes in the water and
sediments. This may be a function of differences in the ability of organisms to solubilize MeHg
through digestive processes with different MeHg complexes (Lawrence and Mason, 2001;
Leaner and Mason, 2002). The presence of organic ligands and high concentrations of DOC in
aquatic ecosystems are generally thought to limit MeHg uptake by biota (Driscoll et al., 1995;
Sunda and Huntsman, 1998; Watras et al., 1998).
In fish, MeHg bioaccumulation is a function of several uptake (diet, gills) and
elimination pathways (excretion, growth dilution) (Gilmour et al., 1998; Greenfield et al., 2001).
As a result, the highest mercury concentrations for a given fish species correspond to smaller,
long-lived fish that accumulate MeHg over their life span with minimal growth dilution (e.g.,
(Doyon et al., 1998). In general, higher mercury concentrations are expected in top predators,
which are often large fish relative to other species in a waterbody.
3.2.7 Summary of Findings in the METAALICUS Study
METAALICUS is a whole-ecosystem experiment examining the relationship between
atmospheric mercury deposition and fish mercury concentrations (Harris et al., 2004). Stable,
non-radioactive isotopes of inorganic Hg(II) are being added to an 8.3-ha lake and its 44-ha
watershed in the Experimental Lakes Area (ELA), Ontario, Canada. Using isotopes provides the
ability to follow newly deposited mercury separately from background mercury. Different
Hg(II) isotopes are being applied to uplands, wetlands and the lake surface to distinguish the
contributions of each of these sources to fish mercury levels. Beginning in 2001, and continuing
each year since (3 years to-date), annual wet deposition of atmospheric Hg(II) has been
increased experimentally 3-4 fold relative to long term average wet deposition rates to the area.
Annual mercury additions to the lake surface were 22 \ig m"2 each year. Upland and wetland
areas received isotopes at average annual application rates of 21-25 and 25-28 ng m"2 yr"1
respectively for the 2001-2003 period.
During the first season of additions (2001), concentrations of inorganic mercury in the
surface waters of Lake 658 nearly doubled as a result of the mercury isotope (202Hg) added
directly to the lake surface. This represented a nearly proportional response for inorganic
mercury, relative to the increase in mercury loading to the lake as a result of the spikes to the
3-14
-------
lake surface. Inorganic 202Hg added to the lake surface was also detected as MeHg in the first
season in the water column, sediments and biota, including fish. Concentrations of 202Hg-
labelled MeHg in water, sediments and biota continued to increase in 2002 and 2003. Different
response dynamics were observed for the buildup of added mercury as MeHg in different
compartments. Results to-date suggest that the system has not yet stabilized in response to the
annual isotope additions directly to the lake surface. Inorganic mercury isotopes added to the
terrestrial system were measured in the uplands and wetlands and observed at near-detection
levels in lake waters by late 2002 (upland isotope only), but were not yet detectable in fish as of
2003. Initial efforts to simulate the L658 experiment with a mass balance model of aquatic Hg
cycling were unable to match the rate at which the isotope applied to the lake as inorganic
mercury was observed as MeHg in the system. The apparent higher bioavailability for
methylation of newly added mercury compared to "older" mercury may be a factor.
3.2.8 Summary of Florida Everglades Study
The Florida Everglades TMDL Pilot Study is one of the best-known investigations of the
temporal response of an aquatic system to reduced mercury loading. In the Everglades, elevated
mercury concentrations in fish are caused by a combination of atmospheric loading, net
methylation in water column periphyton, and food web dynamics(Atkeson, 2003). Periphyton
and macrophytes influence fish levels through their control of available divalent and methyl
mercury in the water column.
Incinerator mercury emissions in southern Florida have declined approximately 99%
since the mid-1980's as a result of pollution prevention and control policies. In general accord,
mercury in fish and wildlife of the Everglades has declined by approximately 60% since the
mercury peaked in biota in the mid-1990's.
In 1999 Florida DEP and USEPA began a modeling analysis of the environmental cycle
of mercury to explore the tools and data needed to perform a Total Maximum Daily Load
analysis (TMDL) for an atmospherically derived pollutant. Extensive Everglades specific data
are available to support a linked, multi-media modeling analysis through the auspices of the
South Florida Mercury Science Program, a 10-year multi-agency program of research, modeling
and monitoring studies.
The dynamic mercury cycling model (E-MCM) was applied to investigate changes in fish
tissue Hg (at site at WCA 3 A-15) with declines in atmospheric mercury deposition as part of the
Pilot TMDL study for that site (Tetra Tech Inc., 2000). Model simulations showed that
regardless of the magnitude of the load reduction, fish mercury concentrations were predicted to
change by 50% of the ultimate response within 8-9 years. Within 25-30 years, 90% of the
ultimate predicted response has occurred. In all cases, the actual magnitude of the change in fish
Hg was dependent on the magnitude of the load reduction. In the above simulations, a 3 cm
thick active sediment layer was assumed and the model did not distinguish between new and old
or "legacy" Hg.
At steady state, E-MCM forecasts a linear relationship between atmospheric mercury
deposition and mercury concentrations in largemouth bass, with a small residual mercury
concentration in fish at zero atmospheric mercury deposition: for any reduction in mercury
3-15
-------
inputs to the Everglades a slightly lesser reduction in fish mercury concentrations may be
anticipated. Furthermore, the E-MCM predicts near equivalence between the change in
atmospheric mercury deposition rate and the change in largemouth bass mercury concentration
over the likely range for current estimates of atmospheric deposition of mercury. The slight
offset from a 1:1 relationship results from slow mobilization of historically deposited mercury
from deeper sediment layers to the water column. Until buried below the active zone, this
mercury can continue to cycle through the system. In addition, because mercury is a naturally
occurring element, fish tissue mercury concentrations can never be reduced to zero. Further, the
model showed that absent changes to the system other than mercury loading (e.g. sulfur or
nutrient cycling, or hydrology), an -80% reduction from the ca. 1996 peak total annual mercury
atmospheric deposition would be needed for mercury concentrations in a 3-year old largemouth
bass in the central Everglades to be reduced to less than Florida's present fish consumption
advisory action level of 0.5 mg/kg.
Despite the quality of the modeling done as part of the Everglades study, this analysis has
limited applicability to other aquatic systems because of unusual attributes of the Everglades.
These attributes are listed below.
• Physiography of the waterbody — As a flat, shallow, vegetated marshland the Everglades
is atypically vulnerable to atmospheric deposition because of its great surface-to-volume
ratio.
Climate — Year-round high temperature and insolation stimulates chemical and physical
processes, promoting rapid aquatic cycling and unusually high production of MeHg.
• Meteorology — Easterly trade winds typify the synoptic transport regime during the
summer when ~ 85% of rainfall and ~ 90% of mercury deposition occurs. This pattern
efficiently brings emissions from the southeast coastal counties of Florida out over the
Everglades where frequent thunderstorms focus deposition there.
• Sources — Incineration was the largest emissions category in south Florida through the
mid-1990's. A predominance of emissions was 'reactive gas-phase mercury' (ROM or
Hg(II)) which tends to deposit on a local scale.
Synergy — Coupled with meteorology described above, the dominance of emissions as
RGM has resulted in an unusually tight local-scale coupling between emissions in
southern Florida and local-scale deposition.
3.3 Overview of Models Used in This Study
The general approach taken in this project was to couple outputs from atmospheric fate
and transport models with a set of watershed and water body models that are parameterized
using empirical data from well-characterized ecosystems.
3.3.1 Atmospheric Models
Over the past decade, EPA has used a variety of analytical and numerical simulation
tools to project the atmospheric transport, chemistry, and deposition of both criteria (e.g., ozone,
fine particles, etc.) and toxic (e.g., Hg) air pollutants. These models range in complexity from
simple, one-layer Gaussian dispersion models (e.g., Industrial Source Complex (ISC3) model) to
more complex, multi-layer Lagrangian puff-type trajectory models (e.g., Hybrid Single Particle
3-16
-------
Lagrangian Integrated Trajectory (HYSPLIT) model), and finally to complex three dimensional
(3-D) Eulerian grid models (e.g., Community Multiscale Air Quality (CMAQ) model). EPA
and others have been using a suite of complex numerical models to assess the transport and fate
of Hg emissions in the local, regional, and global atmosphere. In the Utility Report to Congress,
EPA relied heavily on the ISC3 dispersion model to assess nearfleld Hg deposition effects. The
HYSPLIT model has also been used extensively in the Great Lakes and Chesapeake Bay
watersheds to analyze source-receptor relationships for Hg deposition in these areas (Cohen et
al., 2004).
A review of the strengths and weaknesses of atmospheric mercury fate and transport
modeling is beyond the scope of this project. However, projected deposition scenarios for
different water bodies obtained from the models described above serve as inputs to the aquatic
fate and transport modeling described in this chapter. Modeled deposition rates (2001) using the
CMAQ and REMSAD models for each of the five ecosystems were compared to site-specific
data. Local monitoring data and sedimentary records of total mercury deposition obtained from
dated sediment cores (where available) were used obtain the best possible estimates of overall
loading to individual ecosystems. Outputs from atmospheric mercury cycling models
(REMSAD/CMAQ) were then used to forecast the relative change in mercury inputs to each
ecosystem modeled when contributions from coal-fired utilities were removed (e.g., the percent
difference in atmospheric deposition for each ecosystem between 2001 and scenario removing
coal-fired utilities as a source) to isolate their contribution the fraction of mercury accumulation
in fish. Details of the atmospheric deposition scenarios are described below in Section 3.4.
3.3.2 Ecosystem Models
EPA has developed a set of watershed, water body, and food web models that describe
the speciation, transport, and bioaccumulation of mercury as a function of the physical and
chemical properties of a specific ecosystem. The selected watershed and water body models
have been recently applied to various case studies. In this project, they were refined for
consistency and used to construct ecosystem specific mass balances for the three principle
mercury components - inorganic divalent mercury, Hg(II), elemental mercury, Hg°, and
monomethyl mercury, CH3Hg(II) (or MeHg). We compare the results from the SERAFM and
WASP waterbody fate and transport models (Section 3.4) as an internal check on consistency of
the modeling results. For more information on the specific models described below, please see
http://www.epa.gov/athens and www.epa.gov/crem.
3.3.2.7 Overview of the SERAFM Model
The SERAFM model incorporates more recent advances in scientific understanding
described above and implements an updated set of the IEM-2M solids and mercury fate
algorithms described in detail in the Mercury Study Report to Congress (USEPA, 1997). These
updates provide more realistic representations of the processes governing mercury fate and
transport in aquatic systems. Major differences between the SERAFM model and the IEM-2M
model are as follows:
3-17
-------
• Dynamic calculations: SERAFM can describe the temporal response of fish mercury
concentrations to changes in mercury loading, while the IEM-2M model calculated
expected fish tissue mercury concentrations at steady state.
Watershed Loading: Both IEM-2M and SERAFM model soil erosion into the water body
using the Revised Universal Soil Loss Equation (RUSLE). However, in SERAFM
mercury loading from the watershed to the water body is modeled using run-off
coefficients. SERAFM defines four land-use types: impervious, upland, riparian, and
wetland/forest. The user defines the percentage of each type in the watershed. The
model uses run-off coefficients to describe mercury from atmospheric deposition to each
land type as loadings to the water body. IEM-2M calculates mercury concentrations in
soils, and calculates erosion and transport to the water body.
• Two-Layer: SERAFM has the capability to model a layered lake system with an
epilimnion and hypolimnion, while IEM-2M used a single, well mixed layer to represent
the water column.
• Photo-reactions: Recent research has demonstrated the photo-reactions of mercury.
These have been incorporated into SERAFM but were not part of the original IEM-2M
model. The oxidation and reduction of mercury as functions of visible and UV-B light
are included.
• Speciation: Speciation of mercury with hydroxides, chlorides, and sulfides has been
included in the SERAFM model but was not incorporated in the IEM-2M model.The
abiotic oxidation rate constant for Hgll is multiplied by the fraction of dissolved divalent
mercury and the fraction of Hgll present as Hg(OH)2.
Equilibrium Partitioning: SERAFM models equilibrium partitioning between multiple
compartments or phases: aqueous phase, abiotic particles (silts/fines), biotic particles
(phytoplankton, zooplankton, seston), and DOC-complexation. In SERAFM, the biotic
demethylation rate constant is multiplied by the sum of the fraction dissolved and the
fraction DOC-complexed, as suggested by previous research (Matilainen and Verta,
1995).
• Trophic status: Trophic status of the lake has been incorporated into the SERAFM
model and was not a component of the IEM-2M model. Trophic status is used to
calculate visible light attenuation in the lake, the turnover of biomass, and the
phytoplankton and zooplankton concentration in the SERAFM model framework.
• Suspended particle types in the water column: The SERAFM model accounts for both
zooplankton and phytoplankton as biotic materials in the system, while IEM-2M only
accounted for one biotic particle type.
• Reaction rates: The SERAFM model incorporates more recent reaction rate coefficients,
and the understanding of the variability of these rates with different conditions.
• Partition coefficients: The SERAFM model incorporates more recent values for mercury
partition coefficients for each mercury species. Future versions of the SERAFM model
will calculate site-specific partitioning as a function of sediment organic matter and the
organic carbon content of suspended materials.
State variables in both the IEM-2M and SERAFM models include three mercury species,
Hg°, Hg(II), and MeHg. As mentioned above, SERAFM includes four solids types (abiotic
solids, phytoplankton solids, zooplankton solids, and detrital solids) and dissolved organic
carbon, DOC. Both IEM-2M and SERAFM simulations are driven by external mercury loadings
delivered from the atmosphere, from watershed tributaries, and from point sources, or by internal
3-18
-------
loadings from contaminated sediments. SERAFM calculates the time-dependent solids and
mercury species concentrations in the water column and sediments of the specified water body
reach. Hg(II) and MeHg are partitioned to suspended and benthic solids and to dissolved DOC
with user-specified partition coefficients for each sorbent type.
In the SERAFM model, mercury species are subject to several transformation reactions,
including photo-oxidation and dark oxidation of Hg° in the water column, photo-reduction and
methylation of Hg(II) in the water column and sediment layers, and photo-degradation and
demethylation of MeHg in the water column and sediment layers. Water column oxidation,
reduction and demethylation reactions are driven by sunlight, and so their input rate constants
are attenuated through the water column using specified light extinction coefficients. Hg° is
subject to volatile exchange between the water column and the atmosphere governed by a
transfer rate calculated from velocity and depth, and by its Henry's Law constant.
A preliminary comparison of the SERAFM model to the IEM-2M model using the
parameter values for the model ecosystem described in the RtC suggests that updates to the IEM-
2M model incorporated into the SERAFM model result in lower values for fish mercury
concentrations (Table 3-1). However, the model ecosystem described in the RtC uses a lower
dry deposition rate than estimated based on more recent understanding and assumes that there is
no watershed MeHg loading. When these parameters are updated to reflect current knowledge,
forecasted fish mercury concentrations are higher than the original IEM-2M results (see
Table 3-1).
Table 3-1. Comparison of SERAFM and IEM-2M Forecasted Mercury Concentrations
Using Parameter Values for Model Ecosystem Described in the Mercury Study Report to
Congress (RtC) and a 50% Reduction in Atmospheric Deposition
Parameters
Model
Water Column
MeHg Unfiltered
Water Column HgT
Unfiltered
Trophic Level 4
Fish
Trophic Level 4
Fish BAF:
FishHg/MeHg
RtC Model
Ecosystem
IEM-2M
0.08
1.16ngL-'
0.44 ug g'1
RtC Model
Ecosystem
SERAFM
0.03 IngL'1
2.50 ng L'1
0.21ugg-'
6.8xl06
Updated
Parameters
SERAFM
0.12ngL-'
1.17ngL-'
0.80 ug g'1
The "Updated Parameters" column refers to modification of the original model
ecosystem described in the RtC to incorporate more recent knowledge on the
magnitude of dry deposition and inputs of MeHg from the catchment.
3.3.2.2 Overview of the WASP Model
WASP (Water Quality Analysis Simulation Program) is a dynamic, mass balance
framework for modeling contaminant fate and transport in surface water systems. This model
helps users interpret and predict water quality responses to natural phenomena and man-made
3-19
-------
pollution for various pollution management decisions. WASP is an enhancement of the original
WASP (Ambrose, 1987; Ambrose, 1988; Connolly and Thomann, 1985; Di Toro et al., 1983)
and allows the user to investigate 1, 2, and 3 dimensional systems, and a variety of pollutant
types. The time-varying processes of advection, dispersion, point and diffuse mass loading and
boundary exchange are represented in the model. WASP also can be linked with hydrodynamic
and sediment transport models that can provide flows, depths, velocities, and temperature,
salinity and sediment fluxes.
The WASP7 mercury module simulates three mercury species, Hg°, Hg(II), and MeHg,
as well as three solids types (silt, sand, and biotic solids) (e.g., (Ambrose and Wool, 2001).
Simulations are driven by the speciated mercury loadings delivered from the atmosphere, from
watershed tributaries, and from point sources. Throughout the simulation period, WASP
calculates solids and mercury species concentrations in the water column and sediments of each
reach. Transport processes simulated include advection, dispersion, and sediment-water column
exchange. Hg(II) and MeHg are partitioned to silt, sand, and biotic solids, and to dissolved
organic carbon (DOC).
3.3.2.3 Overview of the WCS Model
Although significant progress has been made in recent years on estimations of mercury
transport fluxes in watershed areas within a region in which atmospheric deposition is
presumably constant (Balogh et al., 1998; Hurley et al., 1995; Lawson et al., 2001; Lee et al.,
1995; Tsiros, 1999), only a few watershed studies have focused on the importance of indirect
anthropogenic sources of Hg such as terrestrial runoff, compared to direct atmospheric
deposition. The EPA Region 4 Watershed Characterization System (WCS) is a GIS-based
modeling system for calculating soil particle transport and pollutant fate in watersheds
(Greenfield et al., 2002). Its mercury transport module was developed from the IEM-2M model,
which calculated mercury species concentrations in an idealized watershed and water body based
on steady atmospheric mercury deposition and long-term average hydrology. Similarly, the WCS
calculates long-term average hydrology and sediment yield, but simulates total Hg(II) in a more
realistic, distributed sub-watershed network. A second-generation WCS that operates on a finer
computational grid is under development. Initial background soil mercury concentrations along
with wet and dry atmospheric mercury deposition fluxes are input to the model. For pervious
subwatershed grid elements, WCS calculates surficial soil mercury concentrations over time
using a mass balance. Calculated total mercury in the surficial soil layers is partitioned between
the dissolved and particulate phases (in the soil water and on the soil solids) assuming local
equilibrium, governed by a partition coefficient. Dissolved mercury is lost from the surficial soil
layers through percolation and runoff. Particulate mercury is lost through water runoff erosion.
No wind resuspension is included in the model calculations. A fraction of the soil mercury is
reduced and volatilized back to the atmosphere. Subwatershed mercury loadings in runoff water
and runoff erosion particles are delivered to the watershed tributary system. For impervious
areas and water surface areas within the subwatersheds, atmospheric mercury deposition is
delivered to the tributary system without loss.
Surficial soil concentrations for each of the three mercury components are calculated in
WCS on a daily basis using a mass balance equation driven by an input term for atmospheric
deposition. Mercury output terms include gaseous flux from soil to air by volatilization, vertical
3-20
-------
hydrologic transport through soil by percolation, and horizontal hydrologic transport from
surficial soil by runoff water and runoff erosion particles. A source/sink term for mercury
cycling with first-order kinetics is used to represent oxidation of Hg°, methylation of Hg(II),
demethylation of MeHg, and abiotic reduction of Hg(II) to Hg°, with a functional dependence of
the reduction rate on soil moisture and vegetation cover shading (Tsiros, 2002).
3.3.2.4 Overview of the BASS Model
BASS (Bioaccumulation and Aquatic System Simulator) describes the dynamics of
mercury bioaccumulation in the food chain using algorithms that account for mercury
accumulation among different species and different age classes using species-specific uptake and
elimination terms such as diet composition and growth dilution (Barber, 2001; Barber et al.,
1987; Barber et al., 1988; Barber et al., 1991).
BASS simulates the population and bioaccumulation dynamics of age-structured fish
communities. Although BASS was specifically developed to investigate the bioaccumulation of
chemical pollutants within a community or ecosystem context, it can also be used to explore
population and community dynamics offish assemblages that are exposed to a variety of non-
chemical stressors such as altered thermal regimes associated with hydrological alterations or
industrial activities, commercial or sports fisheries, and introductions of non native or exotic fish
species. Contaminants entering each fish through gill exchange and ingestion are partitioned
internally to water, lipid, and non-lipid organic material. Internal equilibrium among these
phases is assumed to be rapid in comparison with external exchanges. Stability coefficients are
specified for the binding of MeHg to available sulfhydryl groups in the fish's non-lipid organic
material.
BASS's model structure is very generalized and flexible. Users can simulate both small,
short-lived species (e.g., daces, minnows, etc.) and large, long-lived species (e.g., bass, perch,
sunfishes, trout, etc.) by specifying either monthly or yearly age classes for any given species.
The community's food web is defined by identifying one or more foraging classes for each fish
species based on body weight, body length, or age. The dietary composition of each of these
foraging classes is then specified as a combination of benthos, incidental terrestrial insects,
periphyton/attached algae, phytoplankton, zooplankton, and/or other fish species, including its
own. One of the strengths of the BASS model relative to other bioaccumulation model
frameworks is that there are no restrictions on the number of chemicals or the number offish
species that can be simulated, the number of cohorts/age classes that fish species may have, or
the number of foraging classes that fish species may have (Barber, 2003).
3.4 Overview of Case Studies
Case studies presented in this section are used to explore the range in temporal responses
of different ecosystems following reductions in atmospheric Hg emissions and some of the
sources of uncertainty around the proportional relationship used by the MMaps model. To do
this, we provide quantitative examples from five case studies that span a range of freshwater
ecosystem types across the Eastern and Midwestern United States.
3-21
-------
The five case studies selected for this project were constrained to reasonably well-studied
ecosystems where there were sufficient empirical data to parameterize time-dependent aquatic
cycling models. Our goal was to select well-characterized sites that would display a range in the
magnitude and timing of responses to changes in mercury loading because of differences in
ecosystem specific factors that affect the transport and transformation of mercury in different
water bodies. The sites selected (see Section 3.5) cover a range of ecosystem types, sizes and
latitudes.
3.4.1 Ecosystem Characteristics
A brief description of each site is given below and a summary of ecosystem
characteristics used to parameterize the models are detailed in Table 3-2.
• Eagle Butte, South Dakota: Livestock lakes and ponds on the Cheyenne River Sioux
Tribal Lands. The site modeled (Lee Dam), is a shallow, well-mixed system with a water
surface area of 0.2 km2 and a catchment to lake area ratio of 22.6. There are power plants
in the vicinity of this site, although the prevailing meteorology likely transports most
emissions east of the site. Currently, consumption advisories are in place on reservation
lands due to high levels of mercury in piscivorous fish. Mercury dynamics at this site are
currently being studied as part of an EPA Regional Office RARE Grant awarded in 2003
(http://www.epa.gov/osp/regions/RARE_Region8.pdf). Atmospheric deposition data,
total mercury and MeHg concentrations in sediments, water and biota have all been
collected as part of this study. To model this system, we used the empirical data
collected at this site to parameterize both the SERAFM and WASP models. The BASS
model was applied to investigate mercury residue attenuation in length classes of co-
dominant fishes.
• Pawtuckaway Lake, New Hampshire: Medium sized, seepage lake in Nottingham, New
Hamsphire. This lake has a water surface area of 3.6 km2 and a catchment to lake area
ratio of 13.7. Pawtuckaway Lake is characteristic of undisturbed lakes within the
Northeastern Highlands Ecoregion (Omernik, 1987; US EPA, 2000). This lake was part
of a recent study of mercury dynamics across a number of Vermont and New Hampshire
Lakes funded by EPA's Office of Research and Development under the Regional
Environmental Monitoring and Assessment Program (Kamman et al., 2004). Local
mercury sources in the region include a number of utility units (See Figure 3-8) and
several incinerators.
• Lake Waccamaw, North Carolina: Large, bay lake in southeastern North Carolina. Lake
Waccamaw has a water surface area of almost 35 km2 and a catchment to lake area ratio
of slightly more than six (Table 3-2). In 1992, a survey offish mercury concentrations
North Carolina's Department of Environment, Health and Natural Resources in this
region revealed that fish mercury concentrations exceeded 1 ppm in over 60% of the
samples. Waccamaw is a popular destination for recreational fishing and a fish
consumption advisory is currently in place. The area surrounding Lake Waccamaw is
typical of the region: flat terrain with ubiquitous wetlands and waterways. Very little
commercial or industrial activity takes place in the area immediately surrounding the
park, population density is relatively low and roadways are lightly traveled. The nearest
town is Whiteville, NC, located approximately 15 kilometers to the west-northwest of
Lake Waccamaw. A variety of mercury sources are located in this region including at
3-22
-------
least two coal-fired electric utility boilers, a large municipal waste incinerator, several
large coal or oil-fired industrial boilers, and a pulp and paper mill. By far the largest
historic source of mercury emissions was the HoltraChem mercury cell chlor-alkali
operation located in Riegelwood, NC, approximately 25 kilometers east-northeast of
Lake Waccamaw, which ceased operation in the last decade.
• Brier Creek, Georgia: Brier Creek is a coastal plain river, dominated by a watershed that
contains several different types of land uses and is modeled using 11 different sub-
watersheds located in central/eastern portion of Georgia. Unlike the lakes modeled in
this report, the watershed response drives mercury dynamics in Brier Creek. Brier Creek
is a popular destination for recreational fishing. Fish consumption advisories are in place
in this region due to high mercury levels.
• Lake Barco, Florida: Small, seepage lake in northeast Florida. Lake Barco is located
approximately 35 km east of Gainesville, Florida on the Ordway Preserve that is operated
by the University of Florida. The Ordway is protected from direct human impacts,
although some recreational fishing does take place. Lake Barco has a water surface area
of 0.12 km2 and a negligible catchment area. Hydrology and geochemistry of Lake
Barco have been well characterized by past studies (EPRI, 2003; Pollman et al., 1991)
and there are several nearby mercury sources. For example, Gainesville Regional
Utilities (GRU) operates a medium-size coal-fired power plant approximately 40 km
northwest of Lake Barco. Emissions of Hg from the Deerhaven Unit No. 2 facility
averaged approximately 30 kg yr"1 between 1998 and 2002 (range 13 to 47 kg yr"1).
Table 3- 2. Summary of Ecosystem Characteristics Used To Parameterize Mercury
Models
Parameter
Watershed Area (m2)
Percent Impervious
Percent Forest
Percent Riparian
Percent Upland
Lake Area (m2)
Catchment/Lake Ratio
Epilimnion Depth (m)
Hypolimnion
Depth (m)
Hypolimnion Anoxia
Hydraulic Residence Time
(days)
Inflow/Outflow
(mV)
Water pH
Epilimnion DOC
(mgL-1)
Hypolimnion DOC
(mgL-1)
Trophic Status
Lake
Pawtuckaway
5.00 xlO7
1%
88%
10%
1%
3.64xl07
13.7
2.0
3.0
Yes
165
4.05x10'
6.45
5.5
5.6
Dystrophic
Lake
Waccamaw
2.17x10"
1%
72%
0%
27%
3.47x10'
6.3
2.3
n/a
n/a
241
1.20xl08
4.3
25.9
n/a
Mesotrophic
Lake Barco
0
n/a
n/a
n/a
n/a
1.18xl05
0
3.7
n/a
n/a
n/a
0
4.5
0.8
n/a
Oligotrophic
Eagle Butte
4.21 xlO6
40%
0%
20%
40%
1.86 xlO5
22.6
2.0
n/a
n/a
n/a
0
9.0
27.0
n/a
Eutrophic
Brier Creek
2.19 xlO9
2%
47%
12%
39%
n/a
n/a
0.3-2.0
n/a
n/a
12
3.3-7.4
xlO8
n/a
5.0-8.0
n/a
n/a
3-23
-------
3.4.2 Baseline Atmospheric Deposition at Each Site
For each site, measured empirical data were used to characterize the current level of
atmospheric deposition (Table 3-3). Lake Pawtuckaway data were obtained from Kamman and
Engstrom (2002). We estimated total deposition at Lake Waccamaw and Lake Barco by
assuming dry deposition was approximately 50% of total deposition. Wet deposition at Lake
Waccamaw represented averaged cumulative wet deposition at MDN site NC-O8 between 1998
and 2000. Wet deposition of mercury at Lake Barco deposition were from a technical report
(EPRI, 2003). Atmospheric deposition at Eagle Butte was calculated as from ongoing empirical
measurements and data for Brier Creek were from a number of MDN sampling sites. In
Pawtuckaway Lake, Lake Waccamaw, Lake Barco and Eagle Butte, MeHg was assumed to
represent approximately 3% of total deposition (Fitzgerald et al., 1994; Iverfeldt, 1991), which
falls within the range of measured values.
Table 3- 3. Baseline Atmospheric Deposition For Each Model Ecosystem
Lake Lake Lake Barco Eagle Butte Brier Creek
Pawtuckaway Waccamaw
Annual Precipitation
(cm yr-1)
HgT Precipitation (ng I/1)
Wet Deposition
(ngL-1)
Dry Deposition
(Hg m2 yr'1)
Wet Deposition (MeHg)
ng m2 yr"1
Dry Deposition (MeHg)
|igm2yr''
Total Deposition (HgT)
Hgm2yr''
102.0
10.0
10.2
10.2
0.15
0.15
20.7
120.4
12.0
14.4
14.4
0.22
0.22
29.2
134.8
11.5
15.5
15.5
0.23
0.23
31.5
43.0
21.9
9.4
10
0.20
0.15
19.8
120.0
12.2
14.7
12.1
n/a
n/a
26.8
3.4.3 Atmospheric Loading Scenarios Investigated
For each site, we compared deposition rates in the 36x36 km grid cell corresponding to
the ecosystem locations forecasted using the CMAQ and REMSAD atmospheric fate and
transport models (Table 3-4) to the empirically derived loading rates presented in Table 3-4.
Overall, the site specific data suggest somewhat higher deposition rates than the CMAQ and
REMSAD models at the locations chosen for these case studies. This reinforces the need for
additional data sets that can be used to test model-forecasted atmospheric mercury deposition
rates. In Table 3-4, 2001 base case deposition rates are compared to two atmospheric deposition
scenarios. The 2020 projection describes anticipated loading rates at the end of the proposed
rule and the zero out scenario was calculated by removing coal fired utilities from the
atmospheric models as sources of mercury emissions.
Projected differences in atmospheric deposition (Table 3-4) with the zero-out scenario
are similar to 2020 projected deposition rate under the Clean Air Interstate Rule (CAIR)
presumably because the 2020 scenario includes reductions in emissions from mercury sources
other than coal-fired utilities. We therefore modeled only the change in fish tissue
3-24
-------
concentrations associated with the zero out deposition scenarios. To do this, we calculated a
maximum percent difference between 2001 base case deposition the zero out scenarios for the
REMSAD and CMAQ numbers, and investigated the anticipated response of fish mercury
concentrations to these changes. Changes in atmospheric loading under these scenarios across
all ecosystems range from four to fifteen percent. These values are relatively small in magnitude
compared to maximum values observed across the country.
Accordingly, we also considered the range in deposition changes projected to occur
across the country. The highest forecasted relative change in deposition across all the grid cells
covering the United States in the CMAQ and REMSAD atmospheric models after removal of
power plants as a source of mercury was approximately 70% (50 p.g m'2 yr"1 to 15 |j.g m"2 yr"1).
In addition to the zero-out scenarios run using the SERAFM model, we modeled a standard 50%
decline in atmospheric loading for each case study to compare the expected response across
sites.
Table 3- 4. Forecasted Atmospheric Deposition Rates in Case Study Areas Using the
CMAQ and REMSAD Models
Ecosystem
Brier Creek, GA
Eagle Butte, SD
Lake Barco, FL
Pawtuckaway Lake,
NH
Lake Waccamaw, NC
CMAQ Deposition Projections (u,g m"2
yr-1)
Baseline
(2001)
14.2
8.6
15.6
16.1
16.3
2020
Projection
12.6
8.2
14.3
15.0
14.2
Zero Out
Utilities
12.7
8.4
14.8
15.2
14.1
REMSAD Deposition Projections (jig
m1 yr'1)
Baseline
(2001)
13.2
6.7
16.7
10.3
19.0
2020
Projection
11.8
6.5
14.9
8.9
16.2
Zero Out
Utilities
12.0
6.5
15.5
9.4
16.2
Note: The gradients of mercury deposition around model ecosystem watersheds are fairly low, even with a known
EGU source nearby with the exception of Brier Creek. Two CMAQ grid cells that overlap the Brier Creek
•watershed are twice the magnitude of the average grid value of 16 ng m'2yr'.
3.4.4 Summary of Model Evaluation
Models applied to each case study system are listed in Table 3-5. Irrespective of the
quality of their process algorithms, none of the models can be considered a priori predictive
tools. It is understood that all environmental models are under-determined, so that different
combinations of model parameters can cause simulated concentrations to match the limited set of
observations. It is also understood that observed datasets are always incomplete and uncertain
and represent only a snapshot of the real system. Thus, model calibration involves professional
judgment.
3-25
-------
Table 3- 5. List of Model Frameworks Applied to Ecosystems
Ecosystem Model Frameworks Applied
Brier Creek, GA WASP, WCS
Eagle Butte, SD SERAFM, WASP, BASS
Lake Barco, FL SERAFM
Pawtuckaway Lake, NH SERAFM, WASP
Lake Waccamaw, NC SERAFM, WASP
To address model identification uncertainty, we conducted multiple calibrations of
different model frameworks. Results of these model calibrations are described below. We
applied the SERAFM model to all systems but Brier Creek. Because mercury dynamics in Brier
Creek are dominated by a watershed that consists of 11 tributaries, a surface water body model
like SERAFM is not suitable for application. At three sites (Eagle Butte, Lake Waccamaw and
Pawtuckaway Lake) both the WASP and SERAFM models were applied as an internal quality
assurance check.
All models were calibrated using the available empirical data and run to steady state
before investigating any atmospheric loading scenarios. In some cases it was necessary to
supplement the observed site-specific data with parameter values and ranges from the general
scientific literature. The most reasonable set of parameters that best fit the observed data at each
site were used as the base case. Details of these calibrations are contained in the Attachments 1-
5 that describe each of the site-specific applications.
Although there was insufficient time to do a formal sensitivity analysis of these models,
we investigated the effects of parameter values (summarized in Table 3-6) on ecosystem
response times using several different model calibrations. For each ecosystem, we modeled the
fast, medium and slow response scenarios by varying the depth of the active sediment layer and
the macro-dispersion coefficient for the sediments (i.e., exchange rate between surface sediment
and water column. We chose the same upper sediment depth modeled in the U.S. EPA Mercury
Study Report to Congress (US EPA 1997), where a default value of 2 cm and a uniform
distribution of one to three centimeters were justified (see the technical discussion in Volume III,
Appendix B.2.27, p. B-50). The sediment-water dispersion coefficient is known to be greater
than molecular diffusion coefficient, with a commonly accepted value of around 10"6 tolO"5 cm2
sec"1 (Bowie et al., 1985). Dispersion coefficients for sediments subject to bioturbation
(disturbance by benthic organisms) are commonly accepted to fall between 10~5 to 10"4 cm2sec"'
(Schnoor, 1987). We used these alternate calibrations in the scenario projection phase of this
project to provide a semi-quantitative uncertainty envelope for the temporal responses of the
various ecosystems. In addition, we investigated the uncertainty in watershed loading response
times by conducting sensitivity analyses with the WCS model of the Brier Creek watershed.
Given more time, a formal error propagation analysis would be a valuable addition to this study.
3-26
-------
Table 3- 6. Summary of Mercury Parameters Used in the SERAFM Model
HgO
Hgll
MeHg
H
[atm-m3/mol|
T.lxlO'3
7.10x10-'°
4.7xlO'7
Kd^i. [L/kg]
0
250,000
100,000
Parameter
Active Sediment Layer Thickness
Macro-Dispersion in Sediments
Kd,bio [L/kg]
0
399,052
516,313
Kd)Doc [L/kg]
0
251,188.6
100,000
Range
1 -3 cm
SxlO^-lO^cmV
Runoff Coefficients: Ratio of Watershed Export to Deposition Loading
Impervious
Wetland/Forest
Riparian
Upland
HgO
1
0.2
0.2
0.2
Hgll
1
0
.2
0.2
0.2
MeHg
1
4.9
2.0
0.2
H=Henry's Law Constant (used to describe portioning of mercury between air and water), Kd=partition coefficient
between solids and water for abiotic solids (abio), biotic solids (bio) and dissolved organic carbon (DOC)
3.4.5 Baseline Fish Mercury Concentrations
To maintain consistency with the original IEM-2M model, the SERAFM model forecasts
fish mercury concentrations as a function of an empirically derived bioaccumulation factor
(BAF). This BAF was based on site specific fish mercury concentration data and MeHg
concentrations in the water column. Site specific fish mercury concentration data were also used
as the baseline value used in the MMaps model to forecast fish mercury levels with changing
atmospheric deposition scenarios that are described in the next section.
Empirically derived BAFs for these systems fall within the range of those measured at
other sites (Table 3-7). Bioaccumulation factors were calculated for all SERAFM applications
by normalizing residues to the length of the dominant species of Level 4 (top) piscivoire for
which age data were available (e.g., 2 yr old largemouth bass).
Table 3- 7. Empirically Derived BAFs for Each of the Ecosystem Case Studies
BAFs (L/kg)
MeHg (ng/L)
HgT (ng/L)
BAF-MeHg
BAF - HgT
Eagle Butte
Lake Barco
Lake Pawtuckaway
Lake Waccamaw
Hypothetical Ecosystem
(USEPA 1997)
0.82
0.018
0.19
0.48
~
10.2
1.03
2.26
4.79
--
8.90E+05
3.06E+07
1.11E+06
1.24E+06
6.80E+06
8.73E+04
5.34E+05
9.29E+04
3.86E+04
5.30E+05
We chose to base fish mercury responses in the SERAFM model on empirically derived
BAFs to be consistent with the original IBM model, which was the basis for the MMaps
derivation. However, there are some limitations to the BAF approach that deserve mention.
Because MeHg concentrations in the water column are highly variable, the derived BAF is
inherently underdetermined. One less variable approach might be to normalize between model
sites using the fraction of organic carbon in sediments as the fraction of MeHg available for
3-27
-------
uptake by organisms appears to be directly related to the organic carbon content of the sediments
(Lawrence and Mason, 2001; Lawrence et al., 1999; Mason and Lawrence, 1999). More
important though, response times (i.e., rates to and from equilibrium) in aquatic communities to
changes in loading rates are driven by food web dynamics and species metabolic rates
responding to changes in concentrations of MeHg in the water column and sediments.
To cross check empirically calibrated BAFs and to improve the estimated response times
of mercury residues in different species and age/size classes offish, the BASS bioenergetics
based trophic dynamics model was also applied to the Eagle Butte site (Figure 3-1).
1200 —
•g- 1000 -
|- 800 -
g> 600 -
-------
3.5
o
'o
* 2.5
o>
X
5
Pawtuckaway
Lake
2.5 3
Log Hg fish (ng/g) - Observed
3.5
Straight line represents 1:1 relationship between observed and modeled results.
Figure 3- 2. Observed vs. Predicted Fish Mercury Concentrations in Model Ecosystems at
Steady State with No Change in Atmospheric Loading
At steady state, SERAFM overpredicts fish mercury concentrations in Lake Barco and
underpredicts concentrations in Lake Waccamaw. Underprediction of observed fish mercury
levels in Lake Waccamaw is likely a function of the recent closure of the chlor-alkali facility in
proximity to the lake. When the SERAFM model is run to steady state under the current loading
scenario, expected fish mercury concentrations are somewhat lower than observed values. It is
therefore likely that fish mercury concentrations in Lake Waccamaw are not presently at steady
state and will continue to decline if the current level of atmospheric deposition remains constant.
We do not have enough information the recent history of Lake Barco to speculate as to why we
observe the difference between model-predicted and observed fish concentrations. Given
additional time, a more comprehensive sensitivity analysis of factors affecting these results
would be a useful starting point. On a purely speculative level, it is possible that fish mercury
concentrations in Lake Barco are also not at steady state with respect to current deposition
levels. If mercury loading to this ecosystem has increased recently but has not yet been reflected
in observed fish tissue residues, than fish Hg levels would be overpredicted by the SERAFM
model.
Analysis of the degree of corroboration between observed and model predicted fish
mercury concentrations is a useful starting point for discussion of the MMaps model. Like
SERAFM, the MMaps model uses a proportional relationship between declines in deposition
forecasted by atmospheric fate and transport models and concentrations in fish. Unlike
SERAFM, MMaps assumes the empirical data reflect steady state concentrations in the fish,
thereby over and underestimating mercury concentrations in fish in ecosystems that have
experienced recent changes in mercury inputs.
3-29
-------
Initial mercury concentrations in fish used in the MMaps model are therefore critical for
model outputs. As illustrated in Figure 3-2, the error bars around one species of piscivorous fish
from a single water body are large. National fish tissue data coverages for even a five year
period used by the MMaps approach are limited and span different trophic levels and ages of
species. In addition to uncertainty in atmospheric models used to forecast changes in mercury
deposition described above, uncertainty in MMaps forecasts as the result of variability in fish
mercury residues will be significant and is not likely to capture the true variability in mercury
concentrations among ecosystems in the United States.
3.4.6 Magnitude of Changes in Fish Tissue Residues
For all ecosystems but Brier Creek, we ran scenario projections for fish mercury
concentrations r*^^'" n ^AOX. »-*»yii»/*f^/%r'»^^»^»-i/^ ^^•^*-\n«*i*-*»^ ir-v«i*f*«*iAn i-\-f*cc7D ,Ap^^
forecasted atte ike Barco are
shown in Figui sediments
mainly to inve y deposition,
and the effects
0.4
0.35
0.3
BJ0.25
| 0.2
Jd
V)
"- 0.15
0.1
0.05
PawtuckawaylLake, NH
50% Loading Deduction
Fast: ! ;
1 cm sediment, D = JIO-4 cm2/s
^Slow; 3 cm sediment,
D = 5x-10-5cm2/s
Medium: i
21 cm sedimenj, D = 10-4 cm2ls
MMaps
' Proportional
Relationship
50 100
Time [yrs]
150
Figure 3-3. 1
Lake, NH to a
tvtuckaway
3-30
-------
Figure 3-4. T<
to a Decline in
1,0
1 R
i ^
1 1
H
0>
X 08
JC. U'°
VI
U_
0.6
OA
O*\
\ ijake Bared, FL ]
\ 50%iL6ading Reduction \
\ I I ' I
\ V £ast: i cm sediment, j
\ \ D= 10-4cm2/s |
\ ^~-^_ '• \
\v ;' Medium: 2 cm sediment,
\j^_b = 1 6-4 cm2/s !
^^^^ |
Slow: 3 cm sediment, :
D = 5x10-5 cm2/s ••
) 20 40 60
Time [yrs]
MMaps
Proportional
Relationship
ke Barco, FL
3.4.6.1 Zero-Oi
For Paw _ ve simulated
fish mercury concentrations using the forecasted decline in deposition in each ecosystem
associated with the removal of U.S. utilities as a component of atmospheric deposition
(Table 3-8).
Table 3- 8. MMaps and SERAFM Forecasted Fish Mercury Concentration at Steady State
after Removal of Coal Fired Utilities as a Component of Deposition Using the REMSAD
and CMAQ Models (Zero-out Scenario)
Ecosystem
% Difference in Atm. Hg
Deposition
Zero-Out Scenario
MMaps Forecasted Fish
Hg Zero-Out Scenario
(MB g'1)
SERAFM Forecasted
Fish Hg Zero-Out
Scenario (|j.g g'')
Eagle Butte
Lake Barco
Pawtuckaway Lake
Lake Waccamaw
4.3%
6.8%
8.5%
14.8%
0.85
0.51
0.19
0.51
0.86
1.12(0.65-1.58)
0.21 (0.19-0.24)
0.18
Fish mercury concentrations under the "zero-out" scenario for both MMaps and
SERAFM are shown in the columns following the percent differences in deposition at each site
calculated from the REMSAD and CMAQ models (Table 3-4). Although both models use the
proportional relationship described above, comparing the magnitude of forecasted fish mercury
levels illustrates some of the uncertainty in model-predicted values. Overall, it is clear that the
magnitude of the uncertainty in the atmospheric and ecosystem models at this time is much
greater than the signal derived from a change in loading following removal of the coal fired
utilities at these sites. This may not be the case for highly impacted systems described above,
3-31
-------
where atmospheric deposition declines forecasted by the REMSAD and CMAQ models exceed
70% at some sites. Attenuation curves for fish mercury using the 50% decline in loading
scenario may therefore better represent the response of mercury in fishes in similar types of
ecosystems that are highly affected by mercury deposition from coal fired utilities (Figures 3-3,
3-4).
The major differences between the SERAFM model and the MMaps model across all
sites are a function of the initial fish mercury concentrations at steady state that drive the
response in both systems. Although both models assume fish mercury concentrations are at
steady state with respect to baseline atmospheric deposition levels used to calculate the percent
change in deposition (e.g., the same percent change is used in both models but overall loading
rates and initital fish concentrations are different), only SERAFM calculates an expected
concentration of mercury in fish at steady state for each ecosystem. Relative confidence in the
two models is equivocal at this time, although the uncertainties in both approaches are
highlighted by the observed differences in results.
3.4.7 Summary of Observed Temporal Responses to Declines in Loading
Table 3-9 compares the 50% decline scenario projections using the SERAFM and WASP
models. The Brier Creek scenario incorporates response times in the watershed soils as well as
in the water body sediments. The four lake scenarios account for the dynamic internal sediment
response to instantaneous declines in total mercury loading. Accounting for watershed response
times would add to the overall lake response time estimates.
Table 3- 9. Sediment Response Times in Years to Reach 90% of Steady-state
Concentrations Following 50% Mercury Deposition Reductions
Site Fast Medium Slow Fast WASP Medium Slow
SERAFM SERAFM SERAFM WASP WASP
Brier Creek - Upstream
Brier Creek - Downstream
Eagle Butte
Lake Barco
Pawtuckaway Lake
Lake Waccamaw
n/a
n/a
3
14
80
3
n/a
n/a
4
28
125
6
n/a
n/a
6
45
>180
12
40
58
6
n/a
24
10
74
89
11
n/a
44
15
113
132
16
n/a
69
17
Note: Fast = 1 cm active sediment layer, D (macro-dispersion coefficient) = Iff4 em's'1; Medium = 2 cm active
sediment layer, D = 70"* cm2s~', Slow = 3 cm sediment, D = SxlO"3 cm2s~'.
Overall, forecasted time lags were comparable between models, increasing confidence in
the range of lag time obtained from the modeling results. The greatest difference between the
two models was observed in Pawtuckaway Lake. Forecasted recovery of Pawtuckaway Lake
was 2-3 times faster than SERAFM using the WASP model. This can likely be attributed to
differences in the treatment of the sediment solids balance between the two models. We used the
same parameters in SERAFM as the MCM model applied to Lake Barco by EPRI (EPRI, 2003).
The EPRI study forecasts a slight longer temporal response (54-160 years) to achieve 90% of
steady state concentrations in fish, compared to SERAFM estimates of 14-43 years for fish and
14-48 years for sediments.
3-32
-------
The Brier Creek-WASP/WCS model was only used to forecast a response times in water
and sediments with a 50% reduction in atmospheric deposition. Epilimnion mercury response
times for Brier Creek under the fast, medium and slow scenarios were 8, 11, and 11 years
respectively. These response times should be comparable to the response offish in this system,
depending on the predominance of benthic or pelagic food webs in this system.
The response times forecasted by the SERAFM model for fish mercury concentrations to
reach 90% of steady state with a 50% decline in atmospheric deposition and the zero-out
scenario are presented in Table 3-10 and Figure 3-5.
Table 3-10. Fish Tissue Response Times in Years to Reach 90% of Steady-state
Concentrations Following 50% Mercury Deposition Reductions
Site
Eagle Butte
Lake Barco
Pawtuckaway Lake
Lake Waccamaw
Fast
2
14
34
1
Medium
3
28
56
1
Slow
4
43
64
2
n Epilimnion • Sediment u Fish
o>
E
0)
in
I
in
o>
^
•o
0)
*-
u
4)
"2"
Q.
LB PL
Ecosystem
LW
EB = Eagle Butte, LB = Lake Barco, PL = Pawtuckaway Lake, LW = Lake Waccamaw.
Figure 3- 5. Projected Ecosystem Response Times to Zero-out Deposition Scenario Using
the SERAFM Model
Results presented in Figure 3-5 show SERAFM forecasted response times in the water,
sediments and fish. One artifact of the formulation of the SERAFM model is that fish tissue
residues decline faster than total mercury concentrations in the water and sediments in Figure 4-
5. Fish mercury residues are a function of a BAF that is linked to the MeHg concentrations in
the water column in the SERAFM model. Because MeHg concentrations in water respond more
quickly than the total Hg pools in the water and sediments and the BAF multiplier is applied
3-33
-------
directly to the water column MeHg concentration, the model shows an instantaneous response in
fish mercury.,
A more realistic lag time will be obtained from a bioaccumulation model that takes into
account trophic interactions and bioenergetics drive mercury dynamics. For example, the BASS
model forecasts the comparable response for the same age/length class used to develop a BAF
calibration in SERAFM (average concentration trophic level four species) to be roughly twice
the water column response time (See Figure 3-1). Growth dilution and mortality affect the
response of all fish cohorts to decreased atmospheric loading. The oldest, most heavily
contaminated fish (also captured in the BASS model) take a longer period to show the same
decline in these cohorts (i.e., length/size class). For example, the largest size class for Northern
Pike at Eagle Butte indicated that it would take about 10 years for the most heavily contaminated
fish to respond to a decline in loadings. Overall, the response time of mercury in fish for all sites
is expected to be roughly twice that of the water column response. Simple averaging of residues
for all fish of a species using the BAF approach obscures the fact that cohorts with lower body
burdens respond quicker and the most long-lived cohort responds over a much longer period of
time.
3.4.8 Effect of Land Uses Changes
As mentioned above, we investigated the uncertainty in watershed loading response times
by conducting sensitivity analyses with the WCS model of the Brier Creek watershed. The depth
of soil incorporation significantly influences soil response time. The default of 1 cm was varied
plus and minus 50% to get a range of response times. The loading responses for Upper Brier
Creek are given in Figure 3-6. An initial rapid drop-off in loading (due to instantaneous drop in
deposition to water surfaces and impervious runoff) is followed by a slower drop-off in runoff
and erosion fluxes, controlled by soil mercury concentrations. The 50% loading response varied
between 8, 10, and 15 years for incorporation depths of 0.5, 1.0, and 1.5 cm. The 90% loading
response times were much longer, varying between 35, 70, and 100 years, respectively.
Land use changes can also significantly affect future loading response from a watershed.
Urbanization can increase the total impervious areas in a watershed, decrease the amount of
wetlands, and alter the stream's hydrological response. The net effect of these changes on fish
mercury concentrations is uncertain. Reducing wetlands and hydraulic residence time should
reduce net methylation. On the other hand, covering pervious land with impervious surfaces
should increase the delivery of atmospheric deposition fluxes to the water body. In particular,
changing pervious land use areas to impervious areas will directly affect the delivery of
atmospheric deposition fluxes to the water body.
3-34
-------
Upper Brier Creek Loading Flux Attenuation
28
26
24
lux, mg/ha-yr
-* ro »o
=0 0 M
I«
1
14
12
10
(
\
\
\\
\\
^x^ ^"^~-— ^^_
Upper Brier, base case
Upper Brier, z=0.5
Upper Brier, 2=1.5
50% Response
__90% Response
100% Response
) 10 20 30 40 50 60 70 80 90 100
Time, yr
Figure 3- 6. Upper Brier Creek Loading Flux Attenuation
In the WCS model, impervious areas were assumed to deliver 100% of the deposition
load, while pervious areas delivered a much smaller fraction through runoff and erosion. By
comparison, the SWAT model (Neitsch et al., 2002) assigns a curve number that delivers
virtually all rainfall and associated loads from hydraulically-connected impervious areas, which
average between 73% and 97% of the total impervious area, depending on land use. Although
the fraction of the Brier Creek watershed covered by impervious surfaces is small (about 3% of
the upper watershed), even modest growth over many years could increase the total watershed
delivery of deposited mercury, working against the overall reductions in atmospheric emissions.
Upper Brier Creek Loading Flux Attenuation
.Upper Brier, base case
.0.5% Impervious Growth
Rate
1.0% Impervious Growth
Rate
~50% Response
.90% Response
.100% Response
10
10 20 30 40 50 60 70 80 90 100
Time, yr
Figure 3- 7. Watershed Loading Flux Attenuation Considering Land-use Change
3-35
-------
The loading response of Upper Brier Creek to 3 land use scenarios is shown in the next
Figure 3-7. All scenarios assume an immediate 50% cut in atmospheric deposition. The base
case assumes present land use patterns. The other two scenarios assume modest impervious
surface growth rates of 0.5 % per year and 1 % per year, reaching a total of 4.7% and 7.9%
impervious coverage in 100 years. For the 0.5% scenario, total watershed loads reach a
minimum in 80 years, with watershed loadings stalling at 40% of present levels. For the 1%
scenario, total watershed loads reach a reduction level of 33% in 50 years, and then increase
significantly. After 100 years, the 50% cut in atmospheric deposition would translate into a 25%
drop in ambient watershed loading (Figure 3-7). These simulations are meant to suggest
possible responses to land use changes. If impervious areas deliver only 50% of deposited
mercury, then the 1% growth scenario should follow the 0.5% trend line in Figure 3-7.
3.4.9 Summary
Among the five ecosystems investigated in this study, the range in temporal response to
declines in mercury loading ranged from less than five years to an order of decades. Response is
gauged by the time required for mercury to reach equilibrium at 90% of the initial
concentrations. Fast and slow response scenarios for the five ecosystems ranged from 1 to over
180 years in sediments. The medium response scenarios also varied widely but were generally
on the order of one to three decades for fish mercury concentrations. Forecasted response times
to changes in mercury inputs were longest for Brier Creek, a system strongly influenced by the
watershed mercury loading, and Pawtuckaway Lake, a stratified cold water lake. Shallow, well-
mixed systems like Lake Waccamaw and Lee Dam are projected to respond to changes in
atmospheric deposition in less than a decade.
Consistent with other investigations (e.g., (EPRI, 2003), this analysis showed that
individual ecosystems are highly sensitive to uncertainty in model parameters. Long term
mercury response is strongly influenced by the sediment balance, which varies greatly among
water bodies. Key parameters include watershed erosion and sediment delivery, water body
production, deposition, resuspension and burial rates, and surface sediment mixing depth (active
sediment layer). The effect of epistemic uncertainty (i.e., lack of knowledge) about key mercury
process variables, such as the functional form of equations used to quantify methylation rate
constants, is a major contributor to overall uncertainty that cannot be quantified at this time.
Although scientific understanding of these process variables is rapidly progressing, no modeling
framework can be considered a priori predictive at this time. Accordingly, all models must be
parameterized to specific ecosystems being investigated to develop a credible set of rate
constants and process terms.
Our best available, practicable science suggests that over the long-term (i.e., at steady
state), the change in mercury concentrations in freshwater fish will be proportional to changes in
mercury inputs. Thus, in systems where atmospheric deposition of inorganic mercury is the
major source of mercury to surface waters, long-term changes in fish mercury concentrations
will be proportional to declines in atmospheric deposition as suggested by the MMaps approach.
There is some discrepancy between results from our modeled scenarios based on the MMaps
model that can be attributed to uncertainty in both the models and data. Where there are
sources of mercury in the watershed other than atmospheric loading, the response offish tissue
concentrations will be affected in a manner that is not proportional to changes in atmospheric
3-36
-------
deposition. Preliminary findings of the METALLICUS study show negligible concentrations of
deposited mercury in fish three years after the addition of labeled mercury isotopes to the
watershed, supporting the supposition that mercury deposited on the watershed takes
significantly longer to accumulate in fish than mercury deposited directly on the surface of a
waterbody (Pers. Comm., R. Harris).
A preliminary assessment of the expected effect of land-use changes on fish mercury
concentrations for a watershed dominated system illustrates changes like urbanization within a
watershed can alter the magnitude and timing offish mercury concentrations. A number of peer-
reviewed studies have shown that other environmental changes such as enhanced nutrient
loading in Long Island Sound (Hammerschmidt et al., 2004; Hammerschmidt and Fitzgerald,
2004), sulfate deposition in the Everglades (Krabbenhoft et al., presentation to EPA 2/7/05) and
reservoir flooding can also dramatically change fish mercury concentrations independently of
changes in total mercury additions to the ecosystem. With further advances in knowledge and
new data on mercury fate and transport, EPA's long term goal is to provide tools that have the
capability to provide a priori forecasts of ecosystem responses to changes in mercury deposition.
3.5 National Scale Ecosystem Variability
As mentioned in the introductory sections of this report, different ecosystems exhibit
dramatically different responses to changes in mercury loading depending on their chemical and
physical attributes. Results from individual case studies must therefore be qualified by their
representativeness of ecosystem condition and variability across the United States. Using
georeferenced empirical databases that describe some of the watershed and waterbody
characteristics across the United States, we present the ecosystem case studies along the gradient
of data known to be important for MeHg formation and bioaccumulation in fish. Although this
analysis has not been completed, the concept is demonstrated by a. preliminary qualitative
assessment of how much of the ecosystem variability in MeHg formation and bioaccumulation
has been captured by the modeling case studies. Developing broad categories of ecosystem
types based on their propensity for MeHg formation and bioaccumation in fish and their
frequency of occurrence is an iterative effort. By combining the frequency of each category of
ecosystem type with the magnitude and time lag in fish tissue reductions modeled using
dynamic, ecosystem scale models, such an analysis could ultimately provide a different
methodology for a national scale assessment of expected changes in fish MeHg concentrations
resulting from reductions in atmospheric mercury deposition.
3.5.1 United States Lakes Distribution
We used the USGS National Hydrography Database (NHD)(1:100,000 scale) to obtain
information on the frequency of different lake sizes across the country
For each Electricity Generating Unit (EGU) we constructed a circular buffer area having
a 50 km radius using GIS software (ArcMap, ArcView). This buffer zone was chosen to
reflect the geographic extent of the majority of elevated deposition from a local source
(Expert Panel on Mercury Atmospheric Processes, 1994). This distance is not meant to
conclusively define the extent of local deposition, but is used to approximate near-field
characteristics of the areas surrounding EGUs.
The NHD lake coverage was then clipped using the 50km radii.
3-37
-------
• The lake area was divided into three different classes and the total number in each class
was calculated across all of the buffered areas (see Table 3-11).
Table 3-11. Frequency of Different Lake Sizes Across the United States
Lake Type Frequency All U.S. 50 km Radius Around EGUs
Small 1 (0.01 -0.1 km2) 328,564 119,492
Medium (0.1-10 km2) 64,260 23,563
Large Large (>10 km2) 1,200 442
From this analysis, it is apparent that there are a large number of lakes (<0.1 km2)
surrounding the EGUs across the country. Small, well-mixed shallow systems (like Eagle Butte
and Lake Barco), tend to respond more rapidly to changes in mercury deposition than larger,
deeper lakes. The temporal response of these systems is also driven by the catchment to lake
area ratio of these systems, which was beyond the scope of our initial analysis. Generally,
systems where mercury dynamics are dominated by the response of the watershed to changes in
loading (like Brier Creek), will respond more slowly to changes in atmospheric deposition.
Ecosystems differ dramatically in their propensity to convert incoming inorganic
mercury to methylmercury. Other environmental factors that are generally accepted as
reasonable ancillary variables for MeHg formation in waterbodies include: dissolved organic
carbon content of surface waters, pH, sulfate, % wetland area of the catchment, sulfide
concentrations and sediment organic matter. National scale coverages of these data are currently
under development by a number of groups. In this assessment, we assembled the available data
on sulfate deposition, % wetland coverage, and total mercury deposition projected by the CMAQ
model for 2001. Over the long term, by combining these data layers, a statistical model of
MeHg production will be developed to characterize ecosystems affected by atmospheric inputs
of mercury.
Data layers used to generate Figures 3-8 through 3-13 are described below:
Model ecosystems relative to EGUs across the country. We used EGRID 2002 Version
2.01 Plant File (Year 2000 data) to generate the data points representing the EGUs. We
used lat/long to create point coverage for the U.S.
WETLANDS: We used the NLCD 1993 land cover data.
• Mean wet sulfate deposition 1987-1999 (kg/ha/yr). Sulfur deposition aggregated by
HUC based on data from "Enhanced wet deposition estimates using modeled
precipitation inputs", Grimm, Jeffrey, W., James A. Lynch. 2004. Environmental
Monitoring and Assessment Vol. 90 P. 243-268.
CMAQ deposition 2001 scenario. Source: Russ Bullock, NOAA-ARL and
EPA/ORD/NERL
FISH tissue data for all US. Figure 3-12 is based on raw data from the USGS EMMA
database (http://emmma.usgs.gov/fishHgAbout.aspx and
http://emmma.usgs.gov/datasets.aspx ). Figure 3-13 shows the normalized fish tissue
data from NLFA and NLFTS databases used in the Regulatory Impact Assessment.
3-38
-------
3.5.2 Summary
Extrapolating the results of the case studies to other systems is an ongoing challenge.
Qualitatively, the ecosystem case studies fall in the mid-range of sulfate deposition and do not
capture much of the variability across the country. Sites capture a wide range of % wetland
coverage and appear to be in regions of predominantly moderate mercury deposition rates, but
they do not capture the tails of this distribution. Based on this preliminary assessment it is likely
that the case studies represent ecosystem types of moderate methylation potential across the
United States. This is supported by fish tissue concentrations measured in these regions, which
generally fall within the impacted category that is above EPA's 0.3 ppm tissue residue guideline
(US EPA, 2001) but do not represent the extremes in fish mercury concentrations observed on a
national scale. Therefore, while these ecosystem case studies cover the bulk of the distributions
of the key environmental characteristics that will affect MeHg production, they may miss the
tails of the distributions for some characteristics.
3-39
-------
Model Ecosystems for Mercury Benefits Analysis
1000
.„..„ .„. . • Model Systems
1000 Kilometers Mod* watersh
• EGUs
I I National B oundary (Comeiiiitnous US(
r~~l CMAQ Deposition GiM (3ikm2)
Figure 3- 8. Model Ecosystem Locations with CMAQ Grid Cells and Locations of Electricity Generating Units (EGUs)
3-40
-------
Estimated Mean Sulfate Deposition to Watersheds
Aggregated by 8 Digit Hydrologic Unit Code
18.4
1000
1000 Kilometers
w
| | Model w«t*r jhedi
Sort** Deposition (kg* 100V)
I 1 1.31 10821
10321 - 13213
15213- 18,631
19631- 23221
23.221- 33.609
National Boundaty (Conterminous US)
d
1
Figure 3- 9. Model Ecosystem Locations with Gradient in Sulfate Deposition Across the Eastern US
3-41
-------
Watershed Percent Wetlands Land Cover
(Aggregated by 8 Digit Hydrologic Unit Code)
1000
1000 Kilometers
I | Mod«t Watersheds
P«re«nt W«tland Ar»a
| | 0 - 4.94
B4.94 • 16.434
16.434-34.647
BB 34.647 - 59.59
• 59.59-94.549
Figure 3- 10. Model Ecosystem Locations with Percent Wetland Area Aggregated for Each HUC
3-42
-------
CMAQ Total Mercury Deposition
Year 2001 (ug/m2/yr)
• Mod«) SrtM
I I Mod«l Wit«r»h«d»
1000
1000 Kilometers
3.348 - 10.127
10.127- 14.787
1*.787. 20.413
20.413 • 32.234
32.234- 133 .223
Figure 3- 11. Model Ecosystem Locations with CMAQ 2001 Total Mercury Deposition
3-43
-------
Methylmercury Residues in Fish Tissue
• Mou«US>
: *.!)•».»
• «.» 1.SJ
• 1.W-4.5
• 4.5 ^».«
Figure 3- 12. Model Ecosystem Locations with Measured Fish Tissue Concentrations
3-44
-------
03 to 0.5
• 0.5 to 1
• 1 to 16
* 12,325 post 1995 samples over 0.3 ppm.
* 86 samples ate over 3 ppm.
Source: EPA NLFA and NLFTS
Figure 3- 13. Measured 1995-2001 Fish Hg Concentrations > 0.3 ppm
3-45
-------
3.6 References
Ambrose, R.B., 1987. Modeling Volatile Organics in the Delaware Estuary. Journal of
Environmental Engineering, 113(4): 703-721.
Ambrose, R.B. and Wool, T.A., 2001. Modeling Tools Used for Mercury TMDLs in Georgia
Rivers. In: K. Hatcher (Editor), Proceedings of the 2001 Georgia Water Resources
Conference, March 26-27, 2001. University of Georgia: Athens, Georgia, Athens, GA,
pp. 532-535.
Ambrose, R.B.e.a., 1988. WASP4, A Hydrodynamic and Water Quality Model-Model Theory,
User's Manual, and Programmer's Guide., U.S. Environmental Protection Agency,
Athens, GA. EPA/600/3-87-039.
Amyot, M., Lean, D.R.S., Poissant, L. and Doyon, M.-R., 2000. Distribution and transformation
of elemental mercury in the St. Lawrence River and Lake Ontario. Canadian Journal of
Fisheries and Aquatic Sciences, 57 (Suppl. 1): 155-163.
Atkeson, T., 2003. The Everglades Mercury TMDL Pilot Study: Final Report by the Florida
Department of Environmental Protection
(http://www.dep.state.fl.us/secretary/news/2003/nov/pdf/mercury_report_summary.pdf).
Babiarz, C.L. et al., 2001. Partitioning of total mercury and methylmercury to the colloidal phase
in freshwaters. Environmental Science and Technology, 35(24): 4773-4782.
Balogh, S.J., M.L., M. and D.K., J., 1998. Transport of Mercury in Three Contrasting River
Basins. Environmental Science and Technology, 32: 456-462.
Barber, M., 2003. A review and comparison of models for predicting dynamic chemical
bioconcentration in fish. Environmental Toxicology and Chemistry, 22(9): 1963-1992.
Barber, M.C., 2001. Bioaccumulation and Aquatic System Simulator (BASS) User's Manual
Beta Test Version 2.1, U.S. Environmental Protection Agency, Office of Research and
Development, EPA/600/R-01/035, Athens, GA, USA.
Barber, M.C., Suarez, L.A. and Lassiter, R.R., 1987. FGETS (Food and Gill Exchange of Toxic
Substances): A simulation model for predicting the bioaccumulation of nonpolar organic
pollutants by fish, U.S. Environmental Protection Agency, Office of Research and
Development, EPA/600/3-87/038, Athens, GA.
Barber, M.C., Suarez, L.A. and Lassiter, R.R., 1988. Modeling bioconcentration of nonpolar
organic pollutants by fish. Environmental Toxicology and Chemistry, 7: 545-558.
Barber, M.C., Suarez, L.A. and Lassiter, R.R., 1991. Modelling bioaccumulation of organic
pollutants in fish with an application to PCBs in Lake Ontario salmonids. Canadian
Journal of Fisheries and Aquatic Sciences, 48: 318-337.
3-46
-------
Barkay, T., Miller, S.M. and Summers, A.O., 2003. Bacterial mercury resistance from atoms to
ecosystems. FEMS Microbiology Reviews, 27: 355-384.
Benoit, J., Gilmour, C.C., Mason, R.P. and Heyes, A., 1999. Sulfide controls on mercury
speciation and bioavailability to methylating bacteria in sediment pore waters.
Environmental Science and Technology, 33(6): 951-957.
Benoit, J.M., Gilmour, C.C., Heyes, A., Mason, R.P. and Miller, C., 2003. Geochemical and
Biological Controls over Methylmercury Production and Degradation in Aquatic
Systems, Biogeochemistry of Environmentally Important Trace Metals. ACS Symposium
Series 835.
Benoit, J.M., Gilmour, C.C. and Mason, R.P., 2001. The influence of sulfide on solid-phase
mercury bioavailability for methylation by pure cultures of Desulfobulbus propionicus
(Ipr3). Environmental Science and Technology, 35(1): 127-132.
Benoit, J.M., Gilmour, C.C., Mason, R.P., Riedel, G.S. and Reidel, G.F., 1998. Behavior of
mercury in the Patuxent River estuary. Biogeochemistry, 40: 249-265.
Benoit, J.M., Mason, R.P. and Gilmour, C.C., 1999. Estimation of mercury-sulfide speciation in
sediment pore waters using octanol-water partitioning and implications for availability to
methylating bacteria. Environmental Toxicology and Chemistry, 18(10): 2138-2141.
Beyers, D.W., Rice, J.A. and Clements, W.H., 1999. Evaluating biological significance of
chemical exposure to fish using a bioenergetics-based stressor-response model. Canadian
Journal of Fisheries and Aquatic Sciences, 56: 823-829.
Bigham, G.N. and Vandal, G.M., 1994. A drainage basin perspective of mercury transport and
bioaccumulation: Onondaga Lake, New York, Twelfth International Neurotoxicology
Conference, Hot Springs, Arkansas USA.
Bloom, N.S., 1992. On the chemical form of mercury in edible fish and marine invertebrate
tissue. Canadian Journal of Fisheries and Aquatic Sciences, 49: 1010-1017.
Bowie, G. et al., 1985. Rates, Constants, and Kinetics Formulations in Surface Water Quality
Modeling (Second Edition), U.S. EPA, Athens, GA, EPA/600/3-85/040, June, 1985.
Branfireun, B. and Roulet, N., 2002. Controls on the fate and transport of methylmercury in a
boreal headwater catchment, northwestern Ontario, Canada. Hydrology and Earth System
Sciences, 6(4): 785-794.
Bullock, R.O. and Brehme, K., 2002. Atmospheric mercury simulation using the CMAQ model:
Formulation description and analysis of wet deposition results. Atmospheric
Environment, 36: 2135-2146.
3-47
-------
Carpi, A. and Lindberg, S.E., 1997. Sunlight-mediated emission of elemental mercury from soil
amended with municipal sewage sludge. Environmental Science and Technology, 31(7):
2085-2091.
Cohen, M. et al., 2004. Modeling the Atmospheric Transport and Deposition of Mercury to the
Great Lakes. Environmental Researrch, 95: 247-265.
Compeau, G.C. and Bartha, R., 1987. Effect of salinity on mercury-methylating activity of sufate
reducing bacteria in estuarine sediments. Applied and Environmental Microbiology, 53:
261-265.
Connolly, J. and Thomann, R.V., 1985. WASTOX, A Framework for Modeling the Fate of
Toxic Chemicals in Aquatic Environments. Part 2: Food Chain., U.S. Environmental
Protection Agency, Gulf Breeze, FL and Duluth, MN.
Craig, P.J. and Bartlett, P.D., 1978. The role of hydrogen sulphide in environmental transport of
mercury. Nature, 275: 635-637.
Craig, P.J. and Moreton, P.A., 1986. Total mercury, methyl mercury and sulphide levels in
British estuarine sediments-Ill. Water Research, 20(9): 1111-1118.
Di Toro, D.M., Fitzpatrick, J.J. and Thomann, R.V., 1983. Water Quality Analysis Simulation
Program (WASP) and Model Verification Program (MVP) - Documentation.,
Hydroscience, Inc., Westwood, NY, for U.S. EPA, Duluth, MN, Contract No. 68-01-
3872.
Doyon, J.-F., Schetagne, R. and Verdon, R., 1998. Different mercury bioaccumulation rates
between sympatric populations of dwarf and normal lake whitefish (Coregonus
clupeaformis) in the La Grande complex watershed, James Bay, Quebec.
Biogeochemistry, 40: 203-216.
Driscoll, C.T. et al., 1995. The role of dissolved organic carbon in the chemistry and
bioavailability of mercury in remote Adirondack lakes. Water, Air, Soil Pollution, 80:
499-508.
EPRI, 2003. Factors Affecting Predicted Responses of Fish Mercury Concentrations to Changes
in Mercury Loading., Electric Power Research Institute Report 1005521, Palo Alto, CA.
Expert Panel on Mercury Atmospheric Processes, 1994. Mercury Atmospheric Processes: A
Synthesis Report. In: R.H. Osa (Editor), Mercury Atmospheric Processes. Environment
Canada, Tampa, Florida, pp. 23.
Fitzgerald, W.F., Mason, R.P., Vandal, G.M. and Dulac, F., 1994. Air-water cycling of mercury
in lakes. In: C.J. Watras and J.W. Huckabee (Editors), Mercury Pollution: Integration and
Synthesis. Lewis Publishers, Chelsea, MI, pp. 203-220.
3-48
-------
Frescholtz, T. and Gustin, M.S., 2004. Soil and foliar mercury emission as a function of soil
concentration. Water, Air, and Soil Pollution, 155: 223-237.
Gbundgo-Tugbawa and Driscoll, 1998. Application of the regional mercury cycling model
(RMCM) to predict the fate and remediation of mercury in Onondaga Lake, New York.
Water, Air, and Soil Pollution, 105: 417-426.
Gilmour, C.C. and Henry, E.A., 1991. Mercury methylation in aquatic systems affected by acid
deposition. Environmental Pollution, 71: 131-169.
Gilmour, C.C. et al., 1998. Methylmercury concentrations and production rates across a trophic
gradient in the northern Everglades. Biogeochemistry, 40: 327-345.
Gobas, F.A.P.C., Pasternak, J.P., Lien, K. and Duncan, R.K., 1998. Development and field
validation of a multimedia exposure model for waste load allocation in aquatic
ecosystems: application to 2,3,7,8-tetrachloro-p-dioxin and 2,3,7,8-
tetrachlorodibenzofiiran in the Fraser River watershed. Environmental Science and
Technology, 32: 2442-2449.
Gobas, F.A.P.C., Z'Graggen, M.N. and Zhang, X., 1995. Time response of the Lake Ontario
ecosystem to virtual elimination of PCBs. Environmental Science and Technology, 29(8):
2038-2046.
Greenfield, B.K., Hrabik, T.R., Harvey, C.J. and Carpenter, S.R., 2001. Predicting mercury
levels in yellow perch: use of water chemistry, trophic ecology, and spatial traits.
Canadian Journal Fisheries and Aquatic Sciences, 58: 1419-1429.
Greenfield, J., Dai, T. and Manguerra, H., 2002. Watershed Modeling Extensions of the
Watershed Characterization System., Ft. Lauderdale, Florida, USA, February 2002.
Grieb, T.M. et al., 1990. Factors affecting mercury accumulation in fish in the upper Michigan
peninsula. Environmental Toxicology and Chemistry, 9: 919-930.
Gustin, M. et al., 2004. Application of controlled mesocosms for understanding mercury air-soil-
plant exchange. Environmental Science and Technology, 38: 6044-6050.
Hammerschmidt, C., Fitzgerald, W., Lamborg, C., Balcom, P. and Visscher, P., 2004.
Biogeochemistry of methylmercury in sediments of Long Island Sound. Marine
Chemistry, 90: 31-52.
Hammerschmidt, C.R. and Fitzgerald, W.F., 2004. Geochemical controls on the production and
distribution of methylmercury in near-shore marine sediments. Environmental Science
and Technology, 38(5): 1487-1495.
Harris, R., Gherini, S. and Hudson, R., 1996. Regional Mercury Cycling Model: A Model for
Mercury Cycling in Lakes, R-MCM Version 1.0 Draft User Guide and Technical
3-49
-------
Reference, Electric Power Research Institute, Wisconsin Department of Natural
Resources, Lafayette, California.
Harris, R. and Hutchison, D., 2003. Factors Affecting the Predicted Response of Fish Mercury
Concentrations to Changes in Mercury Loading. 1005521, Electric Power Research
Institute, Palo Alto, CA.
Harris, R.C. et al., 2004. The METAALICUS Project: Overview of Study and Results To-Date,
7th Internation Conference on Mercury as a Global Pollutant, Slovenia.
Harris, R.C., Hutchinson, D. and Beals, D., 2002. Lake Superior Mercury Cycling Model.
Presented at the annual meeting of the International Association of Great Lakes Research
(IAGLR), June 2002.
Hintelmann, H. and Evans, R.D., 1997. Application of stable isotopes in environmental tracer
studies - measurement of monomethylmercury by isotope dilution ICP-MS and detection
of species transformation. Fresenius Journal of Analytical Chemistry, 358: 378-385.
Hintelmann, H. et al., 2002. Reactivity and mobility of new and old mercury deposition in a
boreal forest ecosystem during the first year of the METAALICUS study. Environmental
Science and Technology, 36: 5034-5040.
Hintelmann, H., Keppel-Jones, K. and Evans, R.D., 2000. Constants of mercury methylation and
demethylation rates in sediments and comparison of tracer and ambient mercury
availability. Environmental Toxicology and Chemistry, 19(9): 2204-2211.
Hurley, J. et al., 1995. Influences of watershed characteristics on mercury levels in Wisconsin
rivers. Environmental Science and Technology, 29(7): 1867-1875.
Hurley, J.P., Cowell, S.E., Shafer, M.M. and Hughes, P.E., 1998. Tributary loading of mercury
to Lake Michigan: Importance of seasonal events and phase partitioning. Science of the
Total Environment, 213: 129-137.
Iverfeldt, A., 1991. Water, Air, and Soil Pollution, 56: 553-564.
Jackson, T.A., 1998. Mercury in aquatic ecosystems. In: W.J. Langston and M.J. Bebianno
(Editors), Metal Metabolism in Aquatic Environments. Chapman & Hall, London, pp.
77-158.
Kamman, N., Driscoll, C.T., Estabrook, B., Evers, D.C. and Miller, E.K., 2004. Biogeochemistry
of Mercury in Vermont and New Hampshire Lakes An Assessment of Mercury in Water,
Sediment and Biota of Vermont and New Hampshire Lakes Comprehensive Final Project
Report May, 2004, Project Funding Provided by United States Environmental Protection
Agency Office of Research and Development under the Regional Environmental
Monitoring and Assessment Program.
3-50
-------
Karlsson, T. and Skyllberg, U., 2003. Bonding of ppb levels of methyl mercury to reduced sulfur
groups in soil organic matter. Environmental Science and Technology, 37: 4912-4918.
Kidd, K., Hesslein, R., Fudge, R. and Hallard, K., 1995. The influence of trophic level as
measured by delta-N-15 on mercury concentrations in fresh-water organisms. Water, Air,
and Soil Pollution, 80(1-4): 1011-1015.
King, J.K., Kostka, I.E., Frischer, M.E., Saunders, F.M. and Jahnke, R.A., 2001. A quantitative
relationship that demostrates mercury methylation rates in marine sediments are based on
community composition and activity of sulfate-reducing bacteria. Environmental Science
and Technology, 35(12): 2491-2496.
Lalonde, J., Amyot, M., Kraepiel, A. and Morel, F., 2001. Photooxidation of Hg(0) in artificial
and natural waters. Environmental Science and Technology, 35: 1367-1372.
Landis, M. and Keeler, G., 2002. Atmospheric mercury deposition to Lake Michigan during the
Lake Michigan mass balance study. Environmental Science and Technology, 36(21):
4518-4524.
Landis, M.S., Lynam, M. and Stevens, R.K., 2004. The Monitoring and Modeling of Mercury
Species in Support of Local Regional and Global Modeling. In: N. Pirrone and K.R.
Mahaffey (Editors), Dynamics of Mercury Pollution on Regional and Global Scales.
Kluwer Academic Publishers, New York, NY.
Laporte, J.-M., Andres, S. and Mason, R.P., 2002. Effect of ligands and other metals on the
uptake of mercury and methylmercury across the gills and the intestine of the blue crab
(Callinectes sapidus). Comparative Biochemistry and Physiology Part C, 131: 185-196.
Lawrence, A.L. and Mason, R.P., 2001. Factors controlling the bioaccumulation of mercury and
methylmercury by the estuarine amphipodLeptocheirusplumulosus. Environmental
Pollution, 111:217-231.
Lawrence, A.L., Mcaloon, K.M., Mason, R.P. and Mayer, L.M., 1999. Intestinal solubilization of
particle associated organic and inorganic mercury as a measure of bioavailability to
benthic invertebrates. Environmental Science and Technology, 33(11): 1871-1876.
Lawson, N.M. and Mason, R.P., 1998. Accumulation of mercury in estuarine food chains.
Biogeochemistry, 40: 235-247.
Lawson, N.M., Mason, R.P. and Laporte, J.M., 2001. The fate and transport of mercury,
methylmercury, and other trace metals in Chesapeake Bay tributaries. Water Research,
35:501-515.
Leaner, J.J. and Mason, R.P., 2002. Factors controlling the bioavailability of ingested
methylmercury to channel catfish and atlantic sturgeon. Environmental Science and
Technology, 36:5124-5129.
3-51
-------
Lee, Y.H., Bishop, K., Pettersson, C., Iverfeldt, I. and Allard, B., 1995. Subcatchment Output of
Mercury and Methylmercury at Svartberget in Northern Sweden. Water, Air, and Soil
Pollution, 80: 455-465.
Lin, C.-J. and Pehkonen, S.O., 1999. The chemistry of atmospheric mercury: a review.
Atmospheric Environment, 33: 2067-2079.
Lindberg, S.E. and Stratton, J.E., 1998. Atmospheric mercury speciation: concentrations and
behavior of reactive gaseous mercury in ambient air. Environmental Science and
Technology, 32(1): 49-57.
Mason, R.P. and Lawrence, A.L., 1999. Concentration, distribution, and bioavailability of
mercury and methylmercury in sediments of Baltimore Harbor and Chesapeake Bay,
Maryland, USA. Environmental Toxicology and Chemistry, 18(11): 2438-2447.
Mason, R.P. et al., 1999. Mercury in the Chesapeake Bay. Marine Chemistry, 65: 77-96.
Mason, R.P., Morel, F.M.M. and Hemond, H.F., 1995. The role of microorganisms in elemental
mercury formation in natural waters. Water, Air, and Soil Pollution, 80: 775-787.
Matilainen, T. and Verta, M., 1995. Mercury methylation and demethylation in aerobic surface
waters. Canadian Journal of Fisheries and Aquatic Sciences, 52: 1597-1608.
Mierle, G. and Ingram, R., 1991. The role of humic substance in the mobilization of mercury
from watersheds. Water, Air, and Soil Polution, 56: 349-357.
Morel, F., Kraepiel, A.M.L. and Amyot, M., 1998. The chemical cycle and bioaccumulation of
mercury. Annual Reviews of Ecological Systems, 29: 543-566.
Neitsch, S.L., Arnold, J.G., Kiniry, J.R., Williams, J.R. and King, K.W., 2002. Soil and Water
Assessment Tool theoretical documentation - version 2000. TWRI Report TR-191.,
Texas Water Resources Institute, College Station, Texas.
Omernik, J.M., 1987. Ecoregions of the conterminous United States. Map (scale 1:7,500,000).
Annals of the Association ofAmerican Geographers, 77: 119-125.
Pollman, C. et al., 1991. Preliminary analysis of the hydrologic and geochemical controls on
acid-neutralizing capcity in two acidic seepage lakes in Florida. Water Resources
Research, 27(9): 2321-2355.
Rasmussen, P.E., 1994. Current methods of estimating atmospheric mercury fluxes in remote
areas. Environmental Science and Technology, 28(13): 2233-2241.
Ravichandran, M., 2004. Interactions between mercury and dissolved organic matter - a review.
Chemosphere, 55: 319-331.
3-52
-------
Rolfhus, K. et al., 2003. Distribution and fluxes of total and methylmercury in Lake Superior.
Environmental Science and Technology, 37(5): 865-872.
Schaefer, J. et al., 2004. Role of the bacterial organomercury lyase (MerB) in controlling
methylmercury accumulation in mercury-contaminated natural waters. Environmental
Science and Technology, 38: 4304-4311.
Schnoor, J.L., et al., 1987. Processes, Coefficients, and Models for Simulating Toxic Organics
and Heavy Metals in Surface Waters, t U.S. EPA, Athens, GA, EPA/600/3-87/015, June
1987.
Scholtz, M.T., Heyst, B.J.V. and Schroeder, W.H., 2003. Modelling of mercury emissions from
background soils. Science of the Total Environment, 304: 185-207.
Schroeder, W.H., Munthe, J. and Lindqvist, O., 1989. Cycling of mercury between water air and
soil compartments of the environment. Water, Air, and Soil Pollution, 48: 337-347.
Seigneur, C., Jayaraghavan, K., Lohman, K., Karamchandani, P. and Scott, C., 2004. Global
Source Attribution for Mercury Deposition in the United States. Environmental Science
and Technology, 38: 555-569.
Sellers, P., Kelly, C.A., Rudd, J.W.M. and MacHutchon, A.R., 1996. Photodegradation of
methylmercury in lakes. Nature, 380: 694.
Siciliano, S., O'Driscoll, N. and Lean, D., 2002. Microbial reduction and oxidation of mercury in
freshwater lakes. Environmental Science and Technology, 36(14): 3064-3068.
St. Louis, V. et al., 2001. Importance of the forest canopy to fluxes of methyl mercury and total
mercury to boreal ecosystems. Environmental Science and Technology, 35: 3089-3098.
Sunda, W.G. and Huntsman, S.A., 1998. Processes regulating cellular metal accumulation and
physiological effects: Phytoplankton as model systems. Science of the Total
Environment, 219: 165-181.
Sunderland, E.M. et al., 2004. Speciation and bioavailability of mercury in well-mixed estuarine
sediments. Marine Chemistry, 90: 91-105.
Tetra Tech Inc., 2000. Florida Pilot Mercury Total Maximum Daily Load (TMDL) Study:
Application of the Everglades Mercury Cycling Model (E-MCM) to Site WCA 3A-15.
Prepared for the United States Environmental Protection Agency and Florida Department
of Environmental Protection. October 2000.
Tseng, C.M., Amouroux, D., Abril, G. and Donard, O.F.X., 2001. Speciation of mercury in a
fluid mud profile of a highly turbid macrotidal estuary (Gironde, France). Environmental
Science and Technology, 35(13): 2627-2633.
3-53
-------
Tsiros, I., 2002. Modeling assessment of air emission flux of mercury from soils in terrestrial
landscape components: Model tests and sensitivities. Journal of Air and Waste
Management Association, 52(3): 339-348.
Tsiros, I.X., 1999. A Modeling Analysis of Factors Influencing Mass Balance Components of
Airborne Deposited Mercury in Terrestrial Landscapes. J. Environ. Sci. Health A, 34(10):
1079-2005.
U.S. EPA, 2000. Mercury Research Strategy. Unites States Environmental Protection Agency,
Washington, DC.
US EPA, 2000. Level III ecoregions ofthe continental United States (revision of Omernik, 1987).
US Environmental Protection Agency National Health and Environmental Effects
Research Laboratory, Western Ecology Division, Corvallis, OR.
US EPA, 2001. Mercury Maps. A Quantitative Spatial Link Between Air Deposition and Fish
Tissue. Final Report. EPA/823/R-01/009, USEPA, Washington, D.C.
US EPA, 2001. Methylmercury fish tissue residue criterion, United States Environmental
Protection Agency, Office of Water, 4304 EPA-823-F-01-001, January 2001,
www.epa.gov/waterscience/criteria/methvlmercury /factsheet.html.
USEPA, 1997. Mercury Study Report to Congress. EPA-452/R-97-005, Office of Air Quality
Planning and Standards
United States Environmental Protection Agency, Washington.
Watras, C.J. et al., 1998. Bioaccumulation of mercury in pelagic freshwater food webs. Science
ofthe Total Environment, 219(2-3): 183-208.
Watras, C.J. and Bloom, N.S., 1992. Mercury and methylmercury in individual zooplankton:
implications for bioaccumulation. Limnol. Oceanogr., 37: 1313-1318.
Wren, C.D. and MacCrimmon, H.R., 1986. Comparative bioaccumulation of mercury in two
adjacent freshwater ecosystems. Water Research, 6: 763-769.
Zhang, H. and Lindberg, S.E., 2001. Sunlight and Iron(III)-Induced Photochemical Production of
Dissolved Gaseous Mercury in Freshwater. Environmental Science and Technology, 35:
928-935.
3-54
-------
SECTION 4 4-1
PROFILE OF FISHING ACTIVITY IN THE U.S 4=1
4.1 Industry Characterization 4-1
4.1.1 Introduction and Overview of the Fishing Industry 4-1
4.1.2 Commercial Fishing 4-2
4.1.3 Recreational Fishing 4-6
4.2 U.S. Production Statistics for Commercial and Recreational Fishing 4-8
4.2.1 Commercial Fishing 4-8
4.2.2 Recreational Fishing 4-15
4.3 U.S. Demand for Commercial and Recreational Fishing 4-19
4.3.1 Commercial Imports 4-19
4.3.2 U.S. Demand for Commercial Fishery Products 4-24
4.3.3 Recreational Fishing 4-26
4.3.4 Total U.S. Demand 4-35
4.4 Economic Value of Key Species 4-36
4.4.1 Finfish 4-36
4.4.2 Shellfish 4-37
4.4.3 Fish Products 4-39
4.5 Characterization of Fish Consuming Populations 4-39
4.5.1 Fish Consumption Pathways and Associated Fish-Consuming
Populations 4-40
4.5.2 Fish Consuming Populations 4-41
4.5.3 General Fish Consumption Rates for Key Fish Consuming
Populations 4-42
4.5.4 Discussion of Population and Fish Consumption Data in the Context of the
Mercury Benefits Analysis 4-45
4.6 Summary 4-47
4.6.1 Commercial Fish Production, Demand, and Consumption 4-47
4.6.2 Recreational Fishing Activity, and Consumption 4-48
4.6.3 Overall Conclusions 4-49
4.7 References 4-49
-------
Tables
Table 4-1. Finfish Fishing Industry (NAICS code 114111) 4-4
Table 4-2. Shellfish Fishing Industry (NAICS code 114112) £4
Table 4-3. Value of Aquacultural Products Sold, 1998 4-5
Table 4-4. Number of Anglers and Fishing Licenses in the United States 4-7
Table 4-5. Annual Domestic Landings for Commercial Fishing (Finfish and Shellfish) .... 4-8
Table 4-6. Commercial Fish Landings by Region 4-10
Table 4-7. 2002 U.S. Commercial Fish Landings by End Use 4-10
Table 4-8. 2002 U.S. Commercial Landings by Month 4-11
Table 4-9. Finfish Species with the Highest 2002 U.S. Commercial Landings 4-11
Table 4-10. Shellfish Types with the Highest 2002 Commercial Landings 4-12
Table 4-11. Total 2002 U.S. Aquacultural Production 4-13
Table 4-12. 2002 Exports of Edible Finfish Products 4-14
Table 4-13. 2002 Exports of Edible Shellfish Products 4-15
Table 4-14. 2002 Recreational Marine Landings for Selected Finfish Types 4-16
Table 4-15. 2002 Recreational Marine Catch and Harvest by Region and Top Species
Group 4-18
Table 4-16. 2002 Imports of Edible Fresh or Frozen Finfish Products 4-22
Table 4-17. 2002 Imports of Edible Canned and Cured Finfish Products 4-23
Table 4-18. 2002 Imports of Edible Fresh and Frozen Shellfish Products 4-24
Table 4-19. 2002 Imports of Edible Canned Shellfish Products 4-24
Table 4-20. 2002 U.S. Demand for Commercial Finfish (metric tons) 4-25
Table 4-21. 2002 U.S. Demand for Commercial Shellfish (metric tons) 4-25
Table 4-22. 2002 U.S. Consumption of Commercial Fishery Products 4-26
Table 4-23. Number of Anglers and Days of Fishing for 2001 4-27
Table 4-24. 2001 U.S. Recreational Freshwater Fishing: Targeted Species by Region 4-28
Table 4-25. Freshwater Fishing Bag and Size Limits for Selected States and Species 4-31
Table 4-26. Saltwater Fishing Bag and Size Limits for Selected States and Species 4-32
Table 4-27. Percent of Freshwater Anglers by Age Group 4-33
Table 4-28. Percent of Anglers by Sex and Age Group 4-33
Table 4-29. 2001 Demographic Summary, Angler Race (% Non-White) 4-34
Table 4-30. Incidence of Fishing Among White, African American, and Hispanic Recreational
Boaters 4-34
Table 4-31. Percent of U.S. Population Who Fished (By Household Income) 4-35
Table 4-32. 2002 Economic Value of Commercial Landings for Important Finfish Types .. 4-36
Table 4-33. Average 2001 Wholesale Prices for Several Fresh Finfish Types 4-37
Table 4-34. 2002 Economic Value of Commercial Landings of Several Shellfish Types ... 4-38
Table 4-35. Average 2001 Wholesale Prices for Several Fresh Shellfish Types 4-39
Table 4-36. 2002 Economic Value of Commercial Fishery Products 4-39
Table 4-37. Demographic (count) Data for Key Fish Consuming Populations in the U.S. .. 4-43
Table 4-38 Fish Consumption Rates for Key Fish Consuming Populations in the U.S 4-44
Table 4-39. Total Fish Consumption for Recreational Saltwater Anglers, Recreational
Freshwater Anglers and General U.S. Fish Consuming Population 4-46
-------
Figures
Figure 4-1. Sources of U.S. Fish Consumption 4-2
Figure 4-2. Location of U.S. Aquacultural Operations, 1998 4-6
Figure 4-3. Recreational Fishing Participation Rates by State 4-7
Figure 4-4. 2002 Commercial Landings by Distance from Shore 4-9
Figure 4-5. 2002 Recreational Marine Finfish Landings by Distance from Shore 4-17
Figure 4-6. U.S. Commercial Fish Imports by Area, 2002 4-20
Figure 4-7. U.S. Commercial Fish Imports by Country, 2002 4-21
Figure 4-8. 2001 Total Recreational Fishing Days 4-29
Figure 4-9. 2001 Recreational Fishing Days, State Residents 4-29
Figure 4-10. 2001 Recreational Fishing Days, Non-State Residents 4-30
Figure 4-11. 2002 Market Share of Commercial Finfish (% of Total Economic Value) 4-37
Figure 4-12. 2002 Market Share of Commercial Shellfish (% of Total Economic Value) . . 4-38
Figure 4-13. Fish Consumption Pathways 4-40
-------
SECTION 4
PROFILE OF FISHING ACTIVITY IN THE U.S.
4.1 Industry Characterization
Because fish consumption is the primary pathway for exposure to methylmercury, this
section provides background information through a profile of the fishing industry in the United
States. We provide a characterization of the commercial and recreational fishing industries,
followed by information on U.S. commercial and recreational fish production, U.S. commercial
and recreational fishing demand, data on the pricing of affected fish products, and an industry
outlook. We also provide a characterization of populations that consume fish, and the potential
magnitude of quantities consumed in each category offish consumption pathways. Based on
data availability, most of the information in this profile is based on calendar year 2002.
However, in some sections, historic and projected industry data are provided.
This section demonstrates that the recreational freshwater fisher population (28 million)
is significantly larger than the recreational saltwater population (9 million), while both of these
populations are significantly smaller than the general population offish consumers in the U.S.
(184 million) which includes many individuals receiving a significant fraction of their fish from
commercially-produced stocks.
Based on information provided in this section, we see that commercial fish consumption
constitutes a large portion of exposure to methylmercury. However, a large majority of the
commercial fish consumed are imported from foreign sources, or 3-200 miles offshore by
domestic commericial fishermen (with a majority of domestic landings occuring off the Pacific
coast). These sources of exposure are not likely to be impacted by the control of utilities from
the CAMR rule. However, methylmercury concentrations from freshwater sources are likely to
be affected by control domestic electric utilities. Therefore, the quantified benefit analysis in
Section 11 evaluates the benefits of improved health from reduced exposure to methylmercury
from recreational freshwater fishing activities.
4.1.1 Introduction and Overview of the Fishing Industry
The most common mechanism of exposure to mercury for humans and wildlife is through
the consumption of mercury containted in predatory fish. These include saltwater fish such as
tuna, shark, and swordfish, which are most often caught commercially. They also include
freshwater fish such as bass, perch, and walleye, which are often caught recreationally.
The fish that Americans eat come from a variety of sources. Figure 4-1 shows that total
fish consumption is composed of commercial fish and shellfish, aquaculture (or farm raised fish
for commercial sale), fish caught from recreational activities, fish caught for cultural or
traditional practices, and imports from international waters. The types offish consumed vary
greatly and may come from saltwater or freshwater sources. The figure also shows that the
benefits analysis focuses on the consumption of recreationally-caught freshwater fish.
4-1
-------
Total Fish
Consumption
Recreational &
Self-Caught
JL
Domestic
Commercial
& Aquaculture
Commercial
Imports
ence 1—
': - }
: Fres
Salwater
Freshwater
Saltwater
Freshwater
Saltwater
Figure 4-1. Sources of U.S. Fish Consumption
The focus of this profile is a characterization of the commercial fishing industry and
recreational fishing activity. However, some information on industries associated with
commercial and recreational fishing is also included. The industrial categories with information
in this profile are provided below with their North American Industrial Classification System
(NAICS) code:
Description
Finfish Fishing
Shellfish Fishing
NAICS
114111
114112
Description
Seafood Canning
Fresh and Frozen Seafood
NAICS
311711
311712
Finfish Farming and Fish 112511
Hatcheries
Shellfish Farming 112512
Shipbuilding and Repair 336611
Boat Building 336612
Boat Dealers 441222
Processing
Fish and Seafood Wholesalers
422460
Fish and Seafood Retail Markets 445220
Seafood Restaurants 7221, 7222
Fishing Supply Stores 451110
Charter Fishing Boat Services 4872102
The rest of this section provides an industry characterization for the commercial and recreational
fishing industries. Section 4.1.2 covers the commercial fishing industry. Section 4.1.3 provides
information on the recreational fishing industry.
4.1.2 Commercial Fishing
The commercial fishing industry grossed over $3 billion in revenures in 2003 and is
dominated by saltwater species offish (i.e., tuna, shrimp, clams, shark). Over 60% of domestic
commercial saltwater fish are caught between 3 and 200 miles from shore (most of the remaining
4-2
-------
commercial saltwater catch is made within 3 miles from shore. Approximately, two-thirds of the
U.S. commercial landings occur along the Pacific coast. Imports of commercial seafood account
for 35% of U.S. demand for commercial fishery products.
4.1.2.1 Industry Characterization
The commercial fishing industry is classified under the standard industrial classification
system (SIC) as 0912 for fmfish firms and 0913 for shellfish firms, and the NAICS under codes
114111 and 114112, respectively. Although detailed industry gross domestic product (GDP)
data were not available at the five digit NAICS code level, the gross industrial output for NAICS
code 114100 was $3.19 billion in 2002 (BEA, 2004). Gross industrial output for the years 1998
through 2003 were:
Year Gross Output (Billion $)
1998 3.28
1999 3.56
2000 3.65
2001 3.39
2002 3.19
2003 3.45
The estimated value of year 2002 U.S. edible and nonedible fish products was $7.67
billion, which was 0.07% of U.S. GDP in 2002. Production and marketing for the commercial
fishing industry generated an estimated $28.4 billion in value added to the U.S. Gross National
Product. In addition to primary, secondary, and imports processing, value added activities
include retail trade from food service and retail trade from stores. Total consumer expenditures
for commercial fishing products were estimated to be $55.1 billion in 2002 (NMFS, 2003a).
The commercial fishing industry includes products from fmfish fishing, shellfish fishing,
and aquaculture. Each sub-industry is discussed further below. As expected, the majority of
U.S. commercial fish landings occur off the coast of the United States. Some commercial fishing
products are caught within three miles of the U.S. coast, while the majority of commercial
products are caught further from shore (i.e., between 3-200 miles off the U.S. coast). In fact,
most commercial fishing occurs in saltwater from the Pacific and Atlantic Oceans, and in the
Gulf of Mexico, with the largest portion of landings from the Pacific Ocean. Section 4.2
provides more detail on U.S. commercial and recreational fish landings.
4.1.2.1.1 Finfish Fishing
In 2002, there were 1,113 establishments with 4,301 employees for NAICS 114111,
fmfish fishing. Table 4-1 shows the number of firms, number of establishments, employment
and payroll for 1998-2002 from the Bureau of Census (BOC) publication County Business
Patterns and Statistics of U.S. Businesses (BOC, 2004a; 2004b). Statistics of U.S. Businesses
4-3
-------
from the Bureau of Census 2002 have not been released yet, so the number of firms for 2002 is
not available (BOC, 2004b). The data demonstrate that the finfish industry is dominated by one-
facility firms rather than firms that have multiple facilities nationwide and the average firm
employs approximately 3.5 people.
Table 4-1. Finfish Fishing Industry (NAICS code 114111)
Year
1998
1999
2000
2001
2002
Number of
Firms
1,269
1,307
1,350
1,293
NA
Number of
Establishments
1,275
1,312
1,357
1,302
1,113
Number of
Employees
4,630
5,088
4,833
4,586
4,301
Annual Payroll
($1,000)
181,872
191,521
190,500
183,490
190,657
Source: BOC, 2004a; BOC, 2004b.
4.1.2.1.2 Shellfish Fishing
The Census Bureau reports that there were 793 establishments with 2,195 employees in
2002 for NAICS 114112, shellfish fishing. Table 4-2 shows the number of firms, number of
establishments, employment and payroll for 1998-2002 from the County Business Patterns and
Statistics of U.S. Businesses from the Bureau of Census (BOC, 2004a, 2004b). Statistics of U.S.
Businesses from the Bureau of Census 2002 have not been released yet, so the number of firms
for 2002 is not available. Like the finfish industry, the shellfish industry is also dominated by
firms with only one establishment with average employment of 3 people per firm.
Table 4-2. Shellfish Fishing Industry (NAICS code 114112)
Year
1998
1999
2000
2001
2002
Number of
Firms
831
858
940
935
NA
Number of
Establishments
832
859
941
936
793
Number of
Employees
2,535
(G)
2,628
(H)
2,195
Annual Payroll
($1,000)
68,723
(D)
75,414
(D)
60,944
(D) — Withheld to avoid disclosing data for individual companies; data are included in broader industry totals
(A)-(C), (E)-(M) — Employment-size classes are indicated as follows:
A~0tol9 B-20to99 C--100to249 E-250to499
F--500 to 999 G-1,000 to 2,499 H--2,500 to 4,999
1-5,000 to 9,999 J--10,000 to 24,999
K-25,000 to 49,999 L-50,000 to 99,999
M-100,000 or more
Source: BOC, 2004a; BOC, 2004b.
4-4
-------
4.1.2.1.3 Aquaculture
Although firm-level statistics on aquaculture were not available from the BOC for
NAICS codes 112511 (fmfish) or 112512 (shellfish), the U.S. Department of Agriculture
(USDA), National Agricultural Statistics Service (NASS) reports that in 2002 aquacultural sales
totaled $1.13 billion with 6,653 farms (NASS, 2002). Detailed aquacultural data are available
from the 1998 Census of Aquaculture (NASS, 1998). According to USDA, there were 4,028
aquacultural farms in 1998 with 3,252 used for freshwater fish and 815 used for saltwater fish
farming. Total sales in 1998 were $978 million with food fish accounting for over 70% of sales.
These data indicate a large increase in the number of farms from 1998 to 2002, but a relatively
small increase in sales.
An estimated 68% of these farms were in the southern states of the U.S. Mississippi was
the top State in aquacultural sales, capturing nearly 30 percent of the $978 million dollar total in
1998. Arkansas, Florida, Maine, and Alabama came in second through fifth, respectively, in
sales. Food fish (e.g., catfish, trout, salmon, tilapia, hybrid striped bass, etc.) accounted for
about two-thirds of the aquacultural sales. Other aquaculture activities include raising fish to
stock waterbodies with particular species. Figure 4-2 provides a map of aquacultural farming
operations in the U.S. based on total freshwater and saltwater acreage from the U.S. Department
of Agriculture's 1998 Census of Aquaculture.
Table 4-3. Value of Aquacultural Products Sold, 1998
Product
Catfish
Trout
Food fish, other than catfish and trout
Baitfish
Ornamental fish
Sport or game fish
Other fish
Crustaceans
Mollusks
Other animal aquaculture, algae, and sea vegetables
Total
Farms
1,370
561
435
275
345
204
11
837
535
216
4,028
Sales ($1,000)
450,710
72,473
168,532
168,532
68,982
7,390
267
36,318
89,128
46,734
978,012
Source: NASS, 1998.
4-5
-------
Total Freshwater and Saltwater Acreage
Figure 4-2. Location of U.S. Aquacultural Operations, 1998
4.1.3 Recreational Fish ing
Recreational fishing is characterized by individuals fishing for sport/recreational
purposes, and/or for subsistence.
4.1.3.1 Industry Characterization
4.1.3.1.1 Information on Recreational Anglers in the United States
In 2001, 29.5 million fishing licenses were sold in the United States, while the number of
anglers (aged 16 and over) is estimated to be 34.1 million (USFWS, 2002). Of these 34.1
million anglers, 28.4 million participated in freshwater fishing and 9.1 million participated in
saltwater fishing. Table 4-4 shows the number of fishing licenses sold, the number of anglers,
and the participation rate (percent of population) for each region of the United States. Figure 4-3
shows the number of anglers per capita (aged 16 and over) by state. Data on recreational fishing
demographics are provided in Section 4.3.3.2 below.
Approximately 16% of the U.S. population participates in recreational fishing activities.
The states with the greatest numbers of anglers are California and Texas. However, when taking
population into account, these states have relatively low participation rates (9% and 15%,
respectively). As shown in Figure 4-3, the states with the highest levels of fishing participation
4-6
-------
are in the Northern Plains region: Minnesota (36%), Montana (32%), and Wyoming (32%) and
Alaska (41%).
Table 4-4. Number of Anglers and Fishing Licenses in the United States
Region
United States Total
Great Lakes
Northeast
Northern Plains
South Central
Southeast
West
Fishing Licenses
(thousands)
29,452
5,395
4,578
4,355
4,556
5,463
5,106
Number of
Anglers*
(thousands)
34,071
6,182
6,433
3,722
5,572
7,192
4,947
Population*
(thousands)
212,298
36,294
51,975
13,858
28,880
40,241
40,624
Participation
Rate
16%
17%
12%
27%
19%
18%
12%
*Ages 16 and over only
% of Population That Fish
Figure 4-3. Recreational Fishing Participation Rates by State
4-7
-------
4.2 U.S. Production Statistics for Commercial and Recreational Fishing
Data are provided in this section on production estimates for the commercial and
recreational fishing sectors. Primarily, these data are estimates of the amount offish landed by
either commercial or recreational anglers. Available data are often aggregated as either finfish
or shellfish or by groups of individual fish species. Production data for recreational freshwater
fishing are lacking. More detailed production data are provided in the report titled "Profile of
the Fishing Industry in the United States" available in the docket for this rulemaking (Pechan,
2005).
4.2.1 Commercial Fishing
This section summarizes available U.S. commercial production information (fish
landings), including landings of finfish or shellfish, fish types, locations offish landings, edible
and non-edible uses offish products, and seasonal production information. These data were
taken from the National Marine Fisheries Service (NMFS) of the National Oceanic and
Atmospheric Administration (NOAA).
The NMFS provides both monthly and annual summaries of commercial fishery landings
that are updated weekly. Domestic fishery landings are those fish and shellfish that are landed
and sold in the 50 states by U.S. fishermen and do not include landings made in U.S. territories
or by foreign fishermen. Landings are provided in round (whole) weight even though many fish
may be processed while at sea. Landings do not include aquaculture products except for clams,
mussels and oysters.
4.2.1.1 Domestic Commercial Fishery Landings
Table 4-5 provides a summary of total domestic commercial landings of finfish and
shellfish for the years 1993 - 2003 (landed by U.S. fishermen and sold in the U.S.). Although
production data for 2003 are available, the year 2002 was selected for use in this profile, since
the NMFS indicates that data for 2003 may be incomplete (NMFS, 2004). Pechan (2005)
provides detailed 2002 landings data by fish species for finfish and shellfish, respectively.
Table 4-5. Annual Domestic Landings for Commercial Fishing (Finfish and Shellfish)
Year
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
Metric Tons
4,630,853
4,762,637
4,496,420
4,374,409
4,514,152
4,233,291
4,267,982
4,147,069
4,314,492
4,270,030
4,311,663
Source: NMFS, 2004.
4-8
-------
NMFS categorizes commercial landings by distance to shore and identifies these
distances as inland, state territorial sea, and federal exclusive economic zone with Florida and
Puerto Rico having unique location identifiers. Inland landings refer to commercial landings in
bays, estuaries, and sounds within the United States. State territorial sea represents the area that
extends out three nautical miles from shore for all states with the exception of Florida and Puerto
Rico whose territorial sea extends to ten nautical miles. Federal exclusive economic zone refers
to the ocean area that extends to 200 nautical miles beyond the state territorial sea boundary
(NMFS, 2003b).
Figure 4-4 is a chart showing the amount of commercial fish landed in inland waters and
state and territorial seas (0-3 miles from shore) to be 36% of the total, while 61% of
commercial fish were landed in the Federal exclusive economic zone (3 - 200 miles offshore).
Only 3% of domestic commercial fish landings occurred in high seas locations or off foreign
shores. Pechan (2005) provides additional details on the commercial fish species caught by
distance from shore.
61%
36%
• 0- 3 Miles
• 3 - 200 Miles
nHigh Seas/Foreign Shores
Source: NMFS, 2004.
Figure 4-4. 2002 Commercial Landings by Distance from Shore
A review of the commercial fish landings by region demonstrated that the Pacific coast
accounted for 65% of total landings (see Table 4-6) with Alaska and California accounting for
the largest portion. Alaska alone accounted for over 26% of total U.S. landings. The next
highest landing region in percent, the Gulf Coast (18%), had significantly less landings than the
Pacific Coast. Among the Gulf Coast landings, Louisiana was the state with the highest landing
totals. New England and the Chesapeake Bay contributed 6% and 5% to U.S. landings,
respectively, with total commercial landings in Virginia and Massachusetts (686 million pounds)
having the highest landing numbers among their groups. The Mid-Atlantic, South Atlantic, and
the Great Lakes regions combined for the remaining 4.5% of total commercial landings. Within
the United States, the ports registering the highest number offish landings in 2002 were Dutch
Harbor-Unalaska, AK, Empire-Venice, LA, and Reedville, VA.
4-9
-------
Table 4-6. Commercial Fish Landings by Region
Region
New England
Middle Atlantic
Chesapeake Bay
South Atlantic
Gulf Coast
Pacific Coast
Great Lakes
TOTAL
Metric Tons
264,859
93,756
224,834
97,431
778,428
2,784,260
8,096
4,251,664
Percent
6.2
2.2
5.3
2.3
18.3
65.5
0.2
100
Source: NMFS, 2004.
Fish landings can be disaggregated into edible fish (fish used for human consumption)
and non-edible products (often used for animal consumption, fish meal, fish oil, or bait). Fish
landings ready to be used for human food consumption are generally either fresh or frozen, while
fish that are canned, cured, or processed for both human and non-human consumption are
categorized as landings for industrial purposes.
Of the 4.3 million tons landed by U.S. fishermen in 2002, an estimated 3.4 million metric
tons were to be used for human food while the remaining 0.9 million metric tons offish landings
were designated to be used for other purposes (fish meal, oil, or other including bait and animal
food). These data are shown in Table 4-7 below. Among the canned and cured products, about
320,000 metric tons were to be used for human food. About 4% of total landings were used for
bait and other animal consumption. About 19% of landings were used for reduction to fish meal,
oil, or other fish products. The amount of this last end use destined for human consumption (i.e.
fish oil) was not available.
Table 4-7. 2002 U.S. Commercial Fish Landings by End Use
End Use
Fresh and Frozen:
For human food
For bait and animal food
Subtotal
Canned:
For human food
For bait and animal food
Subtotal
Cured:
Cured for human food
Reduction to meal, oil, other
Subtotal
Total
Metric Tons
2,940,000
150,000
3,090,000
260,053
24,926
284,979
53,024
816,666
869,690
4,244,669
Percent
69.3
3.5
72.8
6.1
0.6
6.7
1.2
19.2
20.4
100
Source: NMFS, 2003b.
Totals do not add exactly or match between tables due to rounding.
In 2002, almost one-third of all commercial landings for human consumption occurred
during the months of July and August (see Table 4-8). An additional 25% of total commercial
.4-10
-------
landings occurred during the months of February and March while roughly 11% offish were
landed during September.
Table 4-8. 2002 U.S. Commercial Landings by Month
Month
January
February
March
April
May
June
July
August
September
October
November
December
Total
Landings for Human Food
Metric Tons
149,234
432,734
400,982
107,050
154,224
216,367
484,445
597,845
358,344
218,635
90,720
57,607
3,268,188
Percent
4.6
13.2
12.3
3.3
4.7
6.6
14.8
18.3
11.0
6.7
2.8
1.8
100
Landings for Industrial
Purposes
Metric Tons
24,494
20,412
16,783
54,432
90,266
136,534
203,213
168,739
117,936
107,503
28,123
25,855
994,291
Percent
2.5
2.1
1.7
5.5
9.1
13.7
20.4
17.0
11.9
10.8
2.8
2.6
100
Total
Metric Tons
173,729
453,146
417,766
161,482
244,490
352,901
687,658
766,584
476,280
326,138
118,843
83,462
4,262,479
Percent
4.1
10.6
9.8
3.8
5.7
8.3
16.1
18.0
11.2
7.7
2.8
2.0
100
Totals do not match between tables due to rounding.
Source: NMFS, 2003b.
Table 4-9 provides a summary of the species of finfish with the highest landings by U.S.
commercial fishermen during 2002 (tuna were added for comparison; most tuna consumed in the
U.S. are imported). Pechan (2005) provides additional detail in 2002 finfish landings by species.
Pollock and Menhaden were by far the finfish types with the highest commercial landings by
U.S. craft in 2002. These two finfish types are almost exclusively caught by commercial
fishermen. Pollock are often used in making frozen fish products (e.g., fish sticks), surimi (see
Section 4.2.1.3), and other minced fish products. Menhaden are often used to produce cut or live
bait, fishmeal and fish oil.
Table 4-9. Finfish Species with the Highest 2002 U.S. Commercial Landings
Finfish Type
Pollock
Menhaden
Salmon
Cod
Hake
Sole
Herring
Sardine
Tuna*
Other Finfish
Total 2002 Finfish
Metric Tons
1,519,101
793,486
254,613
245,705
141,642
116,779
99,173
97,527
22,513
377,904
3,668,443
% of Total Finfish Landed
41.4
21.6
6.94
6.70
3.86
3.18
2.70
2.66
0.61
10.3
100
Source: NMFS, 2004.
Note: These estimates are for domestic landings and do not include quantities of imports.
* There are a number of species that rank higher than tuna, but are not displaed in this table.
Tuna is added for comparison purposes only to other species.
4-11
-------
Table 4-10 provides a summary of the shellfish types with the highest commercial
harvests in 2002 by the domestic fleet. Nearly half of the 2002 landings were for shrimp and
crab. Of the total finfish and shellfish landings in 2002, shellfish represent about 14% of the
total. Details on the commercial landings of shellfish in 2002 are provided in a separate report
(Pechan, 2005).
Table 4-10. Shellfish Types with the Highest 2002 Commercial Landings
Shellfish Type
Shrimp
Crab
Squid
Clam
Lobster
Scallop
Oyster
Other Shellfish"
Total 2002 Shellfish
Metric Tons
156,059
139,840
93,217
59,059
39,454
24,047
16,701
73,238
601,615
% of Total Shellfish Landed
25.9
23.2
15.5
9.82
6.56
4.00
2.78
12.2
100
Source: NMFS, 2004.
* NMFS includes some non-shellfish species in the "Other shellfish" group including seaweed
(7.8%), sponges (0.05%) and turtles (0.006%).
4.2.1.2 Aquaculture
Total aquacultural production in 2002 was 393 thousand metric tons with catfish
accounting for 73% or 286 thousand metric tons (NMFS, 2004). Aquaculture provides most of
the commercial catfish production in the U.S. Data for both finfish and shellfish are provided in
Table 4-11. Trout production levels were 6% of the total, while salmon accounted for 3%. An
additional 7% of aquacultural production in 2002 was attributed to raising crawfish.
4.2.1.3 Exports of Finfish and Shellfish
Total exports include both "exports" (exports of fishery products of domestic origin) and
"re-exports" (exports of fishery products of foreign origin). Total edible finfish and shellfish
exports for 2002 were about 970,000 metric tons (product weight), with finfish accounting for
about 85% of this amount (NMFS, 2003b). Table 4-12 provides 2002 summary export data for
finfish. These export data include products of both domestic and foreign origin. The largest
export of a particular finfish product was surimi, which made up 23% of the exported finfish
products. Surimi is a Japanese word meaning "minced fish". It is typically produced from
skinless Alaskan pollack and used in the subsequent manufacture of imitation fish products (e.g.,
imitation crabmeat). Over 91% of the finfish products exported were either fresh or frozen.
(Pechan, 2005) provides additional details of fresh and frozen fish exports in 2002.
Table 4-13 provides 2002 export data for shellfish. Among U.S. 2002 shellfish exports,
the largest export products were as follows: squid (39%), lobster (20%), crabs (10%), and shrimp
(10%). Nearly 89% of shellfish products were exported either fresh or frozen. Pechan (2005)
provides more detailed information on shellfish exports.
4-12
-------
Table 4-11. Total 2002 U.S. Aquacultural Production
Fish Species
Metric Tons
Percent
Finfish:
Baitfish"
Catfish
Salmon0
Striped bass"
Tilapia
Trout
Finfish Subtotal
6,329
286,039
12,734
4,758
9,000
24,699
343,559
1.6
72.7
3.2
1.2
2.3
6.2
87.3
Shellfish:
Clams"
Crawfish
Mussels8
Oysters"
Shrimp8
Miscellaneous8
Shellfish Subtotal
Total
4,473
27,825
627
8,413
4,080
4,425
49,843
393,402
1.1
7.1
0.2
2.1
1.0
1.1
12.7
100
Source: NMFS, 2003b.
' Saltwater species. The baitfish and and all shellfish species (except crawfish) are believed to be
primarily saltwater species, although definitive data were not available from the source material.
4-13
-------
Table 4-12. 2002 Exports of Edible Finfish Products
Product
Fresh or Frozen Finfish:
Freshwater (whether or not whole):
Eels
Tilapia
Trout
Total Freshwater (whether or not whole)
Saltwater (whether or not whole):
Flatfish8
Groundfish"
Salmon
Tuna
Other0
Total Saltwater Finfish (whether or not whole)
Freshwater (fillets and steaks):
Catfish
Tilapia
Total Freshwater (fillets and steaks)
Saltwater (fillets and steaks):
Flatfish3
Groundfish"
Other
Total Saltwater (fillets and steaks)
Blocks, Regular And Minced
Surimi
Fish Sticks And Similar Products
Total Finfish: Fresh And Frozen
Canned Finfish
Anchovy
Herring
Sardine
Mackerel
Salmon
Tuna
Total Finfish: Canned
Cured Finfish
Dried
Pickled or Salted
Smoked or Kippered
Total Finfish: Cured
Total Finfish Exports'1
Metric Tons
3,063
2,331
600
5,994
67,519
103,900
78,539
15,302
159,613
424,873
73
2,065
2,138
760
80,514
16,397
97,671
26,372
190,911
21,332
769,291
333
3,313
16,190
1,049
44,708
1,628
67.221
843
4,554
503
5,900
842,412
% of Total
0.36
0.28
0.07
0.73
8.01
12.3
9.32
1.82
18.9
51.5
0.01
0.25
0.06
0.10
9.56
1.95
11.4
2.93
23.1
2.55
91.3
0.04
0.39
1.92
0.12
5.31
0.19
7.98
0.10
0.54
0.06
0.70
100
Source: NMFS, 2003b. Includes exports of edible products of both domestic and foreign origin.
' Flatfish includes halibut.
b Groundfish include cod and pollock.
c Other include atka mackerel, butterfish, herring, lingcod, mackerel, monkfish, mullet, sablefish, sardine, scorpionfish,
sea bass, shark and dogfish, toothfish, and unclassified.
d Excludes other fish products such as fish balls, cakes, and puddings, fish/shellfish juice, soups and broths, caviar and
roe, prepared fish meals, and other.
4-14
-------
Table 4-13. 2002 Exports of Edible Shellfish Products
Product
Metric Tons
% of Total
Fresh and Frozen Shellfish Products
Crabs and Crabmeat
Crawfish (freshwater)
Lobsters and Lobster Meat
Shrimp
Conch
Clams
Cuttlefish
Mussels
Oysters
Scallops
Octopus
Squid
Snails
Sea Urchins
Unclassified
Frog Legs and Meat3
Total Fresh and Frozen Shellfish Products
16,089
199
30,748
15,060
444
834
145
645
1,788
4,589
528
60,151
36
1,505
5,472
79
138,312
10.3
0.13
19.7
9.67
0.29
0.54
0.09
0.41
1.15
2.95
0.34
38.6
0.02
0.97
3.51
0.05
88.8
Canned Shellfish Products
Crabmeat
Lobsters and Lobster Meat
Shrimp
Clams
Squid
Total Canned Shellfish Products
Total Shellfish
538
31
1,507
1,762
13,575
17,413
155,725
0.35
0.02
0.97
1.13
8.72
11.2
100
Source: NMFS,2003b.
" NMFS includes frog legs and meat in the shellfish category.
4.2.2 Recreational Fishing
Data on recreational freshwater fish harvest are not available for this profile. Information
on the freshwater species most often targeted by recreational anglers are provided later in
Section 4.3.3.1 (see Table 4-24).
NMFS provided data on estimated recreational marine (saltwater) finfish landings based
on the Marine Recreational Fishing Statistical Survey (MRFSS; NMFS, 2003b). Similar data for
recreational shellfish landings are not available. The purpose of the MRFSS is to establish a
reliable data base for estimating the impact of marine recreational fishing on marine resources.
4-15
-------
This annual survey has been conducted since 1979. The MRFSS covers all coastal states except
Texas and Alaska. The survey provides coverage of saltwater sport fishing (including estuarine
and brackish water) from private/rental boats, charter and head boats, and the shore on the
Atlantic Coast (Maine-East Florida), Gulf Coast (Louisiana-West Florida), and Pacific coast
(Washington through California).
A summary of recreational marine finfish landed for 2002 is provided in Table 4-14
below. This summary excludes the fish that were caught and subsequently released alive. More
detailed data are provided in a separate report (Pechan, 2005). It should be noted that
recreational marine finfish landings in 2002 were less than 20% of the average annual landings
estimated by NMFS from 1999 - 2003 (Pechan, 2005). NMFS indicates that recreational boat
fishing trips were not included for 2002 for Washington and Oregon, but other reasons for the
large discrepancy were not identified.
Table 4-14. 2002 Recreational Marine Landings for Selected Finfish Types
Finfish Type
Dolphinfish
Drums
Flounders
Pacific Barracuda
Sea Basses
Temperate Basses
Tunas and Mackerels
Other Finfish
Total
2002
Landings
(Metric Tons)
6,712
20,366
5,997
9,292
6,896
8,903
14,103
31,262
103,531
%of
Total
6.5
20
5.8
9.0
6.7
8.6
14
30
100
1999-2003 Average
Landings
(Metric Tons)
35,826
116,503
37,533
3,393
29,866
44,561
90,921
200,714
559,317
Figure 4-5 shows that roughly 32% of the 103,529 metric tons of marine finfish landed in
U.S. waters in 2002 occurred in the Federal Exclusive Economic Zone (distances of 3-200 miles
from shore) while 29% of total finfish landings were within 3 miles from shore (State Territorial
Sea). The remaining 39% of finfish were landed in Inland Waters. For shellfish, 65% of the
534,608 metric tons landed were made within Inland Waters and State and Territorial Sea
(< 3 miles from shore). A separate report provides more details on the types of marine
recreational fish caught by distance from shore (Pechan, 2005).
4-16
-------
39%
29%
32%
State Territorial Sea
Federal Exclusive
Economic Zone
Inland Waters
Source: NMFS,2004.
Figure 4-5. 2002 Recreational Marine Finfish Landings by Distance from Shore
Table 4-15 provides 2002 recreational marine catch data by region and top harvested
species group. This summary data was compiled based on the MRFSS and summarized by U.S.
region. The regions were constructed by the American Sportfishing Association (ASA). The
data are summarized this way to be consistent with recreational fishing data provided later in this
report.
The northeast region includes the coastal New England States, Delaware, District of
Columbia, Maryland, New Jersey, New York, and Virginia. The south central region includes
data only for Louisiana, since data for Texas were not included in the MRFSS. The southeast
region includes Alabama, Florida, Georgia, Mississippi, North Carolina, and South Carolina.
The west region includes California, Oregon, Washington, and Hawaii (Alaska was not included
in the MRFSS). Harvested weight was not available for all state-species groups. Therefore,
average weights per fish for each region were estimated using the available data. These average
weights per fish were used to estimate harvested weight where unavailable.
The greatest amount of recreational marine catch was harvested in the Southeast region
followed by the Northeast. Harvest values for the West region would likely be much higher if
the harvest for Alaska were included, and values in the south central region would be higher if
data for Texas were included in the MRFSS.
4-17
-------
Table 4-15. 2002 Recreational Marine Catch and Harvest by Region and Top Species
Group
Species Group
Northeast
Total for Region
Drums
Tunas and Mackerels
Bluefish
Porgies
Flounders
South Central'
Total for Region
Drums
Porgies
Flounders
Catfishes
Snappers
Southeast
Total for Region
Herrings
Drums
Porgies
Mullets
Jacks
West'
Total for Region
Rockfishes
Smelts
Flounders
Sea Basses
Herrings
# of Fish
Caught
128,019
28,675
4,356
10,411
7,641
17,326
24,565
18,724
2
320
3,123
190
225,303
53,400
42,101
22,793
10,240
13,796
42,015
5,435
4,186
5,464
6,931
2,399
# of Fish
Harvested3
41,934
14,361
3,821
3,684
3,661
3,550
10,457
8,953
652
272
194
130
111,486
46,693
15,566
10,448
8,442
6,839
24,067
4,270
4,174
3,875
2,403
2,216
Harvested
Weight"
(metric tons)
37,013
5,568
3,711
4,326
1,682
3,802
9,974
8,143
679
144
127
307
40,882
211
6,671
2,593
1,103
2,813
15,944
2,797
142
1,378
1,538
151
"Harvest equals catch minus fish released alive.
""Weight estimated for some state/species groups.
'Does not include Alaska (west) and Texas (south central).
4-18
-------
4.3 U.S. Demand for Commercial and Recreational Fishing
This section presents information on the U.S. demand for commercial and recreational
fishing and related industries.
4.3.1 Commercial Imports
This section provides information on commercial imports of finfish and shellfish
products. Included are data on the type of products and supplying countries. These products
include those produced through commercial catches, as well as foreign aquaculture. Total edible
imports for 2002 were 4.0 million metric tons with finfish contributing 65% of the total. A
detailed breakdown, by product weight, of edible fish imports is provided in a separate report
(Pechan, 2005). The following sections provide a summary of edible finfish and shellfish
imports. In comparison to domestic production (net of exports), commercial imports represent a
large component of total U.S. demand for fish products. It should be noted that the CAMR is not
likely to substantially affect the level of methylmercury in fish that is imported to the U.S.,
because of the small contribution of U.S. emissions to the global pool that impacts MeHg in
imported fish.
Figures 4-6 and 4-7 provide a breakdown of commercial fishery products by region and
country, respectively (NMFS, 2003b). An estimated 46% of US imports were from Asia while
26% were from other regions in North America. South America could be credited with landing
17% of U.S. imports while Europe, Africa, and Oceania combined to total 11% (Figure 4-6).
Although the largest portions of U.S. fish imports were from Asian landings, Canada represents
the largest share at 18% when compared on a per country basis (Figure 4-7).
4-19
-------
Oceania
Europe 4o/o
6%
South America
17%
North America
26%
Figure 4-6. U.S. Commercial Fish Imports by Area, 2002
4-20
-------
Other | • Thailand
49% .'«4H 13%
China
10%
Chile
Ecuador go/
4%
Source: NMFS, 2003b.
Figure 4-7. U.S. Commercial Fish Imports by Country, 2002
4.3.1.1 Finflsh
Table 4-16 provides a breakdown of edible fresh and frozen finfish products imported to
the United States in 2002 (see detailed breakdown in Pechan (2005)). Fresh and frozen finfish
are categorized into the following products: whether or not whole; regular blocks; minced
blocks; fillets and steaks; surimi; and fish sticks and similar products. The total imported fresh
or frozen finfish products in 2002 was almost 1 million metric tons. Of the freshwater species
imported, tilapia was the most common and accounted for over 6% of the total finfish product
imports. Tuna was the largest group of saltwater fish species imported at nearly 17% of the fresh
or frozen imports.
Table 4-17 provides information on imported canned and cured finfish products imported
in 2002 (see details in Pechan, 2005). In 2002, about 288,000 metric tons of canned and cured
finfish products were imported into the United States for consumption. Nearly 60% of this
amount was canned tuna.
4-21
-------
Table 4-16. 2002 Imports of Edible Fresh or Frozen Finfish Products
Finfish Product
Metric Tons
% of Total Finfish
by Weight
Whether or Not Whole
Freshwater:
Tilapia
Unclassified
Total Freshwater Fish
Saltwater:
Flatfish8
Groundfishb
Salmon
Tuna
Other
Total Saltwater Fish
Total Whether Or Not Whole
40,748
17,642
58,390
21,111
24,615
82,665
162,252
116,840
407,483
465,873
4.19
1.81
6.00
2.17
2.53
8.50
16.68
12.01
41.9
47.9
Blocks, Regular
Total Freshwater Fish
Saltwater:
Flatfish3
Groundfishb
Other
Total Saltwater Fish
Total Regular Blocks
127
1,460
50,572
2,494
54,256
54,653
0.01
0.15
5.20
0.26
5.61
5.62
Blocks, Minced
Total Freshwater Fish
Saltwater:
Flatfish1
Groundfishb
Other
Total Saltwater Fish
Total Minced Blocks
1
13
1,934
10,091
12,038
12,039
0.00
0.00
0.20
1.04
1.24
1.24
Fillets And Steaks
Freshwater:
Tilapia
Other
Total Freshwater Fillets
Saltwater:
Flatfish3
Groundfishb
Other
Total Saltwater Fillets
Total Fillets And Steaks
Surimi
Fish Sticks And Similar Products
Total Fresh And Frozen Finfish
26,440
24,134
50,574
23,369
104,985
239,535
367,889
418,463
3,559
18,271
972,858
2.72
2.48
5.20
2.40
10.8
24.6
37.8
43.0
0.37
1.88
100
Source: NMFS, 2003b.
" Flatfish include halibut, Greenland turbot, plaice, sole, and other.
b Groundfish include cod, haddock, hake, Pollock and ocean perch.
4-22
-------
Table 4-17. 2002 Imports of Edible Canned and Cured Finfish Products
4.3.1.2 Shellfish
Finfish Product
Metric Tons
Canned
Anchovy
Herring
Mackerel
Salmon
Sardines
Tuna
Balls, Cakes, and Puddings
Other (includes Finfish and Shellfish)
Total, Canned
3,298
3,814
9,928
4,542
22,220
171,523
9,014
28,880
253,219
Cured
Dried
Pickled or Salted
Smoked or Kippered
Total, Cured
Total Canned and Cured Imports
7,468
20,952
6,498
34,918
288,137
Source: NMFS, 2003b.
Table 4-18 provides data on 2002 edible fresh and frozen shellfish imported into the
United States. Total shellfish imports were about 692,000 metric tons in 2002. By far, shrimp
was the largest fresh and frozen shellfish product imported (over 60% of total shellfish). Crab
and crabmeat was the next highest product imported at over 11% of total shellfish. Pechan
(2005) contains details of the fresh and frozen shellfish products imported in 2002.
Table 4-19 provides data on 2002 edible canned shellfish product imports (for this table,
it was assumed that all cured edible fish products are finfish products and they were placed in
Table 4-17 above). Over 33,000 metric tons of canned shellfish products were imported for
consumption in 2002. About 60% of this amount was canned crabmeat.
4-23
-------
Table 4-18. 2002 Imports of Edible Fresh and Frozen Shellfish Products
Shellfish Product
Total Crab and Crabmeat
Total Lobster and Lobster Meat
Shrimp
Clam
Mussels
Scallops
Octopus
Squid
Unclassified or Other8
Total Shellfish
Metric Tons
79,805
50,586
427,454
8,800
20,727
21,868
14,164
46,759
21,774
691,937
% of Total Shellfish
11.5
7.30
61.8
1.27
3.00
3.16
2.05
6.76
3.15
100
Source: NMFS, 2003b.
a Other includes Abalone, krill, crawfish, conch, cuttlefish, oysters, snails, sea urchins, and
frog legs and meat.
Table 4-19. 2002 Imports of Edible Canned Shellfish Products
Shellfish Product
Clams
Crabmeat
Lobsters
Oysters
Shrimp
Total
Metric Tons
5,330
20,545
47
5,825
1,849
33,596
% of Total Shellfish
15.9
61.2
0.14
17.3
5.50
100
Source: NMFS, 2003b.
4.3.2 U.S. Demand for Commercial Fishery Products
Table 4-20 provides a summary of commercial finfish demand per capita for several
important finfish types. These estimates were derived from the production data provided in
Section 4.2, the import data summarized above, and a 2002 U.S. population estimate of 288
million (BOC, 2005). These estimates should not be considered per capita consumption estimates
for several reasons. First, not all fish species landed commercially are intended for human
consumption (about 23% are for industrial purposes). Most of the non-edible landings are for
menhaden (about 80%). This is noted at the bottom of Table 4-20. In addition, existing U.S.
inventory of fishery products are not included in these estimates. Conversions have not been
made to estimate the edible portion of each type of edible finfish or shellfish. Finally, landings
are provided in units of live weight, whereas product import and export data are provided as
product weights. Table 4-21 provides similar estimates of per capita demand for shellfish
products.
4-24
-------
Table 4-20. 2002 U.S. Demand for Commercial Finfish (metric tons)
Finfish Type
Pollock
Menhaden
Salmon
Cod
Hake
Sole
Herring
Sardine
Tuna
Other Finfish
Total Finfish
2002
Production*
1,519,101
793,486
275,382
245,705
141,642
1 16,779
99,173
97,527
22,513
693,821
4,005,129
2002 Exports
331,976b
0
133,380
103,934
8,886
48,117
18,565
61,288
16,930
168,460
891,536
2002 Imports
80,889
0
210,735
78,777
10,622
12,353
7,846
24,794
333,775
435,722
1,195,513
2002 U.S.
Demand
1,268,014
793,486
352,737
220,548
143,378
81,015
88,454
61,033
339,358
961,083
4,309,106
Demand per
capita (Ib)
9.71
6.08
2.70
1.69
1.10
0.62
0.68
0.47
2.60
7.36
32.99
Source: NMFS, 2004.
° Production = Commercial Landings + Aquacultural Production.
b Assumes that surimi and regular/minced fish blocks are all made up of pollock.
Table 4-21. 2002 U.S. Demand for Commercial Shellfish (metric tons)
Shellfish Type
Shrimp
Crab
Squid
Clam
Lobster
Scallop
Oyster
Other Shellfish
Total Shellfish
2002
Production"
159,666
139,840
93,217
63,584
39,454
24,047
24,330
92,260
636^98
2002 Exports
16,567
16,627
73,726
2,596
30,779
4,589
1,788
15,289
161,961
2002 Imports
429,303
85,135
46,759
14,130
50,633
21,868
5,825
56,665
710,318
2002 U.S.
Demand
572,402
208,348
66,250
75,118
59,308
41,326
28,367
133,636
1,184,755
Demand per
capita (Ib)
4.38
1.60
0.51
0.58
0.45
0.32
0.22
1.02
9.07
Source: NMFS, 2004.
" Production = Commercial Landings + Aquacultural Production.
The data in Table 4-20 show that most of the U.S. tuna demand is met through imports. In
2002, less than 7% of the tuna demand is met through commercial landings by the domestic fleet.
Total 2002 per capita demand for commercial fishery products is estimated at 42.1
Ib/person. This estimate can be compared to an NMFS estimate of "per capita use" of 66.0
Ib/person in 2002, which includes both edible and industrial uses but does not include exports,
beginning/ending year inventory, or defense purchases (NMFS, 2003b).
NMFS developed estimates of 2002 U.S. per capita consumption for edible commercial
fishery products (NMFS, 2003b). To do this, a "disappearance model" was developed to account
for the edible portion of commercial landings and imports, exports of edible products, and
inventories of edible products. NMFS caveats their estimates by noting that the data sources for
the model are not always reported completely, and that incorrect model assumptions can lead to
significant changes in estimated consumption. NMFS estimated a total 2002 U.S. per capita
4-25
-------
consumption of 15.6 Ib edible meat/person. Additional consumption data are provided in
Table 4-22 below.
Table 4-22. 2002 U.S. Consumption of Commercial Fishery Products
Commercial Product
Fresh and Frozen Finfish and Shellfish
Canned Finfish and Shellfish
Cured Finfish and Shellfish
Total
2002 Consumption
(Ib/person)
11.0
4.3
0.3
15.6
Consumption of Specific Products
Canned:
Salmon
Sardines
Tuna
Shellfish
Other
0.5
0.1
3.1
0.3
0.3
Fresh and Frozen:
Fillets and Steaks
Sticks and Portions
Shrimp (all preparations)
4.1
0.8
3.7
Source: NMFS, 2003b.
4.3.3 Recreational Fishing
4.3.3.1 U.S. Consumption of Recreationally-Caught Fish
Data on recreational saltwater fishing catch is available from the Marine Recreational
Fishing Statistics Survey (MRFSS) from the NOAA Fisheries Statistics Division. These data
were provided in Section 4.2.2. This survey provides data on catch (all fish caught) and harvest
(all fish not released alive).
Data for freshwater recreational harvest are not available; however, the number of anglers
and number of days of fishing is available from the 2001 National Survey of Fishing, Hunting,
and Wildlife-Associated Recreation (USFWS, 2002). This survey is a partnership effort between
the USFWS, States and national conservation organizations. The purpose of the survey is to
quantify the economic impact of wildlife-based recreation. The 2001 survey was the tenth in a
series that began in 1955. Information from a total of 25,070 sportspersons (anglers and hunters)
were gathered for the 2001 survey. Data from the survey were summarized by the American
Sportfishing Association (ASA, 2003). Table 4-23 shows the number of anglers for the top
freshwater, Great Lakes, and saltwater species.
4-26
-------
Table 4-23. Number of Anglers and Days of Fishing for 2001
Type of Fish Targeted
Anglers
(millions)
Days of Fishing
(millions)
Freshwater except Great Lakes
Black bass
Panfish
Trout
Catfish/bullhead
Crappie
White bass, striped bass, and striped bass
hybrids
Total Freshwater except Great Lakes
10.7
7.9
7.8
7.5
6.7
4.9
45.5
160
103
83
104
95
62
607
Great Lakes
Perch
Walleye, sauger
Black bass
Salmon
Lake trout
Steelhead
Total Great Lakes
0.7
0.6
0.6
0.5
0.3
0.3
3.0
7
6
6
4
4
4
31
Saltwater
Flatfish (flounder, halibut)
Striped bass
Sea trout
Bluefish
Salmon
Mackerel
Total Saltwater
Total - All Fish
2.3
1.7
1.5
1.1
0.7
0.6
7.9
56.4
21
17
17
12
5
6
78
716
Average #
Days/Angler
15.C
13.(
10.6
13.9
14.2
12.7
13.3
10.C
10.C
10.C
8.C
13.3
13.3
10.3
9.1
10.C
11.3
10.9
7.1
10.C
58.4
12.7
Source: USFWS, 2002.
Table 4-24 shows the most targeted freshwater species, measured by angler participation,
for each region of the country. Bass (large and small mouth), pan fish, trout, catfish, and crappie
were the most popular target species for freshwater anglers nationwide. The region with the
largest number of anglers for all targeted species was the southeast followed by the south central
United States. In the western part of the United States, trout were by far the most targeted
freshwater species.
4-27
-------
Table 4-24. 2001 U.S. Recreational Freshwater Fishing: Targeted Species by Region
Targeted Species
Crappie
Panfish
Bass (white, striped)
Bass (large &
smallmouth)
Catfish
Walleye
Sauger
Pike
Trout
Salmon
Steelhead
Other
Anything
Total"
Number of Anglers (thousands)
West
186
230
526
873
612
36
a
2,645
932
370
644
392
4,127
South
East
2,027
1,907
1,387
2,897
2,275
108
a
a
583
a
a
1,176
1,413
6,107
South
Central
1,822
1,177
1,217
2,320
2,330
99
a
a
749
31
a
409
856
5,208
North
East
477
876
718
2,139
651
333
a
336
1,926
189
59
440
963
4,587
Great
Lakes
1,478
2,893
904
2,144
1,295
1,142
63
763
458
99
96
296
812
5,388
Northern
Plains
880
1,062
327
796
517
1,601
63
914
1,768
111
a
211
445
4,344
All
Regions
6,648
7,894
4,925
10,694
7,494
3,214
174
2,060
7,797
1,368
536
3,176
4,689
27,913
" Sample size is too small to report results with any reliability, (N = 30-39); or the species is not present in this
region.
bAngler numbers reported here are the number of anglers who fished in the given region. Summing across the
regions to derive a U.S. total will result in an overestimate as some anglers fished in more than one region. The
All Regions totals reported in this table eliminates all double-counting and reports the actual number of anglers in
the United States.
A separate report provides information on the number of days spent fishing by anglers in
each State (Pechan, 2005). Figures 4-8 through 4-10 are maps showing the total number of days
spent fishing in 2001 by all anglers, resident anglers, and non-resident anglers, respectively. High
levels of recreational fishing are shown to occur in the Great Lakes States, Florida, New York,
Texas, and California. Pechan (2005) provides additional information on the expenditures on
fishing equipment and related recreational angling activities across the U.S.
4-28
-------
Total Days of Fishing
< 5,000
5,000 -10,000
10,001 -20,000
j 20,001 - 30,000
> 30,000
Figure 4-8. 2001 Total Recreational Fishing Days
I «-.">•
S^t
/
Days of Fishing by Residents
< 5,000
5,001 -10,000
10,001 -15,000
15,001 -20,000
> 20,000
Figure 4-9. 2001 Recreational Fishing Days, State Residents
4-29
-------
Days of Fishing by Nonresidents
Zj 4.000
Figure 4-10. 2001 Recreational Fishing Days, Non-State Residents
A number of surveys have been conducted to determine the amount of sport fish consumed
by anglers. A discussion of consumption rates for freshwater recreational anglers are provided in
Section 4.5 and in Section 11 of this report. Tables 4-25 and 4-26 provide information on bag
limits and size limits for recreational fishing.
4-30
-------
Table 4-25. Freshwater Fishing Bag and Size Limits for Selected States and Species
Species
Black Bass
White bass
Striped and Hybrid Bass
Panfish
Trout
Catfish
Bullhead
Crappie
California
Bag Limil
5-10°
none
2b
none
0-5"
none
none
25"
Min. Size
(inches)
12-no limit"
none
18"
none
8-no limit"
none
none
none1"
Texas
Bag
Limit
5"
25b
5"
none
5b
channel/blue: 25b
flathead: 5b
none
25b
Min. Size (inches)
14"
10"
18b
none
noneb
channel/blue:
12"
flathead: 18b
none
10"
Minnesota
Bag Limit
6b
30
none
none
Lake: 2b
Stream: 0-5"
5, 1 fish over 24'
100
10b
Min. Size
(inches)
noneb
none
none
none
noneb
none
none
noneb
Florida
Bag
Limit
only 1 fish may be
over 22" b
only 6 fish may be
over 24"
only 6 fish may be
over 24" b
noneb
none
none
none
noneb
Min. Size
(inches)
12-24"
none
none
none
none
none
none
noneb
"Limit depends on location.
'Special limits apply in certain areas.
4-31
-------
Table 4-26. Saltwater Fishing Bag and Size Limits for Selected States and Species
Species
Flounder
Halibut
Striped Bass
Spotted Seatrout
Bluefish
Salmon
Mackerel
Florida
Bag Limit
10
none
none
4-5°, Irish
over 20"
10
none
none
Min. Size
(inches)
12
none
none
15
12 (FL)
none
none
California
Bag
Limit
none
California: 3-5°
Pacific: 1
2
3 (all trout)
none
2
none
Min. Size
(inches)
none
California: 22
Pacific: 32
18-no limit"
none
none
20-24"
none
Texas
Bag Limit
10
none
5
10, 1 over
25"
none
none
King: 2
Spanish: 15
Min. Size
(inches)
14
none
18
15
none
none
King: 27
Spanish: 14
North Carolina
Bag
Limit
none
none
0-3"
10
15; only 6 over 24"
(coastal waters)
none
King: 3
Spanish: 15 (FL)
Min. Size (inches
13-14 (TL)a
none
18-28(TL)a
12 (TL)
none
none
King: 24
Spanish: 12 (FL)
"Limit depends on location.
'Special limits apply in certain areas.
Total length (TL) is measured from tip of snout with mouth closed to tip of compressed tail.
Fork length (FL) is measured from tip of snout to middle of fork in tail.
4-32
-------
4.3.3.2 Recreational Fishing Demographics
The USFWS currently maintain statistics on U.S. freshwater angler participation rates in
addition to demographic characteristics of U.S. anglers. Angler participation rates are usually
measured in number of anglers and the number of angler fishing days for each state (Pechan,
2005). Additional survey data are available that provide demographic information for U.S.
anglers. Below are some of the highlights of the demographic statistics followed by a break down
of angler participation by region.
Based on USFWS survey data, 67% of recreational fishermen were between the ages of 25
to 54. Male anglers outnumbered women by a ratio of 3 to 1. Freshwater recreational anglers
fished in ponds, lakes, and reservoirs over rivers and streams at a ratio of 2 to 1. Table 4-27
shows the percentage of freshwater anglers by age group. Table 4-28 shows the percentage of
anglers by sex and age. These data show that women of childbearing age (16-44) make up 60%
of female anglers. Female anglers make up 26% of all anglers. Thus, women of childbearing age
make up 15.6% of all anglers (i.e. 60% of the 26%).
Table 4-27. Percent of Freshwater Anglers by Age Group
Age Group
16-17
18-24
24-34
35-44
45-54
55-64
65-Older
Percent of
Total Anglers
4
9
19
27
20
12
9
Source: USFWS, 2002.
Table 4-28. Percent of Anglers by Sex and Age Group
Age Group
16 to 17 years
18 to 24 years
25 to 34 years
35 to 44 years
45 to 54 years
55 to 64 years
65 years and older
All Ages
Male
4
7
19
26
20
13
10
74
Female
3
8
21
28
21
11
8
26
Source: USFWS, 2002.
USFWS statistics showed that non-white fishermen encompassed a greater portion of
saltwater recreational fishermen with an average of 12% compared to comprising only 7% of total
freshwater recreational fishermen. From Table 4-29, one can see that the western part of the
4-33
-------
United States had the highest participation rate per capita of non-white anglers followed by the
Southeast. The lowest participation rates among non-white anglers were in the Great Lakes
region and the Northern Plains.
Table 4-29. 2001 Demographic Summary, Angler Race (% Non-White)
Region
West
Southeast
South Central
Northeast
Great Lakes
Northern Plains
Average
Freshwater
12
11
7
5
4
3
7
Saltwater
22
9
7
10
N/A
N/A
12
Great Lakes
N/A
N/A
N/A
3
11
0
5
Source: USFWS, 2002.
According to a Recreational Boating and Fishing Foundation 2003 Boating and Fishing
Attitude, Segmentation Study (RBFF, 2003) on minorities, Hispanics demonstrated to be more
active participants than African Americans, among a group of respondents that were categorized
to be active or prospective fishing participants, with 74% showing some incidence of fishing and
13% considered avid fishermen compared to 55% and 7%, respectively, for African Americans.
Only 17% of Hispanics had never participated in fishing while 25% of African Americans had
not. However, compared to whites, both ethnic groups showed a lower active participation rate
(see Table 4-30). It should be noted that Asian fishing participation statistics were not available.
Table 4-30. Incidence of Fishing Among White, African American, and Hispanic
Recreational Boaters
Participant Description
Active Participant
Avid
Semi-Avid
Occasional
Lapsed (prospective)
Never Participated
(prospective)
White (%)
80
18
19
43
11
10
African
American (%)
55
7
17
31
20
25
Hispanic (%)
74
13
14
47
10
17
Source: RBFF, 2003.
USFWS survey data showed fishing participation rate to be positively correlated with
household income (see Table 4-31). Household income groups of $40 thousand per year or more
had the highest participation rate with over 22% of each income group participating in fishing.
Only 8% of households with annual income less than $10 thousand participated in some kind of
fishing activity.
4-34
-------
Table 4-31. Percent of U.S. Population Who Fished (By Household Income)
Income Group
Less Than $10,000
$10,000- 19,999
$20,000 - 24,999
$25,000 - 29,999
$30,000 - 34,999
$35,000 - 39,999
$40,000 - 49,999
$50,000 - 74,999
$75,000 - 99,999
$100,000 or Greater
Percent of Income
Group that Fished
8
11
14
16
18
20
22
23
23
22
Source: USFWS, 2002.
4.3.4 Total U.S. Demand
NMFS estimated that the total 2002 U.S. commercial finfish and shellfish demand was
66.0 Ib/capita (NMFS, 2003b). Of this total, 15.6 Ib/capita represented edible fish and shellfish
meat. For comparison, Jacobs et al (1998) obtained an estimate of total fish and shellfish
consumption of 15.65 grams/person/day (12.6 Ib/capita) from the Continuing Survey of Food
Intake by Individuals (CSFII) Study. This survey was conducted during the years of 1989-1991.
During those years, NMFS estimated an average U.S. consumption rate of 15.2 Ib/capita.
Therefore, these two sources show reasonably good agreement, although the NMFS estimate does
not include recreational demand.
4.3.4.1 Finfish
From Table 4-14, the total 2002 U.S. recreational marine demand was 103,531 metric tons
(all finfish). From this value, the total demand for recreationally-harvested finfish is 0.79
Ib/capita. From Table 4-20, the total 2002 U.S. commercial finfish demand was estimated to be
33.0 Ib/capita. As mentioned earlier in this section, there are no similar data for recreational
freshwater finfish demand.
4.3.4.2 Shellfish
From Table 4-21, the total 2002 U.S. commercial shellfish demand was estimated to be
9.07 Ib/capita. No data were identified to estimate recreational shellfish demand.
4-35
-------
4.4 Economic Value of Key Species
This section presents available information on the economic value of the commercial and
recreational fishing industries. Available information on the wholesale and retail pricing of
products from the commercial fishing industry are also provided. In other sections, we have
discussed the level of production of commercial fish. In this section, we combine production
levels with economic value and see that in the commercial fishing industry, shellfish sales total
$1.8 million while finfish sales total $1.3 million. Shrimp, crab, and lobster are the top shellfish
species sold, while pollock (used in fish sticks) is the top finfish species sold. Domestic tuna sales
are lower in sales than several other species (however, the U.S. imports large quantities of tuna).
4.4.1 Finfish
Table 4-32 provides a summary of the economic value of commercial landings for several
finfish types. Note that these data exclude aquacultural products, and that for catfish, most of the
production of these fish comes from aquaculture. The economic value data are also shown
graphically in Figure 4-11 as a percentage of the total 2002 commercial finfish landings
($1,368,877,000). Additional details on the economic value of 2002 commercial landings are
provided in a separate report (Pechan, 2005). Table 4-33 provides information on 2001 wholesale
pricing of several fresh finfish (NMFS, 2005). Data for 2002 were not available. The NMFS
gathers wholesale pricing data from the Fulton Fish Market in New York City. Additional data
are provided in Pechan (2005).
Table 4-32. 2002 Economic Value of Commercial Landings for Important Finfish Types
Finfish Type
Pollock
Menhaden
Salmon
Cod
Hake
Sole
Herring
Sardine
Tuna
Other Finfish
Total 2002 Finfish
Economic Value
(thousand dollars)
209,890
105,172
156,082
126,844
26,215
22,437
21,310
10,824
85,478
604,625
1,368,877
% of Total Finfish
Landed by weight
41.4
21.6
6.94
6.70
3.86
3.18
2.70
2.66
0.61
10.3
100
Source: NMFS, 2003b.
4-36
-------
Pollock
15%
Other Finfish
44%
Menhaden
8%
Salmon
11%
Tuna
6%
Hake
Sole 2%
2%
Sardine, Herring
1% 2%
Figure 4-11. 2002 Market Share of Commercial Finfish (% of Total Economic Value)
Table 4-33. Average 2001 Wholesale Prices for Several Fresh Finfish Types
Finfish Type
Pollock
Flounder
Swordfish
Cod
Whiting
Croaker
Price ($/lb)
1.45
1.95
3.91
1.77
0.59
0.28
Source: NMFS, 2005 (Fulton Fish Market).
4.4.2 Shellfish
Table 4-34 provides a summary of the economic value of commercial landings for several
shellfish types. The economic value data are also shown in Figure 4-12 as a percentage of the
total 2002 commercial shellfish landings ($1,808,167,000). Additional details on the economic
value of 2002 commercial landings are provided in Pechan (2005). Table 4-35 provides 2001
wholesale pricing data for several shellfish types as reported by NMFS (NMFS, 2005). Data for
2002 were not available.
4-37
-------
Table 4-34. 2002 Economic Value of Commercial Landings of Several Shellfish Types
Shellfish Type
Shrimp
Crab
Squid
Clam
Lobster
Scallop
Ovster
Other Shellfish"
Total 2002 Shellfish
Economic Value
(thousand dollars)
522,399
397,349
43,540
171,134
316,085
203,494
93,449
60.717
1,808,167
% of Total Shellfish
Landed by Weight
25.9
23.2
15.5
9.82
6.56
4.00
2.78
12.2
100
Source: NMFS, 2003b.
' NMFS includes some non-shellfish species in the "Other shellfish" group including
seaweed, sponges and turtles (Pechan, 2005).
Other Shellfish
3%
Scallop
11%
Shrimp
30%
Figure 4-12. 2002 Market Share of Commercial Shellfish (% of Total Economic Value)
4-38
-------
Table 4-35. Average 2001 Wholesale Prices for Several Fresh Shellfish Types
Shellfish Type
Crab3
Clam"
Lobster
Oyster"
Squid
Price ($/lb)
48.07
106.82
6.85
60.87
0.92
Source: NMFS, 2005 (Fulton Fish Market).
' Pricing is in $/bushel.
4.4.3 Fish Products
Fishery products are processed fish products from commercially caught fish. These
products are categorized as: fresh and frozen (fish fillets and steaks, fish sticks and portion, and
breaded shrimp); canned products (e.g. tuna, etc.); and industrial fishery products (fish meal and
oil). A summary of the 2002 economic value of these products is provided in Table 4-36 below
(NMFS, 2003b). In November of 2004, the wholesale price offish meal was $589/metric ton
(NMFS, 2005). The wholesale price offish oil was $617/metric ton. Wholesale pricing
information for other fishery products was not identified.
Table 4-36. 2002 Economic Value of Commercial Fishery Products
Fishery Product
Economic Value
(thousand dollars)
Weight
(metric tons)
Fresh and Frozen
Fillets and Steaks
Fish Sticks and Portions
Breaded Shrimp
983,900
288.600
475,500
235,464
106,777
67,360
Canned
Salmon
Sardines
Tuna
Clams
Other
295,600
n/a
675,300
117,400
139,600
101,470
n/a
248,119
63,005
165,337
Industrial Fishery Products
Fish Meal
Fish Oils
Other
Total
139,700
41,400
78,900
3,235,900
289,351
95.664
n/a
Source: NMFS, 2003b.
4.5 Characterization of Fish Consuming Populations
This section provides background information on fish consumption in the U.S. including
(a) identification of specific fish consumption pathways (e.g., commercial saltwater fish
consumption, self-caught freshwater fish consumption), (b) the numbers of individuals potentially
exposed through those pathways (i.e., specific fish consuming populations such as recreational
freshwater anglers) and (c) a characterization of the general levels offish consumption associated
with those fish consuming populations. In Section 4.5.1, we begin by identifying a variety offish
4-39
-------
consumption pathways through which the U.S. population can be exposed to mercury that has
bioaccumulated in fish. This section also clarifies the linkage between fish consumption
pathways and specific consuming populations (e.g., consumption of self-caught freshwater fish by
the freshwater recreational angler population). Section 4.5.2 presents some basic demographic
information for the fish consuming populations. Section 4.5.3 presents general fish consumption
rate data for the three key populations covered in Section 4.5.2. Note, that Section 4.5.4 is not
intended to provide an exhaustive review of published fish consumption data, but rather to
provide the reader with a perspective for how consumption rates may differ across pathways and
populations. Section 4.5.4 combines the demographic (count) and consumption information to
provide a characterization of the populations potential affected by mercury exposures and to help
identify those populations that are to be included in the benefits analysis in Section 11 of this
report.
4.5.1 Fish Consumption Pathways and Associated Fish-Consuming Populations
Mercury exposure through fish consumption can be considered in terms of specific fish
consumption pathways such as self-caught freshwater fish consumption and commercial saltwater
fish consumption. A list offish consumption pathways is presented in Figure 4-13. The
commercial fish consumption pathways in Figure 4-13 can be further differentiated as domestic-
versus foreign-sourced fish. Note, it is also possible to further differentiate both commercial and
self-caught pathways by fish species (e.g., foreign-sourced commercial yellow-fin tuna).
Figure 4-13. Fish Consumption Pathways
U.S. Fish Consuming
Population
Consumers of
commercially-
produced fish
Consumers of self-
caught fish
Saltwater
(estuarine, near-
coast, deep
ocean)
Freshwater
Saltwater
(estuarine,
ocean)
Freshwater
(rivers, lakes)
4-40
-------
Each of the pathways listed in Figure 4-13 can be associated with a specific fish
consuming population. For example, commercial saltwater fish is purchased and consumed by a
subset of the U.S. population including those individuals that buy this type offish in foodstores
and restaurants. However, this delineation offish consuming populations according to fish
consumption pathway is somewhat artificial since, in reality, most individuals consume a mixture
offish over time reflecting several of these pathways. For example, recreational anglers may fish
in both freshwater and saltwater waterbodies and consume commercially-produced freshwater and
saltwater fish (e.g., store-bought or restaurant-bought). Therefore, these recreational anglers are
exposed to multiple fish consumption pathways simultaneously.
4.5.2 Fish Consuming Populations
In the previous sections, we present total fish produced and U.S. demand for various fish
species. This information gives the reader a general characterization of the magnitude offish
consumed in the U.S. (and in relation to the various consumption pathways discussed in this
section). In addition, Table 4-37 provides a characterization of the number of people in the
population who eat fish from the different consumption pathways. Table 4-37 provides
population counts for three of the key fish consuming populations of concern from a benefits
standpoint including: (a) the general population who consumes both commercial and (to a lesser
extent) self-caught fish, in both cases including a mix of freshwater, saltwater and estuarine
species (b) recreational freshwater anglers who consume fish obtained from inland lakes, creeks
and rivers and (c) recreational saltwater anglers who fish in the estuarine, near-coastal and open
ocean areas.
Ihe freshwater angler and saltwater angler numbers presented in Table 4-37 include both
"total anglers" (number of consuming and non-consuming anglers) and "total consumers"(the
number of total "consuming" individuals linked to the recreational self-caught fishing activity,
i.e., the total number of individuals including family and friends with which a fisher shares their
catch). These two sets of counts (total anglers and total consumers) were calculated separately for
freshwater and saltwater recreational categories. The "total anglers" estimates are obtained
directly from the National Survey of Fishing, Hunting and Wildlife-Associated Recreation
(USFWS, 2002). However, the "total consumers" numbers are calculated by multiplying the total
recreational fisher number by (a) the fraction of recreational fishers who consumer their catch
(0.84 or 84%) and (b) a factor reflecting the number of family/friends with which the average
recreational fisher shares their catch (2.5).' For purposes of comparing the number of recreational
fishers (freshwater and saltwater) to the number offish consumers in the general population, it is
most appropriate to use the "total consumers" numbers for the recreational groups.
Note that there is overlap in the three populations presented in Table 4-37. For example,
the general population includes both the recreational freshwater and saltwater anglers. In
addition, there is some degree of overlap between freshwater and saltwater anglers. Each of the
1 The "fraction of recreational fishers who consume" factor (0.84) was derived by taking the average of "consuming fraction"
values presented in three key freshwater studies presented in EPA, 1997 including West et al., 1989, Chemrisk 1991, and West et
al., 1993. No comparable "consuming fraction" values were readily available for saltwater anglers, so this freshwater value was
applied to the saltwater angler category. The "sharing" factor of 2.5 was obtained from EPA, 1997 and is based on values
presented in a number of different fishing activity surveys as documented in EPA, 1997.
4-41
-------
three populations in Table 4-37 is further differentiated into: (a) adult consumers and (b) prenatal
infants likely exposed to mercury through maternal consumption of mercury-contaminated fish
(the target group considered in the benefits analysis).
The purpose, in presenting the data in Table 4-37, is to provide the reader with perspective
on the sizes of key fish consuming populations in the U.S.. These data demonstrate that the
recreational freshwater fisher population (28 million) is significantly larger than the recreational
saltwater population (9 million), while both of these populations are significantly smaller than the
general population offish consumers in the U.S. (184 million) which includes many individuals
receiving a significant fraction of their fish from commercially-produced stocks. This suggests
that, while the benefits analysis has captured a key fish consuming population in modeling
recreational freshwater anglers, a potentially large group of individuals (general consumers and
recreational saltwater anglers) are not included in the primary benefits analysis (this issue is
discussed in greater detail in Section 4.5.4.1).
4.5.3 General Fish Consumption Rates for Key Fish Consuming Populations
In presenting perspective on the degree of potential exposure to mercury for different fish
consuming populations in the U.S., in additional to considering the total number of individuals
within each population (information presented in Section 4.5.4), it is also important to consider
fish consumption rates for those populations. Table 4-38 presents general fish consumption rates
(including mean and high-end estimates where available) for key fish consuming populations
including: (a) the general population exposed through commercial and self-caught fish, (b)
recreational freshwater anglers and (c) recreational saltwater anglers. Due to limitations in
available data, it was not possible to identify consumption rates for the full set offish
consumption pathways identified in Figure 4-13. However, the values presented in Table 4-38 do
provide coverage for key populations. All of the values presented Table 4-38 represent long-term
dietary consumption (i.e., annual-averaged daily intake rates), thereby allowing comparison
across pathways/sub-populations (additional characteristics relevant to this discussion are also
included in Table 4-38 as part of the Comments section).
4-42
-------
Table 4-37. Demographic (count) Data for Key Fish Consuming Populations in the U.S.
Population
Population
Count
Comments
References
General population (including consumption of commercial and self-caueht fish including saltwater, freshwater and estuarine species)
Adult fish consumers (>18yrs)
Female adult (1 5-44 yrs) fish
consumers
Infants born to mothers who
consume fish in the general
population
184,000,000
53,000,000
3,430,000
88% of adults surveyed reported
consuming fish or shellfish at least once
within the last month. Fish consumption
likely includes commercial saltwater and
freshwater as well as self-caught
freshwater and saltwater for some fraction
of respondents.
86% of females 1 5-44 yrs (general fertility
range) surveyed reported consuming fish
or shellfish within the last month. Fish
consumption likely includes commercial
saltwater and freshwater as well as self-
caught freshwater and saltwater for some
fraction of respondents.
Developed by applying the general U.S.
fertility rate for 2000 (64.8 births per 1000
females 1 5-44 yrs old) to the number of
adult females aged 15-44vrs.
Percent fish consumption
obtained from NHANES III as
summarized in EPA, 1997.
Demographic count for adults
obtained from US Census 2000.
Percent fish consumption
obtained from NHANES III as
summarized in EPA, 1997.
Demographic count for females
15-44 yrs obtained from US
Census 2000.
Fertility rate data from US
Census 2000.
Freshwater anglers (self-caught freshwater fish consumption only)
Total anglers (>1 5 yrs)
Total consumers of recreational ly
-caught freshwater fish (all ages)
Infants born to mothers who
consume recreationally-caught
freshwater fish
27,900,000
58,590,000
420,000 to
580,000
Includes total number of anglers fishing in
freshwater water bodies including streams,
rivers and lakes (excludes Great Lakes).
Includes consumers and non-consumers
(i.e., catch and release)
Based on application of "consuming"
factor and "sharing" factor to the total
anglers estimate above (see text for details')
Modeled as part of the mercury benefits
analysis conducted for recreational
freshwater anglers
National Survey of Fishing,
Hunting and Wildlife-
Associated Recreation (USFWS,
2002)
Estimate generated using
USFWS, 2002 data, combined
with US Census 2000 data (see
Section 1 1 for additional
details).
Saltwater anglers (self-caught saltwater and estuarine fish consumption only)
Saltwater angler (self-caught
saltwater and estuarine fish
consumption)
Total consumers of recreationally-
caught saltwater fish (all ages)
Infants born to mothers who
consume recreationally-caught
saltwater fish
9,100,000
19,110,000
135,000 to
186,000
Includes consumers and non-consumers
(i.e., catch and release)
Based on application of "consuming"
factor and "sharing" factor to the total
anglers estimate above (see text for details)
National Survey of Fishing,
Hunting and Wildlife-
Associated Recreation (USFWS,
2002)
Estimated using the applicable ratio (between prenatally-exposed infants and
anglers) developed for freshwater recreational anglers (see above)
4-43
-------
Table 4-38 Fish Consumption Rates for Key Fish Consuming Populations in the U.S.
Population
Fish Consumption
(long-term annual-average equivalent in
g/dav)
Comments
References
General population (including consumption of commercial and self-caught fish including saltwater, freshwater and estuarine species)
General population fish
consumers (all ages combined)
General population adults (>18
yrs)
General population females
(15-44 yrs)
- freshwater/ estuarine: 6.0 (mean)
- saltwater: 14.1 (mean)
- total fish: 20.1 (mean); 60.3 (95*%)
- freshwater/ estuarine: 5.6 (mean)
-saltwater: 12.4 (mean)
-total fish: 18.0 (mean)
- freshwater/estuarine: 4.3 (mean)
-saltwater: 10.0 (mean)
-total fish: 14.3 (mean)
- uncooked (but does seem
to represent edible portion to
some extent)
- total population
(consumers and non-
consumers in survey)
- as consumed
- total population
(consumers and non-
consumers in survey)
- as consumed
- total population
(consumers and non-
consumers in survey)
EPA, 1996a(CSFH,
1989- 1991) (as reported
in EPA, 1997)
EPA, 1996a(CSFII,
1989- 1991) (as reported
in EPA, 1997)
EPA, 1996a(CSFII,
1989- 1991) (as reported
in EPA, 1997)
Freshwater anglers (self-caught freshwater fish consumption only)
Anglers (all ages combined)
- Maine 5 (mean); 13 (95th%)
- New York 5 (mean); 1 8 (9Sth%)
- Michigan 12 (mean); 39 (96th%)
-Michigan 17 (mean)
- EPA "recommended" freshwater fish
consumption rate: 8 (mean); 25 (95th%)
- consumers plus non-
consumers (catch and
release)
- uncooked (but does seem
to represent edible
portions)
-Ebertetal., 1992
- Connelly et al., 1987
-Westetal., 1989
- West etal., 1993
(all as reported in EPA,
1997)
Saltwater anglers (self-caught saltwater and estuarine fish consumption only)
Anglers and consumers offish
caught by anglers (e.g., family
members) (all ages combined)
-Atlantic: 5.6 (mean); 18.0 (95th%)
- Pacific: 2.0 (mean); 6.8 (95th%)
- Gulf: 7.2 (mean); 26.0 (95th%)
- edible fraction
(uncooked)
- consumers and non-
consumers
NMFS, 1993 (as reported
in EPA, 1997)
Review of the values presented in Table 4-38 reveals some interesting comparisons and
contrasts between consumption rates for different populations:
The general population has greater average consumption rates for saltwater fish
14.1 g/day) than do the saltwater anglers (2.0 to 7.2 g/day depending on region). It
is important to point out that the saltwater angler values refer to self-caught fish
only. It is likely that they also consume some amount of commercially-produced
saltwater fish, which would mean that in terms of total saltwater fish, the saltwater
anglers might have higher consumption than the general population. However, in
comparing self-caught saltwater fish consumption to general population saltwater
fish consumption, the latter is a larger value on average.
• Self-caught freshwater fish consumption (5 tol? g/day depending on region) is
larger, on average, than self-caught saltwater fish consumption (2 to 7 g/day
depending on region). These data suggest that freshwater anglers may, on average,
have twice the intake of self-caught fish than saltwater anglers.
• Regional data on consumption rates suggests that populations can differ
significantly in their fish consumption depending on where they are located. This
can have important implications for modeling distributional benefits for fish
consumption as part of an equity analysis. If fish consumption rates differ
4-44
-------
significantly across regions, then this may suggest that exposure through fish
consumption may also differ regionally (of course this will also depend on the
regional variability of mercury concentrations in fish consumed). In considering
distributional equity, it may be necessary to conduct more refined exposure
modeling that tracks these different patterns offish consumption regionally (and
links them to spatial distribution of fishers and mercury fish tissue concentrations).
However, in reality, the patchiness of data characterizing regional variability in
fish consumption rates tends to prevent a comprehensive treatment of this issue in
the context of a national-scale benefits analysis. Instead, individual case studies
focusing on specific regions and assessing the potential importance of regional
variability in factors such as fish consumption rates and the spatial distribution of
fishing populations can be conducted.
4.5.4 Discussion of Population and Fish Consumption Data in the Context of the Mercury
Benefits Analysis
Information presented in Section 4.5.4 can be used to gain a perspective on the degree to
which the benefits analysis presented in this RIA has covered key fish consuming populations in
the U.S. from the standpoint of potential benefits linked to mercury emissions reductions.
Specifically, the recreational freshwater angler population modeled for the benefit analysis can be
compared, in terms of total fish consumption, to the other key populations (i.e., recreational
saltwater anglers and the general U.S. fish consuming population).
Table 4-39 presents total fish consumption estimates for the three key populations covered
in Sections 4.5.2 and 4.5.3 (total consumers associated with recreational saltwater angler activity,
total consumers associated with recreational freshwater angler activity and the general fish
consuming population in the U.S.). Note, for the recreational angling scenarios, the total
consumer categories were used in conducting this comparison rather than the total angler
categories, since the former focus on the total number of individuals consuming fish caught by the
anglers. These estimates were generated by multiplying average (mean) fish consumption values
presented in Table 4-38 by the total demographic counts for these populations presented in Table
4-37. While this approach is relatively simplistic and is subject to uncertainty, it is considered
sufficient to provide a general perspective on the magnitude of differences in total fish
consumption by the three populations.
Table 4-39. Total Fish Consumption for Recreational Saltwater Anglers, Recreational
Freshwater Anglers and General U.S. Fish Consuming Population
Population (adults)
General fish consuming population
Recreational freshwater anglers
Recreational saltwater anglers
Population count
184,000,000
58,590,000
19,110,000
Consumption Rate
(g/day)
20.1
8
4.9*
Total fish consumption (kg/year)
1,349,916,000
171,082,800
(13% of general population consumption value]
34,178,235
(3% of general population consumption value)
*The consumption rate presented for recreational saltwater anglers was derived by taking the average of the regional values.
4-45
-------
Total fish consumption values presented in Table 4-39 highlight the fact that the primary
benefits analysis for this RIA is capturing a relatively small fraction of overall fish consumption.
Specifically, the recreational freshwater angler population modeled for the primary benefits
estimates represents only 13% of total fish consumption in the U.S. (comparing self-caught
freshwater fish consumption by recreational anglers to consumption of all fish types by the
general population). It is important to note, that the actual magnitude of IQ benefits for a given
population is a function of (a) that population's fish consumption rate, (b) the baseline fish tissue
concentrations for fish that the population consumes and (c) the magnitude of changes in mercury
deposition to the waterbodies containing fish that a given population catches and the relationship
between those deposition changes and mercury fish tissue concentrations. In short, the likely
difference in IQ benefits between the three populations presented in Table 4-39 is dependent on
several key factors related to mercury concentration in fish in addition to total fish consumption
rates for these populations. While it is still informative to consider that approximately 86% of
fish consumption by the U.S. population is not being covered in the benefit analysis, Section 8 of
this RIA shows that deposition to U.S. waterbodies from coal-fired power plants will
predominently occur in freshwater waterbodies in the Eastern-half of the U.S. Thus, the benfeit
analysis of freshwater recreational anglers captures the primary segment of the affected
population. To the extent that CAMR reductions will impact fish in coastal regions and the
ocean, there remains a potential for a small amount of additional IQ benefits related to the general
population offish consumers (including foreign and domestically-caught commercial fish and
coastal recreationally-caught fish).
4.5.4.1 Potentially High-ExposureSubpopulations
The primary benefits analysis includes consideration for several potentially high-exposure
subpopulations including:
• A high fish consumption rate study population that is defined as "subsistence"
fishers for the purposes of this study;
A low-income high fish consumption study population (an alternative approach to
model subsistence fishers);
• A Southeast Asian ethnic group with high freshwater fish consumption due to
cultural practices; and
• A Native American population with high freshwater fish consumption due to
cultural practices.
While these special subpopulations provide important insights into the issue of
distributional (equity) benefits (i.e., the potential for IQ benefits to be disproportionately
distributed across the U.S. population), they do not represent a significant (net) fraction of overall
benefits due to the size of the subpopulations relative to overall U.S. fish consumption. In the
case of the high fish consumption (subsistence) population (first bullet above), this group is a
subset of the larger recreational freshwater fisher population and therefore, is incorporated as part
of the recreational (self-caught) freshwater angler analysis in Section 10.
4-46
-------
Because these subpopulations are included primarily to provide insights into potential
distributional (equity) issues, and do not contribute a significant fraction to net benefits, they are
not discussed in detail in this section, which is primary concerned with provide the reader with
perspective on the magnitude of total U.S. population consumption through various fish
consumption pathways.
4.6 Summary
Because fish consumption is the primary pathway for exposure to methylmercury, this
section provides background information on fishing activity through a profile of the fishing
industry in the United States. Methylmercury exposure through fish consumption is considered in
terms of specific fish consumption pathways such as commercial and self-caught (recreational)
fish consumption, which is composed offish species from freshwater and saltwater sources.
4.6.1 Commercial Fish Production, Demand, and Consumption
Fish products from commercial fishing activities include: fresh and frozen (fish fillets and
steaks, fish sticks and portion, and breaded shrimp); canned products (e.g. tuna, etc.); and
industrial fishery products (fish meal and oil).
The amount of commercial fish landed in inland waters and state and territorial seas (0-3
miles from shore) is estimated to be 36% of the total commercial landings, while 61% of
commercial fish were landed in the Federal exclusive economic zone (3 - 200 miles offshore).
The Pacific coast region accounted for 65% of total commercial landings with Alaska and
California accounting for the largest portion. Alaska alone accounted for over 26% of total U.S.
landings. The Gulf Coast accounts had significantly less landings than the Pacific Coast at 18%,
followed by New England and the Chesapeake Bay contributing 6% and 5% to U.S. landings,
respectively.
Total commercial finfish landings equal 3.6 million metric tons. Commercial shellfish
landings total approximately 600,000 metric tons. Pollock and Menhaden were by far the finfish
types with the highest commercial landings by U.S. craft in 2002. These two finfish types are
almost exclusively caught by commercial fishermen. Pollock are often used in making frozen fish
products (e.g., fish sticks), surimi (see Section 4.2.1.3), and other minced fish products.
Menhaden are often used to produce cut or live bait, fishmeal and fish oil. Shrimp and crab are
the top species of shellfish caught commercially.
Total aquacultural production from fish farms in 2002 was 393 thousand metric tons with
catfish accounting for 73% or 286 thousand metric tons, followed by trout at 6% of total
aquaculture production.
The U.S. exports total 1.09 million metric tons of commercial fish, with 56% of exports
going to Asia, followed by 20% to Europe and 18% to North America. Total edible imports for
2002 were 4.0 million metric tons with finfish contributing 65% of the total. An estimated 46% of
U.S. imports were from Asia while 26% were from other regions in North America.
4-47
-------
NMFS estimated that the total 2002 U.S. commercial finfish and shellfish per capita
demand was 66.0 Ib/capita (NMFS, 2003b). Of this total, 15.6 Ib/capita represented edible fish
and shellfish meat. Most of the U.S. tuna demand is met through imports (less than 7% of the
tuna demand is met through commercial landings by the domestic fleet).
4.6.2 Recreational Fishing Activity, and Consumption
Recreational fishing activity occurs in saltwater (estuaries, coastal regions, and open
ocean) and freshwater locations (lakes, rivers, and streams). Roughly 32% of the 103,529 metric
tons of marine finfish landed from recreational angling in U.S. waters in 2002 occurred at
distances of 3-200 miles from shore, while 29% of total finfish landings were within 3 miles from
shore. The remaining 39% of finfish were landed in Inland Waters. For shellfish, 65% of the
534,608 metric tons landed were made within Inland Waters and State and Territorial Sea
(< 3 miles from shore).
Data for freshwater recreational harvest are not available; however, the number of anglers
and number of days of fishing is available from the 2001 National Survey of Fishing, Hunting,
and Wildlife-Associated Recreation (USFWS, 2002). There are approximately 28 million
freshwater recreational anglers and 9 million saltwater recreational anglers. The region with the
largest number of anglers was the southeast followed by the south central United States. Bass
(large and small mouth), pan fish, trout, catfish, and crappie were the most popular target species
for freshwater anglers nationwide. Considering the number of days spent fishing, high levels of
recreational fishing activity are shown to occur in the Great Lakes States, Florida, New York,
Texas, and California.
The total demand for recreationally-harvested finfish is 0.79 Ib/capita. From Table 4-31,
the total 2002 U.S. commercial finfish demand was estimated to be 33.0 Ib/capita. The total 2002
U.S. commercial shellfish demand was estimated to be 9.07 Ib/capita. No data were identified to
estimate recreational shellfish demand.
4.6.3 Overall Conclusions
This section demonstrates that the recreational freshwater fisher population (28 million) is
significantly larger than the recreational saltwater population (9 million), while both of these
populations are significantly smaller than the general population offish consumers in the U.S.
(184 million) which includes many individuals receiving a significant fraction of their fish from
commercially-produced stocks.
Based on information provided in this section, we see that commercial fish consumption
constitutes a large portion of exposure to methylmercury. However, a large majority of the
commercial fish consumed are imported from foreign sources, or 3-200 miles offshore by
domestic commericial fishermen (with a majority of domestic landings occuring off the Pacific
coast). These sources of exposure are not likely to be impacted by the control of utilities from the
CAMR rule. However, methylmercury concentrations from freshwater sources are likely to be
affected by control domestic electric utilities. Therefore, the quantified benefit analysis in Section
10 evaluates the benefits of improved health from reduced exposure to methylmercury from
recreational freshwater fishing activities.
4-48
-------
4.7 References
ASA, 2003. American Sportfishing Association. Today's Angler, 2003.
BEA, 2004. Bureau of Economic Accounts. Gross-Domestic-Product-(GDP)-by-Industry Data,
GDPbyInd_GO_NAICS.xls, downloaded from
http://www.bea.doc.gov/bea/dn2/gdpbyind_data.htm. December 2004.
BOC, 2005. U.S. Bureau of Census. Data downloaded from the American Factfinder database at
http://factfinder.census.gov/home. accessed January 2005.
BOC, 2004a. U.S. Bureau of Census. CBP United States Economic Profiles, downloaded from
http://www.census.gov/epcd/cbp/view/cbpus.html. 2004.
BOC, 2004b. Bureau of Census. Statistics of United States Businesses, downloaded from
http://www.census.gov/csd/susb/susb.htm. 2004.
EPA, 1997. U.S. Environmental Protection Agency, Volume I - General Factors Exposure
Factors Handbook Update to Exposure Factors Handbook, EPA/600/8-89/043 - May
1989, EPA/600/P-95/002Fa, August 1997.
Jacobs, H.L., H.D. Kahn, K.A. Stralka, and D.B. Phan (1998). Estimates of per Capita Fish
Consumption in the U.S. Based on the Continuing Survey of Food Intake by Individuals
(CSFII). Risk Analysis, 18 (3):283-291, 1998.
NASS, 2002. National Agricultural Statistics Service, U.S. Department of Agriculture. 2002
Census of Agriculture, 2002.
NASS, 1998. National Agricultural Statistics Service, U.S. Department of Agriculture. 1998
Census ofAquaculture, April, 1998.
NMFS, 2005. National Marine Fisheries Service, National Oceanic and Atmospheric
Administration. Commercial Fishery Landings data, downloaded from
http://www.st.nmfs.gov/stl/market news/, accessed January 2005.
NMFS, 2004. National Marine Fisheries Service, National Oceanic and Atmospheric
Administration. Fisheries of the U.S., 2003, October 2004.
NMFS, 2003a. National Marine Fisheries Service, Fisheries Statistics Division, National Oceanic
and Atmospheric Administration. Statistical Highlights: Fisheries of the United States,
2003, 2003.
NMFS, 2003b. National Marine Fisheries Service, National Oceanic and Atmospheric
Administration. Fisheries of the U.S., 2002, September 2003.
Pechan, 2005. E.H. Pechan & Associates, Inc. Profile of the Fishing Industry in the United
States, Final. Prepared for Lisa Conner, U.S. Environmental Protection Agency, Air
4-49
-------
Quality Strategies and Standards Division, Innovative Strategies and Economics Group,
Research Triangle Park, NC, March, 2005.
RBFF, 2003. Recreational Boating and Fishing Foundation. 2003 Boating and Fishing Attitude,
Segmentation Study, 2003.
USFWS, 2002. U.S. Fish and Wildlife Service, 2001 National Survey of Fishing, Hunting, and
Wildlife-Associated Recreation, 2001.
4-50
-------
SECTION 5 MERCURY CONCENTRATIONS IN FISH M
5.1 Methylmercury Concentrations in Saltwater Fish Species 5-1
5.2 Methylmercury in Freshwater Fish Species 5-2
5.2.1 Sources of Variability in Hg within the NLFA 5-4
5.2.2 National Lake Fish Tissue Study 5-6
5.3 Comparison of the Differences in the NLFA and the NLFTS 5-7
5.4 Combining the NLFA and NLFTS Data 5-9
5.5 Normalization of Hg Fish Tissue Concentration Data 5-10
5.6 Resulting Fish Tissue Concentrations 5-13
5.7 Summary 5-17
5.8 References 5-17
Tables
Table 5-1. Concentrations of Mercury in Marine Life 5-2
Table 5-2. Number of Fish Tissue Samples / Watershed From the National Listing of Fish
Advisories 5-3
Table 5-3. Hg Fish Tissue Concentrations From Various Environments (ppm) 5-5
Table 5-4. Statistical Distribution of Normalized Hg Fish Tissue Concentrations 5-12
Table 5-5. Statistical Distribution of Non-Normalized Hg Fish Tissue Concentrations Shown in
Figure 5-5 5-13
Figures
Figure 5-1. NLFA Sample Locations 5-3
Figure 5-2. Frequency Distribution of Average Watershed Fish Tissue Concentrations (ppm) 5-4
Figure 5-3. Frequency and Average Concentrations of Various NLFA Sample Methods (Cuts of
Fish) 5^4
Figure 5-4. Sample Locations from the NLFTS 5-6
Figure 5-5. Cumulative Distribution Functions (CDFs) for Normalized NLFTS and NLFA Lake
Data 5-8
Figure 5-6. Total Area/Sample in the Combined NLFA and NLFTS Data Set 5-9
Figure 5-7. Locations Where Normalized Fish Tissue Concentrations are Utilized, and Where
Non-Normalized Data are Utilized 5-12
Figure 5-8. Baseline Average Hg Fish Tissue Concentrations 5-13
Figure 5-9. Statistical Distribution of Averages of Hg. Fish Tissue Concentrations in ppm . 5-14
Figure 5-10. Number of Unique Sampling Events Within Each HUC 5-15
Figure 5-11. Average Fish Tissue Concentrations By HUC 5-16
Figure 5-12. Frequency Distribution of HUC Averaged Concentrations (ppm) 5-16
-------
SECTION 5
MERCURY CONCENTRATIONS IN FISH
Because fish consumption is a major pathway for human exposure to methylmercury, it is
important to determine the methylmercury concentrations of consumable fish. A baseline
concentration of consumable fish will enable EPA to determine the reduction in exposure related
to a reduction in Hg in fish tissue resulting from reduction in air deposition of Hg.
As is discussed in Section 4, the fish that Americans consume come from a variety of
sources including saltwater species from the ocean and estuaries, and freshwater species from
lakes, rivers, and streams. Fish may be purchased from commercial sources, or caught
recreationally and consumed. In this section, we describe the level of methylmercury
contamination in shellfish (saltwater species) and finfish (both saltwater and freshwater species).
5.1 Methylmercury Concentrations in Saltwater Fish Species
The EPA's Office of Water has developed the Mercury in Marine Life database, which
provides information on the level of methylmercury contamination in estuarine and marine
species (i.e., commercial and non-commercial seafood)1. The Mercury in Marine Life database
contains over 15,000 records on methylmercury tissue concentrations in approximately 250
different fish and shellfish species. The geographic coverage includes data from all 24 coastal
states, the District of Columbia and Puerto Rico. Data was not evenly distributed by coast.
More tissue samples were taken from the Gulf of Mexico than either the Atlantic or Pacific
Ocean.
The average methylmercury concentration (i.e., mean in ppm) for several species in the
database are presented in Table 5-1. Not surprisingly, many of the species groups with high
methylmercury concentrations are top-level predators. This was expected considering methyl
mercury's tendency to bioaccumulate up the food chain. Twenty-six percent of the samples
contained total mercury concentrations above 0.3 ppm. Five percent of the samples contained
total methylmercury tissue concentrations above 1.0 ppm, the FDA action level for issuance of a
fish advisory. King mackerel and a number of shark species contain the highest means. Other
species with relatively high concentrations include barracuda, jack crevalle, Spanish mackerel,
ladyfish, and seatrout.
1 As discussed in ch. 3, samples are Hg, which are used as a proxy for MeHg.
5-1
-------
Table 5-1. Concentrations of Mercury in Marine Life
Species Group
Marlins
Mackerels
Sharks
Barracudas
Tunas
Jacks
Ladyfish
Groupers
Bluefish
Snook
Drums
Sea basses
Rockfish
Catfishes
Wrasses
Temperate basses
Crabs
Eels
Snappers
Sturgeons
Rays
Tripletails
Porgies
Grunts
Kingfish
Lobsters
Flounders
Dolphin Fish
Croakers
Oysters
Mullets
Salmon
Mussels
Shrimp
Clams
Cods
Herring
Surfperches
Count
11
849
620
54
34
486
194
229
289
496
2544
57
315
385
54
625
385
133
380
10
104
110
292
67
86
18
1006
44
425
2212
203
300
1157
171
51
150
186
391
Mean (ppm)
2.424
0.791
0.776
0.757
0.647
0.509
0.479
0.473
0.445
0.398
0.382
0.352
0.294
0.245
0.234
0.226
0.224
0.223
0.218
0.216
0.197
0.192
0.178
0.177
0.167
0.127
0.117
0.107
0.098
0.073
0.065
0.065
0.059
0.051
0.049
0.045
0.043
0.041
Range (ppm)
0.270 - 6.80
0.013-4.47
0.020 - 6.90
0.076.- 3. 10
0.071-1.57
0.017 - 3.90
0.020 - 2.60
0.045 - 3.30
0.020 - 2.00
0.030 - 2.08
0.001 - 6.62
0.020-1.32
0.004-1.44
0.000-1.80
0.070 - 0.75
0.000-1.25
0.001 - 3.68
0.000 - 0.80
0.004 - 2.80
0.120-0.35
0.013-0.91
0.014-1.28
0.001 - 1.73
0.020 - 0.66
0.017 - 0.78
0.050 - 0.25
0.001- 1.70
0.031-0.49
0.000-1.10
0.002-3.91
0.001 - 1.14
0.015-0.61
0.002 - 0.93
0.000-1.02
0.008-0.12
0.007 - 0.26
0.010 - 0.24
0.002 - 0.26
StDev
1.838
0.703
0.729
0.642
0.421
0.453
0.408
0.418
0.331
0.315
0.499
0.343
0.221
0.305
0.149
0.211
0.427
0.154
0.182
0.07
0.151
0.187
0.191
0.116
0.147
0.067
0.152
0.106
0.124
0.121
0.121
0.051
0.077
0.094
0.022
0.045
0.038
0.039
Source: U.S. EPA, Office of Water; Mercury in Marine Life Database.
5.2 Methylmercury in Freshwater Fish Species
The most extensive source of national-level monitored mercury data for freshwater fish is
the National Listing of Fish and Wildlife Advisories (NLFA) maintained by EPA's Office of
Water. The NLFA includes more than 91,500 samples offish tissue contaminant data collected
by states, Native American tribal governments, territories, and Canada (and submitted to EPA)
from over 10,700 locations nationwide from 1967 through 2003. Figure 5-1 shows the
continental locations of Hg fish tissue samples recorded in the NLFA to date. In some
5-2
-------
watersheds2, freshwater fish were sampled hundreds or even thousands of times, but most areas
have a handful of samples. Watersheds without samples are typically found in the West.
Table 5-2 shows the frequency at which watersheds were sampled.
Sample Locations
Figure 5-1. NLFA Sample Locations
Table 5-2. Number of Fish Tissue Samples / Watershed From the National Listing of Fish
Advisories
Number of Samples
No samples
1-4 samples
5-49 samples
50 - 99 samples
100 - 499 samples
500 - 999 samples
1,000 or more samples
Number of Watersheds
1069
237
547
126
157
13
3
The average watershed Hg fish tissue concentration is .29 ppm., and samples within a
watershed are typically within .81 ppm. of each other. Between watersheds, average watershed
concentrations range from .001 ppm. to over 4 ppm. The frequency distribution shown in Figure
5-2 illustrates the average concentration found in watersheds.
2 A watershed is defined as a USGS 8-digit HUC. HUCs are discussed later in this chapter.
5-3
-------
Frequency Distribution
0.0 0.6 1.2 1.7 2.3 2.9 3.5 4.0 4.B
Figure 5-2. Frequency Distribution of Average Watershed Fish Tissue Concentrations
(ppm)
In the NLFA dataset, each sample is described according to the sample location, sample
date, measured methylmercury concentration, species and size offish, and the part of the fish
sampled. Each of these elements in the data result in large variation across the NLFA samples.
5.2.1 Sources of Variability in Hg within the NLFA
Variation Across Sample Methods
Different states have used various sampling methods (cuts offish) over the years to
determine fish tissue concentrations. Figure 5-3 shows the frequency various sampling methods
were employed in the NLFA, and the average fish tissue concentration of Hg associated with
each sampling method.
>.t«..;
to.**.**;
«-,— ;
^-^^.-a^r^W
a=~i.^^ ,**-»* aa, tPluSK)
•feint*
"•***""-"-»"•.
»«.*»... «-p-~;
•»:
Ilkt.ktxm.l..*.!*,."
"•— ".
^jb*. k»
c"~'
.^o— ^m^n!ir.n^^
1^*1
0
_....'....„,!
=3
n
l
^x
^Hn
•OO 1OJ
Fr
BOO 13 J
iquvn
aoo ao^
ey San
-
»»»<
aplod
-.
X» XOJ
-
XX> 331
•
•
'
100 OJ
czBA
L^^.m.nn.--.
I-FT1
Si—
1'^.U^I ,
D
E3
P
1
13
S3
23SI
1
I
a
i
DO O.
A~vwra
90 1,
^o Cone
DO 1-
entralioc
SO 3J
O>pn»0
CD
Figure 5-3. Frequency and Average Concentrations of Various NLFA Sample Methods
(Cuts of Fish)
5-4
-------
Variation Across Species
There are close to 400 different species offish sampled in the NLFA. Average
concentrations for freshwater species, where samples were geocoded, range from .007 to 1.85
ppm The mean average concentration for a given species (freshwater or saltwater) is .25, with a
standard deviation of .43.
Variation Across Lake/River/Other Types of Ecosystems
About 70% of the samples have geographic coordinates associated with them. Where
samples were geocoded (i.e. a latitude and longitude were recorded based on a written site
description), it was possible to map the sample location. Mapping sample locations allowed the
EPA to classify locations as either lake or river. Locations closer to known non-flowing
waterbodies were assigned as type "lake", and locations closer to known flowing waterbodies
were assigned as type "river". Locations that did not have geographic coordinates associated
with them were mostly saltwater fish species, but are considered unclassified. On average, river
environments were .017 ppm higher in fish tissue concentrations than lake environments,
however river environments typically contained more variability in concentrations. Table 5-3
provides details related to fish tissue samples from unclassified, river and lake environments.
Table 5-3. Hg Fish Tissue Concentrations From Various Environments (ppm)
Type
unclassified
lakes
rivers
Number of Samples
40,965
17,623
33,048
Minima
m
0.00
0.00
0.00
Maximum
29.00
7.59
8.94
Average
0.3455
0.3438
0.3614
Standard Deviation
0.4827
0.3519
0.4386
Variance
0.2330
0.1238
0.1924
Other Potential Sources of Variability in NLFA Hg samples
Sometimes, the same location was sampled multiple times over the span of a few months
or even years. On average, most locations were sampled about 8 times, and same-location repeat
samples typically range within .4 ppm of each other.
Sampling efforts over time are not evenly spatially distributed. There is fairly evenly
distributed sampling efforts in the first half of the 1990's, but the late 1990's and early 2000's
show much heavier sampling in the north and southeast. There are some states that submitted
samples to the NLFA for some years, and not for others.
The NLFA Hg fish tissue concentrations represent total Hg concentration from all
sources. The state of South Carolina is of particular interest because concentrations are
relatively high. This could be due to the historic use of mercury in gold mining that took place
in the middle part of the state.
It is possible that Hg fish tissue concentrations are changing over time, but due to these
other sources of variability, and inconsistent spatial patterns of sampling, it is not possible to
determine if indeed these temporal trends are present, or the strength of their influence.
5-5
-------
5.2.2 National Lake Fish Tissue Study
In addition to the NLFA, EPA's National Lake Fish Tissue Survey (NLFTS) also
provides useful data on concentrations of Hg in fish tissue. Conducted in 1999-2003, this study
sampled fish tissue from 500 randomly selected lakes and reservoirs across the U.S. (from the
estimated 270,000 lakes and reservoirs in the lower 48 States). Figure 5-4 shows locations of
NLFTS sample sites.
• Sample Locations
Figure 5-4. Sample Locations from the NLFTS
In comparison to the NLFA, the design of the NLFTS provides more evenly spatially
distributed sampling sites across the United States, but are much fewer in number than the
NLFA. The NLFTS was statistically designed to be representative and sampled 54 different
species offish using same-species composite sample methods3. For each lake sampled, one
composite sample of a predatory species, and one of a bottom-dwelling species were taken.
Some of the more commonly sampled species in the NLFTS are Largemouth Bass, Carp,
Catfish, Trout, and Walleye. The average length of all the species used in each same-species
composite sample is also recorded. Samples were collected from mid-1999 through 2003. Hg
concentrations in the NLFTS range from .004 to 1.46 ppm The average measured NLFTS Hg
fish tissue concentration is .22 ppm
3 A same-species composite sample is where several fish of the same species are used together to determine the
methyl mercury concentration.
5-6
-------
5.3 Comparison of the Differences in the NLFA and the NLFTS
Of the two major sources of Hg fish tissue concentration data available (the NLFA and
NLFTS), each has it's strengths and weaknesses. The NLFA contains a large number of
samples, but could be biased given the purpose for which it was developed (fishing advisories).
The potential for bias arises because NLFA samples typically collected either at sites that are
known to be popular fishing locations, or because sites are suspected of having elevated levels of
Hg. The NLFTS is known to be unbiased (because sample locations were selected based on a
stratified random sample), but has comparatively fewer samples from fewer locations.
Therefore, it is important to compare the two datasets to determine the presence or absence and
significance of the suspected upward NLFA bias
Our initial data investigations of the NLFA and NLFTS indicated that fish tissue
concentrations varied according to the species sampled, length of the fish sampled, and the
sample method. A straight comparison of the NLFA and NLFTS would be inappropriate given
these other sources of variability in fish tissue concentration. To control for these sources of
variability for a comparison of the data sets, both of the data sets were normalized using the
National Descriptive Model of Mercury and Fish (NDMMF), which is discussed in detail in
section 5.5. The NDMMF is a statistical model that normalizes Hg fish tissue concentration data
to control for species/size/sample method variability4. Once the data has been normalized to
account for these factors, the outputs of the model are then compared by source of data input
(NLFA vs. NLFTS)
The EPA's Office of Water conducted an analysis to determine if the NLFA and NLFTS
provide substantially different estimates offish tissue concentrations. The purpose of the
comparison study was to determine if there is a visible and/or statistically significant bias in the
NLFA data relative to that in the NLFTS.
Since concentrations of methyl mercury in fish may be different in rivers than in lakes,
and the NLFTS study is just of lakes, the lake subset of NLFA normalized data was selected for
use in the comparison. Where multiple regression estimates were available for the same lake, the
normalized concentrations were averaged. Cumulative Distribution Function (CDF) estimates
were generated using the statistical software R v. 1.9.1 (Venables and Smith, 2004) with
Probability Survey Data Analysis Functions (psurvey.analysis v. 2.2) (Kincaid, 2004) module.
CDF estimates were generated for both the NLFA and NLFTS. Stratified random sample weight
adjustments were incorporated in the CDF estimates for the NLFTS data.
4 The NDMMF is a statistical normalization technique that is specifically designed for Hg fish tissue concentrations.
In this sense, the NDMMF is a non-traditional statistical normalization technique.
5-7
-------
Figure 5-5 shows the CDF comparison plot for the NLFA and NLFTS normalized lake
data sets as continuous functions by fish tissue methyl mercury concentrations, with 95*
percentile limits for the NLFTS5. A surprising result that is apparent in the figure, is that there is
little difference between the two distributions for lake fish. However, the upper end of the
distribution (e.g. 95th percentile) shows an upward bias (of 0.14 ppm) in NLFA with respect to
NLFTS. Differences at other points in the distribution (5th, 25*, 50*, and 75* percentiles) are
quite small (range from -0.017 ppm to 0.025 ppm). Four statistical CDF comparison tests were
performed on the two data sets. All four tests indicate the two CDFs can not be considered
statistically different at the 95% level (p values > 0.17). A z-test for difference in the means of
the two data sets indicated they are not different, at the 95% level.
10 20
50 100 200
Mercury Concentration (ppb)
500
1000
Figure 5-5. Cumulative Distribution Functions (CDFs) for Normalized NLFTS and NLFA
Lake Data.
The overall conclusion is that the two data sets, for NLFA and NLFTS lakes, are not
statistically different though there is a clear upward bias at the very upper end of the NLFA
concentration distribution.
5 Since calculations of the 95* percentile confidence limits on the CDF estimates are dependent on an assumption of
simple random sampling, which does not hold for the NLFA data, the accuracy of the confidence limits are affected
and thus are not presented for the NLFA data.
5-8
-------
5.4 Combining the NLFA and NLFTS Data
The lack of statistical difference in the distributions offish tissue concentrations (except
in the highest 95th percentile) in the NLFA supports its use along with the NLFTS data for a
benefits analysis for this rule.
When the two data sets (NLFA and NLFTS) are combined, on average, there is one
sample location /1,000 sq. km. The sampling network is more-dense in the East and on the
West Coast, and less dense in the Midwest. Figure 5-6. illustrates the variable spatial density of
the combined NLFA and NLFTS sample locations.
# sq. km/sample
• Less than 1,000
• 1,000-2,000
m 2,000 -10,000
1 10,000 - 100,000
100,000 - 126,290
Figure 5-6. Total Area/Sample in the Combined NLFA and NLFTS Data Set
An examination of the combined NLFA and NLFTS fish tissue concentrations reveals
further data problems that need to be addressed. In the combined data set, there are over 92,224
records of concentrations sampled in the late 1960's through 2003, and 35 different sampling
methods were used to obtain Hg concentrations from over 450 different species offish that range
in length from .01 to 379 in. (over 30 ft., and an obvious data entry error). Samples were
obtained from over 11,000 different locations across the U.S., and concentrations range from
below detection limits to over 10 ppm
To eliminate errors within the data due to data entry and focus benefit analysis efforts,
the following criteria were applied to the data that were selected for further analysis:
Sampling date must be later than 1990.
• The sample must be geo-referenced.
5-9
-------
• The sample length must have a recorded length67 that is reasonable when
compared to world record lengths.8
• All Hg concentration units that were recorded in .ppb were converted to .ppm to
maintain consistency with the majority of the samples.
• Only freshwater species samples are used.
5.5 Normalization of Hg Fish Tissue Concentration Data
Hg fish tissue concentration variability due to species/size/sample method can be seen at
a global database scale in averages reported previously in this report, and on an individual
sample level. For example, at Abbots Creek in North Carolina, at least 20 largemouth bass were
sampled using the same sample method on the same day in 1992. The lengths of these fish
ranged from 9 to 19 inches, and fish tissue concentrations ranged from .24 ppm (9 inch fish), to
1.4 ppm (19 inch fish). In addition, a 12.41 in black crappie and a 12.41 in. largemouth bass
were sampled at Abbots Creek. The concentration of the black crappie was .34 ppm, while the
concentration of the largemouth bass was .19 ppm An example of variability potentially
introduced by sample method can also be found from the Abbots Creek samples where two 8 in.
goldfish were sampled four months apart. One was sampled using the whole fish, and one was
sampled using a fillet. The whole fish sample was .32 ppm and the fillet sample was .96 ppm
Due to these sources of variability, a computation of the simple average of existing fish
tissue concentrations at a location would not provide a representative estimate of Hg in fish for
use in a benefits analysis. Averaging would include many species that are not typically targeted
by anglers, and very small fish which would misrepresent Hg fish tissue concentrations
consumed by the public. Because of this, we considered focusing on key fish species that are
frequently targeted by anglers and consumable sizes representative of sizes typically found in
lakes and rivers.
We considered using a subset of the combined NLFA and NLFTS consisting of these
species larger that a minimum length, but we found that this would reduce the already sparse
data set. Based on size alone, the removal of sampled fish less than 6 inches long would reduce
the data set by about 20%. Controlling for species offish that are typically targeted by anglers
would reduce the data set even further.
6 Where a fish weight was recorded, the length was predicted using a regression for each sampled fish species of the
log of the length as a dependent variable, and the weight of a fish as the independent variable. Typical residual error
is approximately 10% across all species.
7 It is of interest to note that almost all of the samples from the states of IA, NE, OH, TN, VA, and PA, were
removed because the length of the fish was not recorded.
8 Must not be more than 10% longer than a recorded world record length for that fish species. An additional 10%
over current world records was allowed so that only obvious data entry problems are filtered out of the data set. A
10% threshold over the world record represents, for this study, fish lengths considered outside the realm of
possibility. The recorded length must also be less than 108 inches (108 in. is 10% greater than the length of the
largest world record of any freshwater species). Background research on world record lengths for each species was
extremely time consuming. To focus QA/QC efforts, only species that were sampled at least 30 times in the U.S.
were investigated.
5-10
-------
To take full advantage of the Hg fish tissue concentration samples within the NLFA and
NLFTS, it is important to control for variability due to species/fish size/sample method. Using
information on species, size, sample method, location, and date of sampling event enables the
use of statistical procedures to estimate what Hg fish tissue concentration would have been, had
a different species/size/sample method been used to collect samples and detect the Hg
concentrations.
The USGS developed a procedure called the National Descriptive Model of Mercury and
Fish (NDMMF) (Wente 2004). The NDMMF is a statistical model related to covariance. It is
calibrated using the NLFA and NLFTS data. The model is designed to allow the prediction of
different species, cuts, and lengths offish for sampling events, even when those
species/lengths/cuts of fish were not sampled during those sampling events.
The idea is to model methyl mercury concentration as a power function offish length,
i.e., y = axb, where y = methyl mercury concentration, x = length, and a,b are parameters.
Potentially, that could be done for each species at each site, but the data are far too sparse for
that, and, further, we wanted to devise some way to predict mercury concentrations for a species
that was not even collected at some given site. So, the assumptions are made that: (i) a
universal, national value of "b" for each tissue type from each fish species, and (ii) the value of
"a" at a given site at a given time is the same for all species. Thus a prediction of mercury
concentration for a given tissue from a fish of an arbitrary species of length x could then be
predicted by "looking up" the right value of "b" for that species and tissue and the value of "a"
for that site and time.
To implement the NDMMF, the NLFA and NLFTS fish tissue concentrations are further
prepared for analysis. If the method of sample collection was "fillet skin on scales off or
"composite fillet skin on", these were assigned to the "fillet skin on" sample method type.
Various methods of "Fillet skin off sample types were treated in a similar manner.
"Composite" types were treated as "whole" sample types. Sample types that were "tissue
carcass, crab, meat, dorsal muscle plug with skin, dorsal muscle plug without skin ,edible
portion, eggs, gills, lipid, liver, organism without head or viscera, shellfish, tissue, turtle
muscle/somatic , unknown, or viscera" were deleted.
Those species sampled 29 or fewer times were excluded from the analysis due to the lack
of quality assurance background information collected for these samples (see section 5.4
description of data criteria). In total, the steps discussed in section 5.4 to remove data entry
errors and to select relevant data for benefits analysis leave a total of 42,756 remaining samples
from the original population of 92,224 samples.
The remaining samples are log-transformed to model the relationship as log y = b * log x
+ log a, 1 is added to x and to y to escape the problem of taking the log of zero. The SAS
program LIFEREG is then used to estimate the parameters. More detail about the NDMMF can
be found at http://pubs.water.usgs.gov/sir20045199/.
The NDMMF was used to generate estimates of fish tissue concentrations for six species
that are both commonly targeted by freshwater fish anglers and were frequently sampled in the
NLFA and NLFTS data bases (largemouth bass, catfish, brown trout, white crappie, white perch,
5-11
-------
and walleye). The target sizes selected for each species is representative of the typical adult size
of that fish species found in the wild (Schultz 2004).
For sampling events where samples were removed during data preparation, and the target
species were sampled, the sampled target species Hg concentrations were averaged for every
unique sampling event.
This means that for some sample locations, we cannot control for variability resulting
from the size of the sample fish through the normalization process. In these cases, we retain the
raw data from the NLFA and NLFTS. Figure 5-7 shows locations where it was possible to
compute NDMMF estimates, and where raw data was used. Table 5-4 gives the average, max,
min, and standard deviation of Hg fish tissue concentrations of the normalized data using the
NDMMF. Table 5-5 below gives the average, max., min., and standard deviation of the non-
normalized fish tissue concentrations used for this analysis.
NDMMF estimated
concentrations
Raw data concentrations
Figure 5-7. Locations Where Normalized Fish Tissue Concentrations are Utilized, and
Where Non-Normalized Data are Utilized
Table 5-4. Statistical Distribution of Normalized Hg Fish Tissue Concentrations
SPECIES
Largemouth Bass
Catfish
White Crappie
White Perch
Brown Trout
Walleye
Average
.31
.23
.14
.27
.11
.42
Maximum
4.08
2.98
1.89
3.45
1.47
5.47
Minimum
0
0
0
0
0
0
Standard Deviation
.36
.26
.17
.31
.13
.48
5-12
-------
Table 5-5. Statistical Distribution of Non-Normalized Hg Fish Tissue Concentrations
Shown in Figure 5-5
SPECIES
Largemouth Bass
Catfish
White Crappie
White Perch
Brown Trout
Walleye
Average
.57
.14
.13
.10
.09
.66
Maximum
4.22
1.30
1.11
.84
.56
7.59
Minimum
0
0
0
0
0
0
Standard Deviation
.68
.13
.18
.15
.09
1.21
Number of Samples
558
556
60
50
56
55
5.6 Resulting Fish Tissue Concentrations
To develop a baseline of Hg fish tissue concentrations for the benefits analysis of this
rule, both the normalized and non-normalized data values shown in Tables 5-4 and 5-5 were
merged to form a single data set, and the average of all the fish tissue concentrations (either
normalized or non-normalized), by sampling event was computed. To do this we specify a
sampling location and date as an "event", and then calculate an average for each event. For
example, at event X, estimates were computed for largemouth bass, catfish, white crappie, white
perch, brown trout, and walleye. All six species estimates were then averaged at event X to
compute a single representative Hg fish tissue concentration for that event. For the raw data,
where multiple target species were sampled in a particular event, these were averaged.
Figure 5-8 shows the statistical distribution of averages. Figure 5-9 is a map of sample locations
and representative Hg fish tissue concentrations in ppm
Figure 5-8. Baseline Average Hg Fish Tissue Concentrations
-0.0 8.T W 22 2M 3.7 *.4 5.1 5.9 1.6 F.3
000
rm
Standard Deviation: 03D
5-13
-------
-0.001000-0,100000
0.100001-0,200000
. 0.200001-0.300000
. 0.300001-1.000000
i 1.000001-7.590000
f
f
Figure 5-9. Statistical Distribution of Averages of Hg. Fish Tissue Concentrations in ppm
The species - averaged concentrations at each event were then averaged for each unique
location. These averages were then averaged by US Geological Survey (USGS) 8-digit
watershed units called Hydrologic Unit Code (HUC) Classification Areas. Just like states can be
subdivided into counties, large watersheds can be subdivided into smaller and smaller
watersheds. For example, the Chesapeake Bay Watershed is composed of 104 small 8-digit
HUCs. The 8-digit HUC is the smallest USGS Hydrologic Unit Code (HUC) Classification.
Figure 5-10 geographically shows the USGS 8-digit HUCs and the frequency they were sampled
Approximately 35% of HUCs in the states east and south of North Dakota (excluding North
Dakota) were not sampled.
5-14
-------
Number of Unique
Location/Bttt Sampling
Ev«nt»
no
Cil-5
• 6-15
• 16-50
• 51-234
Figure 5-10. Number of Unique Sampling Events Within Each HUC
For the most part, higher concentrations averaged by HUC can be found in the
southeastern coastal plains of North Carolina, south and west in the coastal plains around to
Mississippi. Maine, New Hampshire, and some areas of New York, Pennsylvania and Ohio
show some higher concentrations. Mercury Maps has recorded the presence of gold mines in
South Carolina, chlor-alkali plants in Ohio, and mercury mines in Arkansas that may, in part,
explain their higher Hg fish tissue concentrations. Figure 5-11 shows a map of mercury fish
tissue concentrations averaged by HUC.
Average HUC concentration is .25 ppm The overwhelming majority of HUCs have
concentrations below 1 ppm Figure 5-12 shows a frequency distribution of average HUC
concentrations.
5-15
-------
Avenge Fish
Tissue Concentration
(.ppm)
less than.1
.1-.2
M.2-.3
• .3-1
• 1 - 2.7
* Where no samples fell within «HUC, the HUC is shown in while.
Figure 5-11. Average Fish Tissue Concentrations By HUC9
10
Frequency Distribution
-0.0 0.3 0.7 1.0 1.4 1.7 2.1 2.4
Figure 5-12. Frequency Distribution of HUC Averaged Concentrations (ppm)
9 226 8-digit HUC watersheds have average concentrations higher than the Office of Water Methylmercury water
quality criterion of .3 ppm
111 Where no samples fall within a HUC, the HUC is shown in white. This does not indicate that Hg is not in the fish,
but simply that data do not exist to estimate Hg concentration at this location.
5-16
-------
5.7 Summary
The data of mercury concentrations in flnfish and shellfish from freshwater and saltwater
sources indicate a wide variety of contamination levels in fish species. The data demonstrate the
finding in Section 3 of this report that larger predatory fish in the higher tropic levels tend to
have higher levels of methyl mercury contamination in fish tissue.
We obtained data for a variety offish species from the Mercury in Marine Life Database,
the National Listing of Fish Advisories (NLFA), and the National Lake Fish Tissue Study
(NLFTS). There are considerably more fish tissue samples from freshwater sources than for
saltwater sources. The mean concentrations for the saltwater fish species obtained from the
Mercury in Marine Life Database ranges from 0.04 ppm to 2.4 ppm, but some samples from
predatory fish such as marlin, mackerel, shark, and barracuda indicate levels as high as 6.9 ppm
Data for the freshwater fish species from the NLFA and NLFTS are used in the quantified
benefit analysis provided in Section 11 of this report. Because the concentration of mercury in
fish tissue can vary by species, the length of the fish, the sampling method (i.e., fillet, whole fish,
skin on/off), and by location, EPA normalized the data from the NLFA and the NLFTS to
control for variability from factors other than location of the fish tissue sample. The resulting
normalized fish tissue samples were then average to the 8-digit HUC to provide a
characterization offish tissue concentrations in a watershed. The mean mercury concentrations
in the HUCs ranges from 0 ppm to 7.59 ppm, with an overall mean of 0.25 ppm
5.8 References
EPA. 1997. Mercury Study Report to Congress.
http://www.epa.gov/airprogm/oar/mercury.html
Kincaid, Thomas, 2004. User Guide for psurvey.analysis, version 2.4. Probability Survey Data
Analysis Functions. Available from:
http://www.epa.gov/nheerl/arm/analvsispages/techinfoanalysis.htm
Schultz, Ken, 2004. Field Guide to Freshwater Fish. John Wiley and Sons, Hoboken N.J.
Venables, W.N. and Smith, D.M., 2004. An Introduction to R: Notes on R: A Programming
Environment for Data Analysis and Graphics Version 1.9.1 (2004-06-21) Available from:
http://cran.us.r-project.org/
Wente, S.P., 2004, A Statistical Model and National Data Set for Partitioning Fish-Tissue
Mercury Concentration Variation between Spatiotemporal and Sample Characteristic
Effects: U.S. Geological Survey Scientific Investigations Report 2004-5199. Available
from: http://pubs.water.usgs.gov/sir20045199/
5-17
-------
SECTION 6 PROFILE OF THE UTILITY SECTOR 6J.
6.1 Power-Sector Overview 6-1
6.1.1 Generation 6-1
6.1.2 Transmission 6-3
6.1.3 Distribution 6-3
6.2 Deregulation and Restructuring 6-3
6.3 Pollution and EPA Regulation of Emissions 6-4
6.4 Pollution Control Technologies 6-5
6.5 Regulation of the Power Sector 6-6
6.6 Cap and Trade 6-7
6.7 Clean Air Interstate Rule 6-8
Table 6-1. Existing Electricity Generating Capacity by Energy Source, 2002 6-1
Table 6-2. Total U.S. Electric Power Industry Retail Sales in 2003 (Billion kWh) 6-2
Table 6-3. Electricity Net Generation in 2003 6-2
Table 6-4. Emissions of SO2 and NOX in 2003 and Percentage of Emissions in the CAIR
Affected Region (tons) 6-8
Table 6-5. Current Electricity Net Generation and EPA Projections for 2010 and 2015 6-9
Figure 6-1. Status of State Electricity Industry Restructuring Activities (as of February 2003)6-4
Figure 6-2. Emissions of Hg, SO2, and NOX from the Power Sector (2003) 6-5
-------
SECTION 6
PROFILE OF THE UTILITY SECTOR
This section discusses important aspects of the power sector as they relate to CAMR,
including the types of power-sector sources affected by CAMR, and provides background on the
power sector and EGUs. In addition, this section provides some historical background on EPA
regulation of and future projections for the power sector.
6.1 Power-Sector Overview
The functions of the power sector can be separated into three distinct operating activities:
generation, transmission, and distribution.
6.1.1 Generation
Electricity generation is the first process in the delivery of electricity to consumers. The
process of generating electricity, in most cases, involves creating heat to rotate turbines which, in
turn, create electricity. The power sector is comprised of over 16,000 generating units,
consisting of fossil-fuel fired units, nuclear units, and hydroelectric and renewable sources
dispersed throughout the country (see Table 6-1).
Table 6-1. Existing Electricity Generating Capacity by Energy Source, 2002
Energy Source
Coal
Petroleum
Natural Gas
Dual Fired
Other Gases
Nuclear
Hydroelectric
Other Renewables
Other
Total
Number of Generators
1,566
3,076
2,890
2,974
104
104
4,157
1,501
41
16,413
Generator Nameplate Capacity (MW)
338,199
43,206
194,968
180,174
2,210
104,933
96,343
18,797
756
979,585
Source: EIA
These electric-generating sources provide electricity for commercial, industrial, and
residential uses, each of which consumes roughly one-third of the total electricity produced (see
Table 6-2).
6-1
-------
Table 6-2. Total U.S. Electric Power Industry Retail Sales in 2003 (Billion kWh)
N %
Residential
Commercial
Industrial
Other
All Sectors
1,280
1,119
991
109
3,500
37%
32%
28%
3%
100%
Source: EIA
In 2003, electric-generating sources produced 3,848 billion kWh to meet electricity
demand. Roughly 70 percent of this electricity was produced through the combustion of fossil
fuels, primarily coal and natural gas, with coal accounting for more than half of the total (see
Table 6-3).
Table 6-3. Electricity Net Generation in 2003
(billion kWh)
N %
Coal
Petroleum
Natural Gas
Other Gases
Nuclear
Hydroelectric
Other
Total
1,970
118
629
11
764
275
81
3,848
51%
3%
16%
0.3%
20%
7%
2%
100%
Source: EIA
Note: Retail sales and net generation may not correspond exactly because net generation data may include net exported
electricity and loss of electricity.
Coal-fired generating units typically supply "base-load" electricity, which means these
units operate continuously throughout the day. Coal-fired generation, along with nuclear
generation, meet the part of electricity demand that is relatively constant. Gas-fired generation,
however, typically supplies "peak" power, when there is increased demand for electricity (e.g.,
when businesses operate throughout the day or when people return home from work and run
appliances and heating/air-conditioning, versus late at night or very early morning when demand
for electricity is reduced).
6-2
-------
6.1.2 Transmission
Transmission is the term used to describe the movement of electricity, through use of
high voltage lines, from electric generators to substations where power is stepped down for local
distribution. Transmission systems have been traditionally characterized as a collection of
independently operated networks or grids interconnected by bulk transmission interfaces.
Within a well-defined service territory, the regulated utility has historically had
responsibility for all aspects of developing, maintaining, and operating transmission of
electricity. These responsibilities typically included system planning and expanding,
maintaining power quality and stability, and responding to failures.
6.1.3 Distribution
Distribution of electricity involves networks of smaller wires and substations that take the
higher voltage from the transmission system and step it down to lower levels to match the needs
of customers. The transmission and distribution system is the classic example of a natural
monopoly because it is not practical to have more than one set of lines running from the
electricity-generating sources to neighborhoods or from the curb to the house.
Transmission and distribution have been considered differently than generation in current
efforts to restructure the industry. Transmission has generally been developed by the larger
vertically integrated utilities that typically operate generation and distribution networks.
Distribution is handled by a large number of utilities that often only sell electricity. Electricity
restructuring has focused primarily on converting the industry to fully compete the sale of
electricity production or generation and not the transmission or distribution of electricity. The
restructuring of the industry is, in large part, the separating of generation assets from the
transmission and distribution assets into separate economic entities in many state efforts.
Transmissions and distribution remain price regulated throughout the country based on the cost
of service.
6.2 Deregulation and Restructuring
The ongoing process of deregulation of wholesale and retail electric markets is changing
the structure of the electric power industry. In addition to reorganizing asset management
between companies, deregulation is aimed at the functional unbundling of generation,
transmission, distribution, and ancillary services the power sector has historically provided to
competitors in the generation segment of the industry.
Beginning in the 1970s, government policy shifted against traditional regulatory
approaches and in favor of deregulation for many important industries, including transportation,
communications, and energy, which were all thought to be natural monopolies (prior to 1970)
that warranted governmental control of pricing. Some of the primary drivers for deregulation of
electric power included the desire for more efficient investment choices, the possibility of lower
electric rates, reduced costs of combustion turbine technology that opened the door for more
companies to sell power, and complexity of monitoring utilities' cost of service and establishing
cost-based rates for various customer classes (see Figure 6-1). The pace of restructuring in the
6-3
-------
electric power industry slowed significantly in response to market volatility and financial turmoil
associated with bankruptcy filings of key energy companies in California. By the end of 2001,
restructuring had either been delayed or suspended in eight states that previously enacted
legislation or issued regulatory orders for its implementation. Another 18 other states that had
seriously explored the possibility of deregulation in 2000 reported no legislative or regulatory
activity in 2001 (DOE, EIA, 2003a). Currently, there are 17 states where price deregulation of
generation (restructuring) has occurred. The effort is more or less at a standstill; however, at the
federal level, there are efforts in the form of proposed legislation and proposed Federal Energy
Regulatory Commission (FERC) actions aimed at reviving restructuring. For states that have not
begun restructuring efforts, it is unclear when and at what pace these efforts will proceed.
Figure 6-1. Status of State Electricity Industry Restructuring Activities (as of February
2003)
6.3 Pollution and EPA Regulation of Emissions
The burning of fossil fuels, which generates about 70 percent of our electricity
nationwide, results in air emissions of Hg, SO2 and NOX, important precursors in the formation
of fine particles and ozone (NOX only). The power sector is a major contributor of these three
pollutants, and reductions of SO2 and NOX emissions are critical to EPA's efforts to bring about
attainment with the fine particle and ozone NAAQS through programs like CAIR, and critical to
6-4
-------
EPA's effort under CAMR because control of SO2 and NOX also can result in Hg reduction. In
2003, the power sector accounted for 40% of total nationwide Hg emissions, 67 percent of total
nationwide SO2 emissions and 22 percent of total nationwide NOX emissions (see Figure 6-2).
Figure 6-2. Emissions of Hg, SO2, and NOX from the Power Sector (2003)
Mercury Sulfur Dioxide Nitrogen Oxides
Pomrffl ^k
*
Mercury emissions from the power sector come mainly from coal-fired units, and CAMR
represents the first time Hg emissions from these units will be regulated. Mercury emissions can
vary by coal type, especially for units with existing PM, NOX, and SO2 controls. In general,
given the different properties of coal, these existing controls are best able to capture Hg from
bituminous coals, with less capture from subbituminous and lignite coals.
6.4 Pollution Control Technologies
There are two primary options for reducing SO2 emissions from coal-burning power
plants. Units may switch from higher to lower sulfur coal, or they may use flue gas
desulfurization (FGD, commonly referred to as scrubbers). According to data submitted to EPA
for compliance with the Title IV Acid Rain Program, the SO2 emission rates for coal-fired units
varied from under 0.4 Ibs/mmBtu to over 5 Ibs/mmBtu depending on the type of coal combusted.
It is generally easier to switch to a coal within the same rank (e.g., bituminous or
sub-bituminous) because these coals will have similar heat contents and other characteristics.
Switching completely to sub-bituminous coal (which typically has a lower sulfur content) from
bituminous coal is likely to require some modifications to the unit. Limited blending of
sub-bituminous coal with bituminous coal can often be done with much more limited
modifications.
The two most commonly used scrubber types include wet scrubbers and spray dryers.
Wet scrubbers can use a variety of sorbents to capture SO2 including limestone and magnesium
enhanced lime. The choice of sorbent can affect the performance, size, and capital and operating
costs of the scrubber. New wet scrubbers typically achieve at least 95 percent SO2 removal.
Spray dryers can achieve over 90 percent SO2 removal.
6-5
-------
One method of reducing NOX emissions is through the use of combustion controls (such
as low NOX burners and over-fired air). Combustion controls reduce NOX by ensuring that the
combustion of coal occurs under conditions that form less NOX. Post-combustion controls
reduce NOX by removing the NOX after it has been formed. The most common post-combustion
control is SCR. SCR systems inject ammonia (NH3), which combines with the NOX in the flue
gas, to form nitrogen and water, and uses a catalyst to enhance the reaction. These systems can
reduce NOX by 90 percent and achieve emission rates of around 0.06 Ibs NCVmmBtu. Selective
noncatalytic reduction also removes NOX by injecting ammonia, but no catalyst is used. These
systems can reduce NOX by up to 40 percent.
Hg capture can occur through existing controls and through Hg-specific control
technologies. Many power plants have existing mercury capture as a co-benefit of air pollution
control technologies for NOX, SO2 and particulate matter (PM). This includes capture of
particulate-bound mercury in PM control equipment and capture of soluble ionic Hg in wet flue
gas desulfurization (FGD) systems. Additional data have also shown that the use of SCR for NOX
control enhances oxidation of Hg to the soluble ionic form, resulting in increased removal in the
wet FGD system for units burning bituminous coal. The range of Hg removal depends on the
control configuration and the coal type burned, and can vary between 0 and 98 percent. (For
further discussion see Control of Emissions from Coal-Fired Electric Utility Boilers: An Update,
EPA/Office of Research and Development, March 2005, in docket.)
Mercury-specific controls, most notably activated carbon injection (ACI), are used on
municipal waste combustor (MWC) and medical waste incinerator (MWI) facilities in the U.S.
and Europe. At present, ACI is the most widely studied of the mercury-specific control
technologies for coal-fired power plants and shows the potential to achieve moderate-to-high
levels of mercury control. EPA's modeling provides ACI as a compliance choice and assumes a
90% removal with the addition of a fabric filter PM control device.
For more detail on the cost and performance assumptions of pollution controls, see the
documentation for the Integrated Planning Model (IPM), a dynamic linear programming model
that EPA uses to examine air pollution control policies for Hg, SO2 and NOX throughout the
contiguous United States for the entire power system. Documentation for IPM can be found at
www.epa.gov/airmarkets/epa-ipm.
6.5 Regulation of the Power Sector
At the federal level, efforts to reduce emissions of SO2 and NOX have been occurring
since 1970. Policy makers have recognized the need to address these harmful emissions, and
incremental steps have been taken to ensure that the country meets air quality standards.
Federal regulation of SO2 and NOX emissions at power plants began with the 1970 Clean
Air Act. The Act required the Agency to develop performance standards for a number of source
categories including coal-fired power plants. The first New Source Performance Standards
(NSPS) for power plants (subpart D) required new units to limit SO2 emissions either by using
scrubbers or by using low sulfur coal. NOX was required to be limited through the use of low
NOX burners. A new NSPS (subpart Da), promulgated in 1978, tightened the standards for SO2
requiring scrubbers on all new units.
6-6
-------
The 1990 Clean Air Act Amendments (CAAA) placed a number of new requirements on
power plants. The Acid Rain Program, established under Title IV of the 1990 CAAA, requires
major reductions of SO2 and NOX emissions. The SO2 program sets a permanent cap on the total
amount of SO2 that can be emitted by electric power plants in the contiguous United States at
about one-half of the amount of SO2 these sources emitted in 1980. Using a market-based cap-
and-trade mechanism allows flexibility for individual combustion units to select their own
methods of compliance. The program uses a more traditional approach to NOX emission
limitations for certain coal-fired electric utility boilers, with the objective of achieving a 2
million ton reduction from projected NOX emission levels that would have been emitted in 2000
without implementation of Title IV.
The Acid Rain Program comprises two phases for SO2 and NOX. Phase I applied
primarily to the largest coal-fired electric generation sources from 1995 through 1999 for SO2
and from 1996 through 1999 for NOX. Phase II for both pollutants began in 2000. For SO2, the
Acid Rain Program applies to thousands of combustion units generating electricity nationwide;
for NOX it generally applies to affected units that burned coal during 1990 through 1995. The
Acid Rain Program has led to the installation of a number of scrubbers on existing coal-fired
units as well as significant fuel switching to lower sulfur coals. Under the NOX provisions of
Title IV, most existing coal-fired units were required to install low NOX burners.
The CAAA also placed much greater emphasis on control of NOX to reduce ozone
nonattainment. This has led to the formation of several regional NOX trading programs as well as
an intrastate NOX trading program in Texas. The Ozone Transport Commission (a group of
northeast states) created an interstate NOX trading program that began in 1999. In 1998, EPA
promulgated regulations (the NOX SIP Call) that required 21 states in the eastern United States
and the District of Columbia to reduce NOX emissions that contributed to nonattainment in
downwind states using the cap-and-trade approach. This program began in the summer of 2004
and has resulted in the installation of significant amounts of selective catalytic reduction.
In addition to federal programs to reduce emissions of SO2 and NOX, several states have
also taken action. Several states, like North Carolina, New York, Connecticut, and
Massachusetts, have moved to control these emissions to address nonattainment. To date, there
have not been any regulations on the utility sector to control mercury emissions.
6.6 Cap and Trade
The cap-and-trade system under CAMR, which is largely based on the Acid Rain Trading
Program and the NOX SIP Call, provides the power sector with considerable flexibility in
meeting the emission reduction requirements. Cap-and-trade regulation is an extremely efficient
tool that allows for environmental goals to be met in the most cost-effective manner, because
firms have economic incentives to achieve emissions reductions where they are cheapest. The
system allows for various compliance options, with each firm determining what option works
best given certain costs, such as fuel costs or costs of pollution controls.
In addition to the pollution control options discussed above, companies can comply with
cap-and-trade programs through more efficient use of the generating fleet to take advantage of
generating sources that emit less and run more efficiently, commonly referred to as dispatch
6-7
-------
changes. By shifting generation to these more efficient units, the power sector is reducing the
cost of compliance because there is a cost to pollute under a cap. Another option is purchasing
additional allowances to cover emissions.
6.7 Clean Air Interstate Rule
The CAIR is a new regulatory action that addresses air quality problems and improves
public health and the environment by substantially reducing emissions of SO2, NOX, and Hg.
The final CAIR requires annual SO2 and NOX reductions in 23 States and the District of
Columbia, and also requires ozone season NOX reductions in 25 States and the District of
Columbia. Many of the CAIR States are affected by both the annual SO2 and NOX reduction
requirements and the ozone season NOX requirements. CAIR allows affected states to adopt a
two-phased cap-and-trade program to meet emissions reduction requirements of roughly 73
percent for SO2 and 61 percent for NOX from 2003 levels.
The rule would affect roughly 3,000 fossil fuel-fired units with a nameplate capacity
greater than 25 MW. These sources accounted for roughly 90 percent of nationwide SO2
emissions and 78 percent of nationwide NOX emissions in 2003 (see Table 6-4).
Table 6-4. Emissions of SO2 and NOX in 2003 and Percentage of Emissions in the CAIR
Affected Region (tons)
SOj NO,
CAIR Region 9,501,201 3,251,980
Nationwide 10,595,069 4,165,026
CAIR Emissions as % of Nationwide Emissions 90% 78%
Source: EPA.
Note: Region includes the States of Alabama, Connecticut, District of Columbia, Florida, Georgia, Illinois, Indiana, Iowa,
Kentucky, Louisiana, Maryland, Massachusetts, Michigan, Minnesota, Mississippi, Missouri, New York, North Carolina,
Ohio, Pennsylvania, South Carolina, Tennessee, Texas, Virginia, West Virginia, and Wisconsin
6-8
-------
EPA modeling1 shows that coal-fired and oil/gas-fired generation will continue to play an
important part of the electricity generating portfolio in the United States. Electricity demand is
anticipated to grow by 1.6 percent a year, and total electricity demand is projected to be 4,198
billion kWh by 2010. Table 6-5 shows current electricity generation and projected levels in
2010 and 2015 using EPA modeling.
Table 6-5. Current Electricity Net Generation and EPA Projections for 2010 and 2015
(billion kWh)
2003 2010 2015
Coal
Oil/Gas
Other
1,970
758
1,119
2,198
111
1,223
2,242
1,026
1,235
Total 3,848 4,198 4,503
Source: 2003 data is from EIA. Projections are from the Integrated Planning Model run by EPA.
'EPA uses the IPM to make power-sector forecasts about emissions, costs, and other key factors of the power sector.
Industry projections presented here are from EPA's base case scenario. For more information about IPM, see
http://www.epa.gov/airmarkets/epa-iprn/index.html.
6-9
-------
CHAPTER 7 COST AND ENERGY IMPACTS 7J.
7.1 Modeling Background 7-1
7.2 Projected Hg Emissions 7-3
7.3 Projected SO2 and NOX Emissions 7-5
7.4 Projected Costs 7-6
7.5 Projected Control Technology Retrofits 7-8
7.6 Projected Generation Mix 7-9
7.7 Projected Capacity Additions 7-9
7.8 Projected Coal Production for the Electric Power Sector 7-10
7.9 Projected Retail Electricity Prices 7-11
7.10 Projected Fuel Price Impacts 7-13
7.11 Social Cost Calculations 7-13
7.12 Limitations of Analysis 7-14
7.13 Significant Energy Impact 7-18
7.14 Sensitivity Analysis on Assumptions for Hg Control Costs 7-18
7.15 Sensitivity Analysis on Assumptions for Natural Gas Prices and Electricity
Growth 7-23
7.16 Small Entity Impacts 7-28
7.16.1 Identification of Small Entities 7-30
7.16.2 Overview of Analysis and Results 7-31
7.16.3 Summary of Small Entity Impacts 7-36
7.17 Unfunded Mandates Reform Act (UMRA) Analysis 7-37
7.17.1 Identification of Government-Owned Entities 7-38
7.17.2 Overview of Analysis and Results 7-39
7.17.3 Summary of Government Entity Impacts 7-44
7.18 List of IPM Runs in Support of CAMR 7-45
Tables
Table 7-1. CAMR Options Annual Emissions Caps (Tons) 7-1
Table 7-2. CAIR Emissions Caps (Million Tons) 7-1
Table 7-3. Projected Emissions of Hg with the Old Base Case3, New Base Case, and with
CAMR Options (Tons) 7-5
Table 7-4. Projected Speciated Emissions of Hg in 2020 with New Base Case (CAIR) and
CAMR Options (Tons) 7-5
Table 7-5. Projected Emissions of SO2 with the Old Base Case8, New Base Case (CAIR), and
with CAMR Options (Million Tons) 7-6
Table 7-6. Projected Emissions of NOx with the Old Base Case8, New Base Case (CAIR), and
with CAMR Options (Million Tons) 7-6
Table 7-7. Annualized National Private Compliance Cost and Present Value Cost ($1999) .. 7-7
Table 7-8. Marginal Cost of Hg, SO2, and NOx Reductions with CAMR Options ($1999) .. 7-7
Table 7-9. Pollution Controls by Technology with the Old Base Case, New Base Case (CAIR),
and with CAMR Options (GW) 7-8
Table 7-10. Generation Mix with the Old Base Case, with New Base Case (CAIR), and with
CAMR Options (Thousand GWhs) 7-9
Table 7-11. Total Coal and Natural Oil/Gas-Fired Capacity by 2020 (GW) 7-10
Table 7-12. Coal Production for the Electric Power Sector with the Old Base Case, New Base
Case (CAIR), and with CAMR Options (Million Tons) 7-10
-------
Table 7-13. Projected National Retail Electricity Prices with the Old Base Case, New Base Case
(CAIR), and CAMR Options (Mills/kWh) ($1999) 7-11
Table 7-14. Retail Electricity Prices by NERC Region with the Old Base Case, New Base Case
(CAIR), and with CAMR Options (Mills/kWh) ($1999) 7-12
Table 7-15. Henry Hub Natural Gas Prices and Average Delivered Coal Prices with the Old
Base Case, New Base Case (CAIR), and with CAMR Options (1999$/mmBtu) 7-13
Table 7-16. Projected Emissions of Hg with New Base Case (CAIR) and CAMR, without and
with Selected Technological Advances (Tons) 7-19
Table 7-17. Projected Emissions of SO2 with New Base Case (CAIR) and CAMR, without and
with Selected Technological Advances (Million Tons) 7-19
Table 7-18. Projected Emissions of NOx with the New Base Case (CAIR) and CAMR without
and with Selected Technological Advances (Million Tons) 7-20
Table 7-19. Annualized Private Compliance Cost and Present Value Cost Incremental to the
New Base Case (CAIR) ($1999) 7-20
Table 7-20. Marginal Cost of Hg, SO2, and NOX Reductions with CAMR without and with
Selected Technological Advances ($1999) 7-20
Table 7-21. Pollution Controls by Technology with the New Base Case (CAIR), and CAMR
without and with Selected Technological Advances (GW) 7-21
Table 7-22. Generation Mix with the Old Base Case, the New Base Case (CAIR), and with
CAMR without and with Selected Technological Advances (Thousand GWhs) 7-21
Table 7-23. Total Coal and Natural Oil/Gas-Fired Capacity by 2020 (GW) 7-22
Table 7-24. Coal Production for the Electric Power Sector with the Old Base Case, New Base
Case (CAIR) , and with CAMR without and with Selected Technological Advances
(Million Tons) 7-22
Table 7-25. Retail Electricity Prices by NERC Region with the Old Base Case, New Base Case
(CAIR), and with CAMR without and with Selected Technological Advances
(Mills/kWh) ($1999) 7-23
Table 7-26. Projected Emissions of Hg for the New Base Case (CAIR) and CAMR with EPA
and EIA Assumptions for Natural Gas Prices and Electric Growth (Tons) 7-24
Table 7-27. Projected Nationwide Emissions of SO2 and NOX under the New Base Case (CAIR)
and CAMR with EPA and EIA Assumptions for Natural Gas and Electric Growth
(Million Tons) 7-25
Table 7-28. Annualized Cost and Present Value Cost Incremental to the New Base Case (CAIR)
with EPA and EIA Assumptions for Natural Gas Prices and Electric Growth (Billion
$1999) 7-25
Table 7-29. Marginal Cost of SO2 and NOX Reductions under the New Base Case (CAIR) and
CAMR with EPA and EIA Assumptions for Natural Gas Prices and Electric Growth
($/ton, in $1999) 7-26
Table 7-30. Pollution Controls under the New Base Case (CAIR) with EPA and EIA
Assumptions for Natural Gas and Electricity Growth (GWs) 7-26
Table 7-31. Generation Mix under the New Base Case (CAIR) and CAMR with EPA and EIA
Assumptions for Natural Gas and Electric Growth (Thousand GWhs) 7-27
Table 7-32. Coal Production for the Electric Power Sector under the New Base Case (CAIR)
and CAMR with EPA and EIA Assumptions for Natural Gas and Electricity Growth
(Million Tons) 7-27
Table 7-33. Retail Electricity Prices by NERC Region for the Base Case (No Further Controls),
CAIR, and CAMR with EPA and EIA Assumptions for Natural Gas and Electricity
Growth (Mills/kWh) ($1999) 7-28
-------
Table 7-34. Potentially Regulated Categories and Entities8 7-29
Table 7-35. Projected Impact of CAMR on Small Entities 7-31
Table 7-36. Summary of Distribution of Economic Impacts of CAIR on Small Entities ... 7-35
Table 7-37. Incremental Annualized Costs under CAMR relative to CAIR, Summarized by
Ownership Group and Cost Category ($1,000,000) 7-36
Table 7-38. Summary of Potential Impacts on Government Entities under CAIR 7-38
Table 7-39. Distribution of Economic Impacts on Government Entities under CAMR .... 7-43
Table 7-40. Incremental Annualized Costs under CAMR Relative to CAIR Summarized by
Ownership Group and Cost Category ($1,000,000) 7-43
Table 7-41. Listing of Runs from the Integrated Planning Model Used in Analyses Done in
Support of the CAMR Final Rule Analyses 7-45
Figures
Figure 7-1. Projected Mercury Emissions in 2020 by State 7-4
Figure 7-2. NERC Power Regions 7-11
-------
SECTION 7
COST AND ENERGY IMPACTS
This chapter reports the cost, economic, and energy impact analysis performed for
CAMR. EPA used the IPM, developed by ICF Consulting, to conduct its analysis. IPM is a
dynamic linear programming model that can be used to examine air pollution control policies for
Hg, SO2, and NOX throughout the contiguous United States for the entire power system.
Documentation for IPM can be found at www.epa.gov/airmarkets/epa-ipm.
7.1 Modeling Background
The analysis presented here covers the electric power sector, a major source of Hg, SO2,
and NOX emissions nationwide. CAMR requires that states control electric generation units
fueled by coal through state Hg emissions reduction requirements. EPA has assumed that states
implement those reductions through a cap-and-trade program. This analysis also assumes that
electric generating units will also comply with CAIR requirements through a cap-and-trade
program. For mercury, the analysis examines three control options, all implemented in multiple
phases. See Table 7-1 for total annual Hg emissions caps for CAMR options examined. For SO2
and NOX, EPA modeled the requirements of the final CAIR. This modeling includes regionwide
annual SO2 and NOX caps on the 23 States and the District and Columbia that are required to
make annual reductions, and includes a regionwide ozone season NOX cap on the 25 States and
the District of Columbia required to make ozone season reductions. See Table 7-2 for total
annual emissions caps under CAIR used in EPA modeling.
Table 7-1. CAMR Options Annual Emissions Caps (Tons)
Option 1 (38/15)
Option 2 (15/15)
Option 3 (24/15)
2010-2014
38
38
38
2015-2017
38
15
24
2018-Thereafter
15
15
15
Table 7-2. CAIR Emissions Caps (Million Tons)
SO2
NOX (Annual)
NOX (Summer)
2010-2014
3.6
1.5
0.6
2015-Thereafter
2.5
1.3
0.5
7-1
-------
The final CAMR requires annual Hg reductions in 50 States and the District of Columbia.
The final CAMR will require a 38 ton cap in 2010 and a 15 ton cap in 2018. Using IPM, EPA
modeled the cost and emissions impacts of three Hg control options to aid in its decision for the
final CAMR. This chapter will provide the analysis conducted for all three options. IPM output
files for the model runs used in CAMR analyses are available in the CAMR docket.
The modeling conducted for this analysis assumes that sources are complying with the
final CAIR control strategy along with a CAMR control strategy. To provide incremental
comparison, the CAIR modeling results are also presented. The CAIR IPM modeling includes
regionwide annual SO2 and NOX caps on the 23 States and the District of Columbia for States
required to make annual reductions, and includes a regionwide ozone season NOX cap on the 25
States and the District of Columbia required to make ozone season reductions. EPA modeled the
final CAIR NOX strategy as an annual NOX cap with a nested, separate ozone season NOX cap.
CAMR was designed to achieve significant Hg emissions reductions from the power
sector in a highly cost-effective manner. EPA analysis has found that the most efficient method
to achieve the emissions reduction targets is through a cap-and-trade system that States have the
option of adopting. States, in fact, can choose not to participate in the optional cap-and-trade
program. However, EPA believes that a cap-and-trade system for the power sector is the best
approach for reducing Hg emissions. As a result, EPA modeling has focused on the cap-and-
trade approach for meeting the CAMR requirements. The modeling done with IPM assumes a
nation-wide Hg cap and trade system on the power sector for the 48 contiguous states. However,
EPA recognizes that states may use a different approach for reducing emissions, given that
CAMR allows States to choose how they will meet their Hg emissions budget through reductions
from utility units. States can elect not to participate in the federal trading program, and pursue
reductions through other means including facility limits and trading limited to inside the state
borders. This would likely impact the cost estimate of the program.
IPM has been used for evaluating the economic and emission impacts of environmental
policies for over a decade. The model's base case incorporates title IV of the Clean Air Act (the
Acid Rain Program), the NOX SIP Call, various New Source Review (NSR) settlements, and
several state rules affecting emissions of SO2 and NOX that were finalized prior to April of 2004.
The NSR settlements include agreements between EPA and Southern Indiana Gas and Electric
Company (Vectren), Public Service Electric & Gas, Tampa Electric Company, We Energies
(WEPCO), Virginia Electric Power Company (Dominion), and Santee Cooper. IPM also
includes various current and future state programs in Connecticut, Illinois, Maine,
Massachusetts, Minnesota, New Hampshire, North Carolina, New York, Oregon, Texas, and
Wisconsin. IPM includes state rules that have been finalized and/or approved by a state's
legislature or environmental agency. The base case is used to provide a reference point to
compare environmental policies and assess their impacts and does not reflect a future scenario
that EPA predicts will occur.
The economic modeling presented in this chapter has been developed for specific
analyses of the power sector. Thus, the model has been designed to reflect the industry as
accurately as possible. As a result, EPA has used discount rates in IPM that are appropriate for
the various types of investments and other costs that the power sector incurs. The discount rates
used in IPM may differ from discount rates used in other EPA analyses done for CAMR,
7-2
-------
particularly the discount rates used in the benefits analysis that are assumed to be social discount
rates. EPA uses the best available information from utilities, financial institutions, debt rating
agencies, and government statistics as the basis for the discount rates used for power sector
modeling. These discount rates have undergone review by the power sector and the Energy
Information Administration. EPA's discount rate approach has not been challenged in court.
EPA's modeling is based on its best judgment for various input assumptions that are
uncertain, particularly assumptions for Hg control technology, future fuel prices and electricity
demand growth. To some degree, EPA addresses the uncertainty surrounding these assumptions
through its sensitivity analysis provided in the chapter. Other uncertainties, like states choosing
not to participate in the trading program, would also impact the cost estimate.
More detail on IPM can be found in the model documentation, which provides additional
information on the assumptions discussed here as well as all other assumptions and inputs to the
model (www.epa.gov/airmarkets/epa-ipm).
7.2 Projected Hg Emissions
Because excess emission reductions are projected to be banked under the first phase of
the Hg program, emissions in the second (or third phase) will be initially higher than the cap that
are required for CAMR. As shown in Figure 7-1, the results of EPA modeling of C AMR show
state-by-state emissions in 2020 for some states do change significantly among CAMR options.
However, for some states, the emissions projections among options follow the same profile as
the national emission projections in 2020.
7-3
-------
New Base Case: CAIR
Option 1
Option 2
Option 3
Scale: 3 37 tons in Pennsylvania under CAIR
Figure 7-1. Projected Mercury Emissions in 2020 by State
Table 7-3 provides projected total Hg emissions levels and Table 7-4 provides projected
speciated Hg emissions levels in 2020. EPA projections of Hg emissions are based on 1999 Hg
ICR emission test data and other more recent testing conducted by EPA, DOE, and industry
participants (for further discussion see Control of Emissions from Coal-Fired Electric Utility
Boilers: An Update, EPA/Office of Research and Development, March 2005, in docket). That
emissions testing has provided a better understanding of Hg emissions and their capture in
pollution control devices. Mercury speciates into three basic forms, ionic, elemental, and
particulate. In general, ionic Hg compounds are more readily adsorbed than elemental Hg. The
presence of chlorine compounds (which tend to be higher for bituminous coals) results in
increased ionic mercury.
Overall the 1999 Hg ICR data revealed higher levels of Hg capture for bituminous
coal-fired plants as compared to low-rank coal-fired plants, large ranges of Hg capture in
existing plants, higher levels of Hg capture in fabric filters (FF) compared to electrostatic
precipitators (ESPs), and a significant capture of ionic Hg in wet SO2 scrubbers. Additional Hg
testing indicates that for bituminous coals SCR has the ability to convert elemental Hg to ionic
Hg and thus allow easier capture in a wet scrubber. This understanding of Hg capture was
incorporated into EPA modeling assumptions (see IPM documentation, Hg EMFs) and is the
basis for projections of Hg co-benefits from installation of scrubbers and SCR under CAIR.
7-4
-------
Table 7-3. Projected Emissions of Hg with the Old Base Case", New Base Case, and with
CAMR Options (Tons)
Old Base Case
New Base Case: CAIR
Option 1 (38/15)
Option 2 (15/1 5)
Option 3 (24/1 5)
2010
46.6
38.0
31.3
30.9
31.1
2015
45.0
34.4
27.9
25.7
27.4
2020
46.2
34.0
24.3
20.1
21.1
Note: The emissions projections are for coal-fired electric power units greater than 25 MW.
" Base case includes Title IV Acid Rain Program, NOX SIP Call, and State rules finalized before March 2004.
Source: Integrated Planning Model run by EPA.
Table 7-4. Projected Speciated Emissions of Hg in 2020 with New Base Case (CAIR) and
CAMR Options (Tons)
1999
New Base Case:
CAIR
Option 1 (38/15)
Option 2 (15/15)
Option 3 (24/1 5)
Elemental Hg
26.2
25.8
17.6
14.3
15.1
Ionic Hg
20.6
7.9
6.6
5.7
5.9
Particulate Hg
1.7
0.8
0.8
0.8
0.8
Total
48.6
34.4
25.0
20.9
21.8
Note: Numbers may not add due to rounding and include un affected units. The emissions data presented here are EPA
modeling results and include some unaffected units. 1999 emissions from 1999 Hg ICR estimate.
7.3 Projected SO2 and NO, Emissions
The addition of Hg cap does not significantly affect SO2 and NOx emissions when
compared to CAIR alone. National SO2 emissions are somewhat lower in 2020 under the
CAMR scenarios because sources are projected to install more scrubbers to achieve compliance
for both CAIR and CAMR. Because of excess emission reductions are projected to be banked
under the title IV Acid Rain Program that sources will be allowed to use under the requirements
of CAIR, emissions in 2010 and 2015 will be higher than the caps that are required for CAIR.
Tables 7-5 and 7-6 provide projected emissions levels for SO2 and NOx.
7-5
-------
Table 7-5. Projected Emissions of SO2 with the Old Base Case8, New Base Case (CAIR),
and with CAMR Options (Million Tons)
Old Base
Case
New Base
Case: CAIR
Option 1
(38/15)
Option 1
(15/15)
Option 3
(24/15)
2010
Nationwide
9.7
6.1
6.1
6.1
6.1
CAIR Region
8.8
5.1
5.1
5.1
5.1
2015
Nationwide
8.9
5.0
4.9
4.9
4.9
CAIR Region
8.0
4.0
4.0
3.9
4.0
2020
Nationwide
8.6
4.3
4.2
4.2
4.2
CAIR Region
7.7
3.3
3.3
3.3
3.3
Note: Emissions projections are for fossil-fired electric power sector.
' Base case includes Title IV Acid Rain Program, NOX SIP Call, and State rules finalized before March 2004.
Source: Integrated Planning Model run by EPA.
Table 7-6. Projected Emissions of NOx with the Old Base Case8, New Base Case (CAIR),
and with CAMR Options (Million Tons)
Old Base
Case
New Base
Case: CAIR
Option 1
(38/15)
Option 2
(15/15)
Option 3
(24/15)
2010
Nationwide
3.6
2.5
2.4
2.4
2.4
CAIR Region
2.8
1.5
1.5
1.5
1.5
2015
Nationwide
3.7
2.2
2.2
2.2
2.2
CAIR Region
2.8
1.3
1.3
1.3
1.3
2020
Nationwide
3.7
2.2
2.2
2.2
2.2
CAIR Region
2.8
1.3
1.3
1.3
1.3
Note: Emissions projections are for fossil-fired electric power sector.
" Base case includes Title IV Acid Rain Program, NOX SIP Call, and State rules finalized before March 2004.
Source: Integrated Planning Model run by EPA.
7.4 Projected Costs
Table 7-7 provides EPA's projections of annual and present value costs incremental to
CAIR. The cost of electricity generation represents roughly one-third to one-half of total
electricity costs, with transmission and distribution costs representing the remaining portion. A
better impact measure of the cost to the consumer is the impact on electricity pricing, which is
shown in a later table.
The presence of an earlier cap under CAMR Option 2 (an the reduction of years of
banking excess emissions) results in higher projected costs than Option 1. CAMR Option 2
costs are projected to be the highest of the options and is reflected by the lowest projected Hg
7-6
-------
emissions in 2020. The intermediate cap of 24 tons under Option 3 also reduces the amount of
banking of excess emissions and results in higher projected costs than Option 1. However,
because the final cap goes into place in 2018, the projected costs are lower than Option 2 and is
reflected by the projected Hg emission in 2020 being higher than Option 2.
The marginal costs for Hg, SO2 and NOX can be found in Table 7-8. EPA projects a
reduction in the SO2 allowance price and changes in the NOx allowance price under CAMR
when compared to CAIR alone. The changes in SO2 and NOx allowance prices are due to the
different set of costs faced by sources under CAMR. In th case of SO2, the ability to control for
both Hg and SO2 effectively through scrubbers results in marginal cost of SO2 being reflected in
th Hg allowance price such that SO2 allowance price falls. In th case of NOx, because SCR is an
effective Hg control when combined with a scrubber, facilities choose to control different units
than they would in th absence of a cap on Hg emissions. Sources will choose to control unit
where they can install a combination of scrubbers and SCR to achieve both mercury and NOx
reductions.
Table 7-7. Annualized National Private Compliance Cost and Present Value Cost ($1999)
Cost (billions)
Option 1 (38/15)
Option 2 (15/15)
Option 3 (24/15)
2010
$0.16
$0.16
$0.16
2015
$0.10
$0.36
$0.18
2020
$0.75
$1.04
$1.04
Present value (2007-2025)
$3.9
$6.0
$5.2
Note: Annual incremental costs of CAIR are $2.4 billion in 2010, $3.6 billion in 2015, and $4.4 billion in 2020, present value
(2007-2025) is $41.1 billion.
Note: Numbers rounded to the nearest ten million for annualized cost.
Source: Integrated Planning Model run by EPA.
Table 7-8. Marginal Cost of Hg, SO2, and NOx Reductions with CAMR Options ($1999)
New Base
Case: CAIR
Option 1
(38/15)
Option 2
(15/15)
Option 3
(24/15)
SO2 ($/ton)
NOx ($/ton)
SO2 ($/ton)
NOx ($/ton)
Hg ($/lb)
SO2 ($/ton)
NOx ($/ton)
Hg ($/lb)
SO2 ($/ton)
NOx ($/ton)
Hg ($/lb)
2010
$800
$1,300
$700
$1,200
$23,200
$700
$1,200
$29,000
$700
$1,200
$26,400
2015
$1,000
$1,600
$900
$1,500
$30,100
$900
$1,500
$37,600
$900
$1,500
$34,200
2020
$1,300
$1,600
$1,200
$1,300
$39,000
$1,100
$1,200
$48,700
$1,100
$1,300
$44,400
Note: Numbers rounded to the nearest hundred for marginal cost.
Source: Integrated Planning Model run by EPA.
Actual costs may be lower than those presented since modeling assumes no
improvements in the cost of mercury control technology. Given that this is the first time
7-7
-------
mercury emission will be regulated at the federal level1 for the coal-fired power sector and given
the current level of research and demonstration of mercury control technologies, control cost are
expected to improve over time. For purposes of options comparisons, EPA has conservatively
assumed no cost improvements in Hg control technologies. Later, in this Chapter, EPA will
present a sensitivity analysis in which we examine impact of mercury technology improvements
by providing a lower cost mercury control option in future years.
7.5 Projected Control Technology Retrofits
Under the modeled Hg options, Hg reduction is projected to result from the installation of
additional flue gas desulfurization (FGD or scrubbers) on existing coal-fired generation capacity
for SO2 control, additional selective catalytic reduction technology (SCR) on existing coal-fired
generation capacity for NOX control, and activated carbon injection (ACI) on existing coal-fired
capacity for Hg-specific control (see Table 7-9). In addition, during the first phase of the Hg
program, some Hg banking of emissions is projected to be attributed to coal switching and
dispatch changes. Most of the NOX reductions achieved in the first phase of the rule can be
attributed to the large pool of existing SCR that are used during the ozone season in the NOX SIP
call region that, for relatively little additional cost, run the SCRs year-round. Due to earlier
second phase cap (Option 2) and the addition of a third phase (Option 3), less banking is
projected in 2010 to 2015 timeframe and results in more ACI in 2020 as emissions approach the
15 ton cap.
Table 7-9. Pollution Controls by Technology with the Old Base Case, New Base Case
(CAIR), and with CAMR Options (GW)
Old Base Case
New Base
Case: CAIR
Option 1
(38/15)
Option 2
(15/15)
Option 3
(24/15)
2010
FGD
110
146
146
146
147
SCR
111
125
126
127
127
ACI
--
-
2
3
3
2015
FGD
116
177
179
179
179
SCR
119
151
153
153
153
ACI
-
-
3
12
5
2020
FGD
117
198
199
198
199
SCR
121
153
156
156
156
ACI
0.3
0.5
13
38
30
Note: Numbers may not add due to rounding. Base case retrofits include existing scrubbers and SCR as well as additional
retrofits for the Title IV Acid Rain Program, the NO, SIP call, NSR settlements, and various state rules.
Source: Integrated Planning Model run by EPA.
1 Some states have enacted Hg reduction requirements for the coal-fired power sector. See IPM documentation for
modeled State Hg regulations.
7-8
-------
7.6 Projected Generation Mix
Table 7-10 show the generation mix with CAMR. Coal-fired generation and natural gas-
fired generation are projected to remain relatively unchanged because of the phased-in nature of
CAMR, which allows industry the appropriate amount of time to install the necessary pollution
controls.
Table 7-10. Generation Mix with the Old Base Case, with New Base Case (CAIR), and
with CAMR Options (Thousand GWhs)
Old Base Case
New Base
Case: CAIR
Option 1
(38/15)
Option 2
(15/15)
Option 3
(24/15)
Coal
Oil/Natural Gas
Other
Coal
Oil/Natural Gas
Other
Coal
Oil/Natural Gas
Other
Coal
Oil/Natural Gas
Other
Coal
Oil/Natural Gas
Other
2010
2,198
777
1,223
2,165
807
1,217
2,160
812
1,216
2,158
813
1,216
2,159
812
1,216
2015
2,195
1,072
1,233
2,197
1,069
1,232
2,194
1,072
1,233
2,191
1,075
1,233
2,193
1,074
1,232
2020
2,410
1,221
1,218
2,384
1,247
1,217
2,365
1,265
1,217
2,365
1,266
1,217
2,367
1,263
1,217
Change From
New Base
Case in 2020
-0.8%
1.5%
0.0%
-0.8%
1.5%
0.0%
-0.7%
1.3%
0.0%
Note: Numbers may not add due to rounding.
Source: 2003 data are from EIA: Coal - 1,970; Oil/Natural Gas - 758; Other -1,120. Projections are from the Integrated
Planning Model run by EPA.
Under all three Hg control options modeled and relative to the new base case, no coal-
fired generation is projected to be uneconomic to maintain under CAMR.
7.7 Projected Capacity Additions
In addition, EPA projects that future growth in electric demand will be met with a
combination of new natural gas- and coal-fired capacity (see Table 7-11).
7-9
-------
Table 7-11. Total Coal and Natural Oil/Gas-Fired Capacity by 2020 (GW)
Pulverized Coal
IGCC
Oil/Gas
Current
305
0.6
395
Old Base
Case
318
8
467
New Base
Case:
CAIR
315
9
469
Option 1
(38/15)
314
8
471
Option 2
(15/15)
314
8
471
Option 3
(24/15)
314
8
471
Source: Current data are from EPA's NEEDS 2004; projections are from the Integrated Planning Model run by EPA.
7.8 Projected Coal Production for the Electric Power Sector
Coal production for electricity generation is expected to increase relative to current
levels, with or without CAIR (see Table 7-12). The reductions in emissions from the power
sector will be met through the installation of pollution controls for Hg, SO2 and NOX removal.
The pollution controls can achieve up to a 95 percent SO2 removal rate, which allows industry to
rely more heavily on local bituminous coal in the eastern and central parts of the country that has
a higher sulfur content and is less expensive to transport than western subbituminous coal.
Table 7-12. Coal Production for the Electric Power Sector with the Old Base Case, New
Base Case (CAIR) , and with CAMR Options (Million Tons)
Supply Area
Appalachia
Interior
West
National
Supply
Area
Appalachia
Interior
West
National
2000
299
131
475
905
2003
275
135
526
936
Old Base
2010 2015
325 315
161 162
603 631
1,089 1,109
Option 1 (38/15)
2010
303
169
589
1,061
2015 2020
310 330
194 224
568 572
1,071 1,127
Case
2020
301
173
714
1,188
New Base Case: CAIR
2010
306
165
607
1,078
Option 2 (15/15)
2010
303
170
587
1,060
2015 2020
309 322
195 231
565 574
1,069 1,127
2015
306
191
586
1,083
2020
331
218
609
1,158
Option 3 (38/15)
2010
304
171
587
1,061
2015
309
194
567
1,070
2020
325
232
570
1,127
Source: 2000 and 2003 data are derived from EIA data. All projections are from the Integrated Planning Model run by
EPA.
7-10
-------
7.9 Projected Retail Electricity Prices
Retail electricity prices for the U.S. are projected to increase a small amount with CAMR
(see Table 7-13). The cap-and-trade approach allows industry to meet the requirements of
CAMR in the most cost-effective manner, thereby minimizing the costs passed on to consumers.
Retail electricity prices by NERC region (see Figure 7-2) are provided in Table 7-14 and show
small increases in retail prices for the NERC regions in the eastern part of the country. By 2020,
national retail electricity prices are projected to be roughly 0.3 percent higher with CAMR when
compared to CAIR.
Figure 7-2. NERC Power Regions
Table 7-13. Projected National Retail Electricity Prices with the Old Base Case, New Base
Case (CAIR), and CAMR Options (Mills/kWh) ($1999)
Year
2010
2015
2020
Old Base Case
58
61
61
New Base Case:
CAIR
61
64
64
Option 1
(38/15)
61
65
65
Option 2
(15/15)
61
65
65
Option 3 (24/15)
61
65
65
Source: Retail Electricity Price Model run by EPA. 2000 national electric price is 66 mills/kWh from EIA's AEO 2003.
7-11
-------
Table 7-14. Retail Electricity Prices by NERC Region with the Old Base Case, New Base
Case (CAIR), and with CAMR Options (Mills/kWh) ($1999)
Option 1
Power Region
ECAR(l)
ERCOT (2)
MAAC(3)
MAIN (4)
MAPP (5)
NY (6)
NE(7)
FRCC (8)
STV(9)
SPP(IO)
PNW(ll)
RM(12)
CALI(13)
National
Primary States Included
OH, MI, IN, KY, WV, PA
rx
PA, NJ, MD, DC, DE
IL, MO, WI
MN.IA, SD.ND.NE
NY
VT, NH, ME, MA, CT, RI
FL
VA, NC, SC, GA, AL, MS, TN, AR, LA
KS, OK, MO
WA, OR, ID
MT, WY, CO, UT, NM, AZ, NV, ID
~A
Contiguous Lower 48 States
2000
57.4
65.1
80.4
61.2
57.4
104.3
89.9
67.9
59.3
59.3
45.9
64.1
94.7
66.0
Base Case
2010 2015 2020
51.7 55.2 56.1
57.9 64.4 62.6
59.3 69.4 72.2
52.6 57.8 61.0
52.8 49.3 47.6
82.8 87.9 88.1
77.4 83.9 82.8
71.2 71.3 69.5
56.2 55.1 55.3
54.2 57.0 56.7
49.6 47.4 46.9
63.9 65.2 64.7
97.1 98.9 99.3
60.3 63.1 63.4
CAIR
2010 2015 2020
53.7 58.6 58.0
59.4 64.5 63.3
61.0 72.0 72.7
53.9 60.4 62.0
52.9 49.6 48.0
83.3 88.9 88.5
77.4 84.7 83.1
71 7 72.3 70.5
57.0 56.2 56.6
54.6 57.5 57.0
49.8 47.5 46.9
64.1 65.6 65.4
97.3 99.1 99.5
61.3 64.5 64.3
Option 1
2010 2015 2020
53.9 58.7 58.1
59.1 64.9 63.4
61.3 72.1 72.9
54.1 60.5 62.3
53.0 49.6 48.2
83.3 89.1 88.7
77.5 84.8 83.1
71.8 72.3 70.7
57.1 56.2 56.8
54.6 57.5 57.2
49.8 47.5 47.1
64.2 65.6 65.1
97.3 99.1 99.7
61.3 64.5 64.5
Change
from
CAIR
2020
0.2%
0.1%
0.2%
0.5%
0.5%
0.4%
0.0%
0.3%
0.3%
0.3%
0.5%
-0.4%
0.3%
0.2%
Option 2
Power Region
ECAR(l)
ERCOT (2)
MAAC (3)
MAIN (4)
MAPP (5)
NY (6)
NE(7)
FRCC (8)
STV(9)
SPP(IO)
PNW(ll)
RM(12)
CALI(13)
National
Power Region
ECAR(l)
ERCOT (2)
MAAC (3)
MAIN (4)
MAPP (5)
NY (6)
NE(7)
FRCC (8)
STV(9)
SPP(IO)
PNW(ll)
RM(12)
CALI(13)
National
Primary States Included
3H,MI,IN,KY,WV, PA
rx
PA, NJ, MD, DC, DE
1L, MO, WI
MN, IA, SD.ND.NE
NY
VT, NH, ME, MA, CT, RI
FL
VA, NC, SC, GA, AL, MS, TN, AR, LA
KS, OK, MO
WA, OR, ID
MT, WY, CO, UT, NM, AZ, NV, ID
CA
Contiguous Lower 48 States
Primary States Included
3H, MI, IN, KY, WV, PA
rx
PA, NJ, MD, DC, DE
1L, MO, WI
MN.IA, SD,ND,NE
NY
VT, NH, ME, MA, CT, RI
FL
VA, NC, SC, GA, AL, MS, TN, AR, LA
KS, OK, MO
WA, OR, ID
MT, WY, CO, UT, NM, AZ, NV, ID
CA
Contiguous Lower 48 States
2000
57.4
65.1
80.4
61.2
57.4
104.3
89.9
67.9
59.3
59.3
45.9
64.1
94.7
66.0
2000
57.4
65.1
80.4
61.2
57.4
104.3
89.9
67.9
59.3
59.3
45.9
64.1
94.7
66.0
Base Case
2010 2015 2020
51.7 55.2 56.1
57.9 64.4 62.6
59.3 69.4 72.2
52.6 57.8 61.0
52.8 49.3 47.6
82.8 87.9 88.1
77.4 83.9 82.8
71.2 71.3 69.5
56.2 55.1 55.3
54.2 57.0 56.7
49.6 47.4 46.9
63.9 65.2 64.7
97.1 98.9 99.3
60.3 63.1 63.4
Option 3
Base Case
2010 2015 2020
51.7 55.2 56.1
57.9 64.4 62.6
59.3 69.4 72.2
52.6 57.8 61.0
52.8 49.3 47.6
82.8 87.9 88.1
77.4 83.9 82.8
71.2 71.3 69.5
56.2 55.1 55.3
54.2 57.0 56.7
49.6 47.4 46.9
63.9 65.2 64.7
97.1 98.9 99.3
60.3 63.1 63.4
CAIR
2010 2015 2020
53.7 58.6 58.0
59.4 64.5 63.3
61.0 72.0 72.7
53.9 60.4 62.0
52.9 49.6 48.0
83.3 88.9 88.5
77.4 84.7 83.1
71.7 72.3 70.5
57.0 56.2 56.6
54.6 57.5 57.0
49.8 47.5 46.9
64.1 65.6 65.4
97.3 99.1 99.5
61.3 64.5 64.3
CAIR
2010 2015 2020
53.7 58.6 58.0
59.4 64.5 63.3
61.0 72.0 72.7
53.9 60.4 62.0
52.9 49.6 48.0
83.3 88.9 88.5
77.4 84.7 83.1
71.7 72.3 70.5
57.0 56.2 56.6
54.6 57.5 57.0
49.8 47.5 46.9
64.1 65.6 65.4
97.3 99.1 99.5
61.3 64.5 64.3
Option 2
2010 2015 2020
53.9 58.8 58.1
59.1 64.9 63.4
61.3 72.1 72.9
54.1 60.6 62.4
53.0 49.7 48.3
83.3 89.2 88.7
77.5 84.7 83.2
71.7 72.3 70.7
57.1 56.4 56.8
54.6 57.6 57.6
49.8 47.4 47.2
64.2 65.6 65.0
97.3 99.1 99.7
61.3 64.6 64.5
Option 3
2010 2015 2020
53.9 58.8 58.1
59.1 64.9 63.4
61.3 72.1 72.9
54.0 60.6 62.5
53.0 49.7 48.3
83.3 89.2 88.7
77.5 84.8 83.1
71.8 72.3 70.7
57.1 56.3 56.8
54.6 57.5 57.5
49.8 47.4 47.2
64.2 65.6 65.1
97.3 99.1 99.8
61.3 64.6 64.5
Change
from
CAIR
2020
0.2%
0.1%
0.2%
0.7%
0.8%
0.3%
0.2%
0.3%
0.4%
1.0%
0.6%
-0.6%
0.3%
0.3%
Change
from
CAIR
2020
0.2%
0.1%
0.2%
0.7%
0.7%
0.3%
0.0%
0.3%
0.4%
0.9%
0.6%
-0.5%
0.3%
0.3%
Source: Retail Electricity Price Model run by EPA. 2000 prices from EIA's AEO 2003.
7-12
-------
7.10 Projected Fuel Price Impacts
The impacts of CAMR on coal prices and natural gas prices before shipment are shown
in Table 7-15.
Table 7-15. Henry Hub Natural Gas Prices and Average Delivered Coal Prices with the
Old Base Case, New Base Case (CAIR), and with CAMR Options (1999$/mmBtu)
Old Base Case
New Base
Case: CAIR
Option 1
(38/15)
Option 2
(15/15)
Option 3
(24/15)
2010
Delivered
Coal
1.05
1.05
1.05
1.05
1.05
Henry Hub
Gas
3.20
3.25
3.25
3.25
3.25
2015
Delivered
Coal
1.01
0.98
0.98
0.98
0.98
Henry Hub
Gas
3.25
3.30
3.30
3.30
3.30
2020
Delivered
Coal
0.96
0.93
0.93
0.93
0.94
Henry Hub
Gas
3.16
3.20
3.25
3.25
3.25
Source: Integrated Planning Model run by EPA. 2000 natural gas data are from Platts GASdata is $4.15/mmBtu. 2000 coal
price from EIA is $1.25/mmBtu.
Note: Coal price changes largely result from changes in the mix of coal types used. Delivered coal prices vary widely, but
large changes in the cost of each type of coal are not projected.
7.11 Social Cost Calculations
The annualization factor used for pure social cost calculations (for annualized costs)
normally includes the life of capital and the social discount rate. For purposes of benefit-cost
analysis of this rule, EPA has calculated the annualized social costs using the discount rates from
the benefits analysis for CAMR (3% and 7% and a 30 year life of capital. The costs of added
insurance was included in the calculations, but local taxes were not included because they are
considered to be transfer payments, and not a social cost). Using these discount rates, the social
costs of CAMR incremental to CAIR are $151 million in 2010 and $848 million in 2020 using a
discount rate of 3%, and are $157 million in 2010 and $896 million in 2020 using a discount rate
of 7%.
Recent research suggests that the total social costs of a new regulation may be affected
by interactions between the new regulation and pre-existing distortions in the economy, such as
taxes. In particular, if cost increases due to a regulation are reflected in a general increase in the
price level, the real wage received by workers may be reduced, leading to a small fall in the total
amount of labor supplied. This "tax interaction effect" may result in an increase in deadweight
loss in the labor market and an increase in total social costs. The limited empirical data available
to support quantification of any such effect leads to this qualitative identification of the costs.
7-13
-------
7.12 Limitations of Analysis
EPA's modeling is based on its best judgment for various input assumptions that are
uncertain, particularly assumptions for Hg control technologies and future fuel prices and
electricity demand growth. To some degree, EPA addresses the uncertainty surrounding these
three assumptions through its sensitivity analysis. Sensitivity analysis on future fuel prices and
electricity demand growth are provided in section 7.15. A discussion on Hg technology cost
uncertainty and sensitivity analysis are provided below in section 7.14. As a general matter, the
Agency selects the best available information from available engineering studies of air pollution
controls and has set up what it believes is the most reasonable modeling framework for analyzing
the cost, emission changes, and other impacts of regulatory controls.
The annualized cost estimates of the private compliance costs that are provided in this
analysis are meant to show the increase in production (engineering) costs of CAMR to the power
sector. In simple terms, the private compliance costs that are presented are the annual increase in
revenues required for the industry to be as well off after CAMR is implemented as before. To
estimate these annualized costs, EPA uses a conventional and widely-accepted approach that is
commonplace in economic analysis of power sector costs for estimating engineering costs in
annual terms. For estimating annualized costs, EPA has applied a capital recovery factor (CRF)
multiplier to capital investments and added that to the annual incremental operating expenses.
The CRF is derived from estimates of the cost of capital (private discount rate), the amount of
insurance coverage required, local property taxes, and the life of capital. The private compliance
costs presented earlier are EPA's best estimate of the direct private compliance costs of CAMR.
The annualized cost of CAMR, as quantified here, is EPA's best assessment of the cost
of implementing CAMR, assuming that States adopt the model cap and trade program. Under
CAMR, States are required to meet Hg emission budget based on reductions from coal-fired
utility units. States have the discretion to participate in the federal cap-and-trade program or to
meet their budget through other options (including facility limits and trading restricted inside
state boarder). These costs are generated from rigorous economic modeling of changes in the
power sector due to CAMR. This type of analysis using IPM has undergone peer review and
federal courts have upheld regulations covering the power sector that have relied on IPM's cost
analysis.
The direct private compliance cost includes, but is not limited to, capital investments in
pollution controls, operating expenses of the pollution controls, investments in new generating
sources, and additional fuel expenditures. EPA believes that the cost assumptions used for
CAMR reflect, as closely as possible, the best information available to the Agency today. The
cost associated with monitoring emissions, reporting, and record keeping for affected sources is
not included in these annualized cost estimates, but EPA has done a separate analysis and
estimated the cost to be about $76 million (see final CAMR preamble Section VLB. Paperwork
Reduction Act).
Furthermore, there are some unquantified costs that EPA wants to identify as limits to its
analysis. These costs include the costs of federal and State administration of the program, which
we believe are modest given our experience with the Acid Rain Program and the NOx Budget
Trading Program and likely to be less than the alternative of States developing approvable State
7-14
-------
Plans, securing EPA approval of those State Plans, and federal/state enforcement. There also
may be unqualified costs of transitioning to CAMR, such as the costs associated with the
retirement of smaller or less efficient electricity generating units, and employment shifts as
workers are retrained at the same company or re-employed elsewhere in the economy. There
are certain relatively small permitting costs associated with Title IV that new program entrants
face (we believe there are far less than 1,000 new entrants who may require one day of additional
work for trading permits). In a separate analysis for the CAIR RIA, EPA estimated the indirect
cost and impacts of higher electricity prices on the entire economy for the CAIR scenario (see
Regulatory Impact Analysis for the Final Clean Air Interstate Rule, Appendix E (March 2005)).
Given the small difference in electricity prices between CAMR and CAIR, analysis for CAMR
would project similar results.
Cost estimates for CAMR are based on results from ICF's Integrated Planning Model.
The model minimizes the costs of producing electricity (including abatement costs) while
meeting load demand and other constraints (full documentation for IPM can be found at
www.epa.gov/airmarkets/epa-ipm). The structure of the model assumes that the electric utility
industry will be able to meet the environmental emission caps at least cost. Montgomery (1972)
has shown that this least cost solution corresponds to the equilibrium of an emission permit
system.2 See also Atkinson and Tietenburg (1982), Krupnick et al. (1980), and McGartland and
Gates (1985).3 4 5 However, to the extent that transaction and/or search costs, combined with
institutional barriers, restrict the ability of utilities to exhaust all the gains from emissions
trading, costs are underestimated by the model. Utilities in the IPM model also have "perfect
foresight." To the extent that utilities misjudge future conditions affecting the economics of
pollution control, costs may be understated as well.
This modeling analysis does not take into account the potential for advancements in the
capabilities of pollution control technologies for SO2 and NOX removal as well as reductions in
their costs over time. Market-based cap-and-trade regulation serves to promote innovation and
the development of new and cheaper technologies. As an example, recent cost estimates of the
Acid Rain SO2 trading program by Resources for the Future (RFF) and MIT's Center for Energy
and Environmental Policy Research (CEEPR) have been as much as 83% lower than originally
projected by the EPA.6 It is important to note that the original analysis for the Acid Rain
2 Montgomery, W. David 1972. "Markets in Licenses and Efficient Pollution Control Programs." Journal of
Economic Theory 5(3): 395-418.
3 S. Atkinson and T. Tietenberg 1982. "The empirical properties of two classes of design for transferable discharge
permit markets," Journal of Environmental Economics and Management 9:101-121
4 Krupnick, A., W. Gates and E. Van De Verg. 1980. "On Marketable Air Pollution Permits: The Case for a System
of Pollution Offsets." Journal of Environmental Economics and Management 10: 233-47.
5 McGartland, A and W. Gates. 1985. "Marketable Permits for the Prevention of Environmental Deterioration,"
Journal of Environmental Economics and Management 12: 207-228.
6 See (1) Carlson, Curtis.; Burtraw, Dallas R.; Cropper, Maureen and Palmer, Karen L. 2000. Sulfur Dioxide Control
by Electric Utilities: What Are the Gain from Trade? Journal of Political Economy 108 (#6): 1292-1326, and (2)
Ellerman, Denny. January, 2003. Ex Post Evaluation of Tradable Permits: The U.S. SO2 Cap-and-Trade Program.
7-15
-------
Program done by EPA also relied on an optimization model like IPM. Ex ante, EPA costs
estimates of roughly $2.7 to $6.2 billion7 in 1989 were an overestimate of the costs of the
program in part because of the limitation of economic modeling to predict technological
improvement of pollution controls and other compliance options such as fuel switching. Ex post
estimates of the annual cost of the Acid Rain SO2 trading program range from $1.0 to $1.4
billion. Harrington et al. have compared estimates of actual costs of many large EPA regulatory
programs to predictions of those costs made while programs were under development and found
a tendency for predicted costs to overstate actual implementation costs for market-based
programs.8 EPA's mobile source programs use adjusted engineering cost estimates to account
for this fact, which EPA has not done in this case.9
As configured in this application, the IPM model does not take into account demand
response (i.e., consumer reaction to electricity prices). The increased retail electricity prices
shown in Table 7-14 would prompt end users to curtail (to some extent) their use of electricity
and encourage them to use substitutes.10 The response would lessen the demand for electricity,
lowering electricity prices and reducing generation and emissions. Because of demand response,
certain unquantified negative costs (i.e., savings) result from the reduced resource costs of
producing less electricity because of lower demand. To some degree, these saved resource costs
will offset the additional costs of pollution controls and fuel switching that we would anticipate
with CAMR. Although the reduction in electricity use is likely to be small, the cost savings from
such a large industry ($250 billion in revenues in 2003) is likely to be substantial. EIA analysis
examining multi-pollutant legislation under consideration in 2003 indicates that the annualized
costs of CAMR may be overstated substantially by not considering demand response.
It is also important to note that the capital cost assumptions for scrubbers used in EPA
modeling applications are highly conservative. These are a substantial part of the compliance
costs. Data available from recent published sources show the reported FGD costs from recent
installations to be below the levels projected by the IPM. In addition, EPA conducted a survey
of recent FGD installations and compared the costs of these installations to the costs used in
IPM. This survey included small, mid-size, and large units. Examples of the comparison of
these referenced published data with the FGD capital cost estimates obtained from IPM are
provided in the Final CAMR docket. There is also evidence that scrubber costs will decrease in
the future because of the learning-by-doing phenomenon, as more scrubbers are installed".
Massachusetts Institute of Technology Center for Energy and Environmental Policy Research.
7 2010 Phase II cost estimate in $1995.
8 Harrington, W. R.D. Morgenstern, and P. Nelson, 2000. "On the Accuracy of Regulatory Cost Estimates," Journal
of Policy Analysis and Management 19(2): 297-322.
9 See recent regulatory impact analysis for the Tier 2 Regulations for passenger vehicles (1999) and Heavy-Duty
Diesel Vehicle Rules (2000).
10 The degree of substitution/curtailment depends on the price elasticity of electricity.
11 Manson, Nelson, and Neumann, 2002. "Assessing the Impact of Progress and Learning Curves on Clean Air Act
Compliance Costs," Industrial Economics Incorporated.
7-16
-------
Another area of uncertainty is the performance of mercury control removal systems, like
the one assumed in the modeling, activated carbon injection with added pulse-jet fabric filters.
ACI systems have shown great promise in demonstrated tests. However, there is uncertainty
about the availability and effectiveness of ACI across all coal types in the 2010 timeframe, since
these systems have not been fully deployed on coal-fired generating plants. EPA's assumption of
90% removal for ACI is based on EPA's Office of Research and Development (ORD) assessment
( for further discussion see Control of Emissions from Coal-Fired Electric Utility Boilers: An
Update, EPA/Office of Research and Development, March 2005, in CAMR docket). Although
modeled in IPM to be available immediately for all coal-fired generation as a simplification of
modeling, ORD assessment concluded that ACI could not be fully deployed on all plants by
2010 timeframe. EPA's modeling projects only a small amount of ACI use in the 2010
timeframe which is consistent with ORD's conclusion about the availability of ACI.
An additional limitation of Hg control assumptions is that we are assuming no
development in control technologies even though we recognize that this is a fast moving area
with new developments nearly monthly. Actual costs may be lower than those presented since
modeling assumes no improvements in the cost of mercury control technology. Given that this is
the first time mercury is regulated for the coal-fired power sector and the current level of
research and demonstration of mercury control technologies, control costs are expected to
improve over time. For purposes of modeling, EPA has conservatively based its cost
assumptions for mercury control on today's knowledge and not included cost improvement
assumptions in the modeling. Later, in this Chapter (section 7.14), EPA presents a sensitivity
analysis in which we examine impact of mercury technology improvements by providing a lower
cost mercury control option in future years. It is important to note that CAMR's cap-and-trade
approach will encourage technological innovation in Hg emissions control and allow sources to
exploit currently unforeseen emissions control technologies.
Further, while there are many choices of technology for mercury control in existence or
under development, several are not offered to model plants in IPM. Plants in IPM cannot retrofit
with a fabric filter or make improvements to existing controls to capture mercury, such as
improving the cloth to air ratio of the fabric filter, up-grading their ESP or injecting carbon. In
addition, research and development continues on other Hg control technologies, including the
use of pre-combustion controls (e.g. K-fuels), or multi-pollutant controls (i.e., one control
removing SO2, NOx, and Hg). Given a cap-and-trade approach, we would expect further
development and innovation of technology.
EPA's latest update of IPM incorporates State rules or regulations adopted before March
2004 and various NSR settlements. Documentation for IPM can be found at
www.epa.gov/airmarkets/epa-ipm. Any State or settlement action since that time has not been
accounted for in our analysis in this chapter.
On balance, after consideration of various unquantified costs (and savings that are
possible), EPA believes that the annual private compliance costs that we have estimated are
more likely to overstate the future annual compliance costs that industry will incur, rather than
understate those costs.
7-17
-------
7.13 Significant Energy Impact
According to E.O. 13211: Actions that Significantly Affect Energy Supply, Distribution,
or Use, this rule is not significant, measured incrementally to CAIR, because it does not have a
greater than a 1 percent impact on the cost of electricity production and it does not result in the
retirement of greater than 500 MW of coal-fired generation.
Several aspects of CAMR are designed to minimize the impact on energy production.
First, EPA recommends a trading program rather than the use of command-and-control
regulations. Second, compliance deadlines are set cognizant of the impact that those deadlines
have on electricity production. Both of these aspects of CAMR reduce the impact of the
proposal on the electricity sector.
7.14 Sensitivity Analysis on Assumptions for Hg Control Costs
This section presents results of cost sensitivity analysis using the IPM. As discussed
earlier in this Chapter, actual costs may be lower than those presented since modeling assumes
no improvements in the cost of mercury control technology. Given that this is the first time
mercury is federally regulated for the coal-fired power sector and given the current level of
research and demonstration of mercury control technologies, control cost are expected to
improve over time. The sensitivity analysis presented examines the impacts of possible
improvements in Hg control costs over time. EPA selected Option 1 as the policy option for the
final CAMR. For that reason, EPA proceeded with sensitivity analyses for that option.
The sensitivity analysis presented includes examination of the impact of mercury
technology improvements by providing a lower cost mercury control option in future years.
Specifically, the sensitivity analysis examines the impact of providing a second ACI option in
2013 with brominated sorbents and lower capital costs. The assumptions of costs and
performance for the sensitivity analysis is based on recent testing sponsored by EPA, DOE, and
industry and more information on these advanced sorbents can be found in white paper by EPA's
Office of Research and Development, available in the docket. For purposes of modeling, EPA
has assumed the availability of two ACI options: (1) ACI using conventional sorbents and
achieving 90% removal with the addition of a fabric filter; and (2) ACI using advanced sorbents
and achieving 80 to 90% removal without the addition of a fabric filter (see memorandum to the
docket entitled "Assumptions used in sensitivity analysis for the Clean Air Mercury Rule"). The
first ACI option is available at the start of the model and the second ACI is available in 2013.
For comparison of impacts, the sorbent sensitivity was modeled based on the reduction levels for
CAMR Option 1 (Hg trading scenario plus CAIR of 38 tons in 2010, 15 tons in 2018).
Tables 7-16 and 7-17 provide Hg, SO2, and NOx emission projections for the sorbent
sensitivity option. Because the banking of excess emission under the first phase of the Hg
program, emissions are projected to be higher than the cap that is required for CAMR in 2020.
However, with lower future Hg technology costs, less banking and higher emissions are
projected in 2010 and 2015 under the sorbent sensitivity option.
Table 7-19 provides annual and present value costs incremental to CAIR and Table 7-20
provides marginal costs. Under the sorbent sensitivity, the second ACI option has higher O&M
7-18
-------
costs, but lower capital costs resulting in overall lower cost projections. Compared with CAMR
option 1, annual costs and present value cost are projected to be lower for the sorbent sensitivity
option and Hg marginal cost are projected to be about 50 percent lower.
The lower costs for ACI technology also results in higher projections of ACI retrofits in
2020 for the sorbent sensitivity option (see Table 7-21). When compared to Option 1, Coal
generation and production are projected to increase under the sorbent sensitivity option (see
Tables 7-22, 7-23, and 7-24). Retail electricity prices are not projected to changes significantly
when comparing the sorbent sensitivity option to Option 1 (see Table 7-25).
Table 7-16. Projected Emissions of Hg with New Base Case (CAIR) and CAMR, without
and with Selected Technological Advances (Tons)
New Base Case: CAIR
Option 1 - Current Technology
Option 1 - Sorbent Sensitivity
2010
38.0
31.3
32.6
2015
34.4
27.9
29.3
2020
34.0
24.3
23.1
Note: The emissions data presented here are EPA modeling results.
Table 7-17. Projected Emissions of SO2 with New Base Case (CAIR) and CAMR, without
and with Selected Technological Advances (Million Tons)
New Base
Case: CAIR
Option 1 -
Current
Technology
Option 1 -
Sorbent
Sensitivity
2010
Nationwide CAIR Region
6.1 5.1
6.1 5.1
6.1 5.1
2015
Nationwide CAIR Region
5.0 4.0
4.9 4.0
4.9 4.0
2020
Nationwide CAIR Region
4.3 3.3
4.2 3.3
4.3 3.3
Source: Integrated Planning Model run by EPA.
7-19
-------
Table 7-18. Projected Emissions of NOx with the New Base Case (CAIR) and CAMR
without and with Selected Technological Advances (Million Tons)
New Base
Case: CAIR
Option 1 —
Current
Technology
Option 1 —
Sorbent
Sensitivity
2010
Nationwide CAIR Region
2.5 1.5
2.4 1.5
2.4 1.5
2015
Nationwide CAIR Region
2.2 1.3
2.2 1.3
2.2 1.3
2020
Nationwide CAIR Region
2.2 1.3
2.2 1.3
2.2 1.3
Source: Integrated Planning Model run by EPA.
Table 7-19. Annualized Private Compliance Cost and Present Value Cost Incremental to
the New Base Case (CAIR) ($1999)
Cost (billions)
Option 1 - Current
Technology
Option 1 - Sorbent
Sensitivity
2010
$0.16
$0.10
2015
$0.10
$0.04
2020
$0.75
$0.56
Present value (2007-2025)
$3.9
$2.2
Note: Annual incremental costs of CAIR are $2.4 billion in 2010, $3.6 billion in 2015, and $4.4 billion in 2020, present value
(2007-2025) is $41.1 billion.
Note: Numbers rounded to the nearest hundred million for annualized cost.
Source: Integrated Planning Model run by EPA.
Table 7-20. Marginal Cost of Hg, SO2, and NOX Reductions with CAMR without and with
Selected Technological Advances ($1999)
New Base
Case: CAIR
Option 1 -
Current
Technology
Option 1 -
Sorbent
Sensitivity
SO2 ($/ton)
NOx ($/ton)
SO2 ($/ton)
NOx ($/ton)
Hg ($/lb)
SO2 ($/ton)
NOx ($/ton)
Hg (S/lb)
2010
$800
$1,300
$700
$1,200
$23,200
$800
$1,200
$11,800
2015
$1,000
$1,600
$900
$1,500
$30,100
$1,000
$1,500
$15,300
2020
$1,300
$1,600
$1,200
$1,300
$39,000
$1,300
$1,400
$19,900
Note: Numbers rounded to the nearest hundred for marginal cost.
Source: Integrated Planning Model run by EPA.
7-20
-------
Table 7-21. Pollution Controls by Technology with the New Base Case (CAIR), and CAMR
without and with Selected Technological Advances (GW)
New Base
Case: CAIR
Option 1 -
Current
Technology
Option 1 -
Sorbent
Sensitivity
2010
FGD
146
146
146
SCR
125
126
126
ACI
—
2
1
2015
FGD
177
179
179
SCR
151
153
153
ACI
—
3
3
2020
FGD
198
199
197
SCR
153
156
155
ACI
0.5
13
25
Note: Retrofits include existing scrubbers and SCR as well as additional retrofits for the Title IV Acid Rain Program, the
NOX SIP call, NSR settlements, and various state rules.
Source: Integrated Planning Model run by EPA.
Table 7-22. Generation Mix with the Old Base Case, the New Base Case (CAIR), and with
CAMR without and with Selected Technological Advances (Thousand GWhs)
Old Base Case
New Base
Case: CAIR
Option 1 -
Current
Technology
Option 1—
Sorbent
Sensitivity
Coal
Oil/Natural Gas
Other
Coal
Oil/Natural Gas
Other
Coal
Oil/Natural Gas
Other
Coal
Oil/Natural Gas
Other
2010
2,198
111
1,223
2,165
807
1,217
2,160
812
1,216
2,161
811
1,217
2015
2,195
1,072
1,233
2,197
1,069
1,232
2,194
1,072
1,233
2,196
1,070
1,233
2020
2,410
1,221
1,218
2,384
1,247
1,217
2,365
1,265
1,217
2,372
1,258
1,217
Change From
New Base Case
in 2020
-0.8%
1.5%
0.0%
-0.5%
0.9%
0.0%
Note: Numbers may not add due to rounding.
Source: 2003 data are from EIA: Coal - 1,970; Oil/Natural Gas - 758; Other -1,120. Projections are from the Integrated
Planning Model run by EPA.
7-21
-------
Table 7-23. Total Coal and Natural Oil/Gas-Fired Capacity by 2020 (GW)
Pulverized Coal
IGCC
Oil/Gas
Current
305
0.6
395
Old Base Case
318
8
467
New Base
Case: CAIR
315
9
469
Option 1 -
Current
Technology
314
8
471
Option 1 -
Sorbent
Sensitivity
314
9
470
Source: Current data are from EPA's NEEDS 2004; projections are from the Integrated Planning Model run by EPA.
Table 7-24. Coal Production for the Electric Power Sector with the Old Base Case, New
Base Case (CAIR) , and with CAMR without and with Selected Technological Advances
(Million Tons)
Supply Area
Appalachia
Interior
West
National
2000
299
131
475
905
2003
275
135
526
936
Supply Area
Appalachia
Interior
West
National
Old Base Case
2010
325
161
603
1,089
Option
2010
303
169
589
1,061
2015
315
162
631
1,109
2020
301
173
714
1,188
1 - Current Technology
2015
310
194
568
1,071
2020
330
224
572
1,127
New
2010
306
165
607
1,078
Option 1
2010
305
168
592
1,065
Base Case:
2015
306
191
586
1,083
- Sorbent
2015
312
191
578
1,081
CAIR
2020
331
218
609
1,158
Sensitivity
2020
333
220
579
1,132
Source: 2000 and 2003 data are derived from EIA data. All projections are from the Integrated Planning Model run by
EPA.
7-22
-------
Table 7-25. Retail Electricity Prices by NERC Region with the Old Base Case, New Base
Case (CAIR), and with CAMR without and with Selected Technological Advances
(Mills/kWh) ($1999)
Option 1
Power Region
ECAR
ERCOT
MAAC
MAIN
MAPP
NY
NE
FRCC
STV
SPP
PNW
RM
CALI
National
Primary States Included
3H, MI, IN, KY, WV, PA
rx
PA,NJ,MD,DC,DE
IL, MO, WI
MM, IA,SD,ND,NE
NY
VT, NH, ME, MA, CT, RI
FL
VA, NC, SC, GA, AL, MS, TN, AR, LA
KS, OK, MO
WA, OR, ID
MT, WY, CO, UT, NM, AZ, NV, ID
CA
Contiguous Lower 48 States
2000
57.4
65.1
80.4
61.2
57.4
104.3
89.9
67.9
59.3
59.3
45.9
64.1
94.7
66.C
Base Case
2010 2015 2020
51.7 55.2 56.1
57.9 64.4 62.6
59.3 69.4 72.2
52.6 57.8 61.0
52.8 49.3 47.6
82.8 87.9 88.1
77.4 83.9 82.8
71.2 71.3 69.5
56.2 55.1 55.3
54.2 57.0 56.7
49.6 47.4 46.9
63.9 65.2 64.7
97.1 98.9 99.3
60.3 63.1 63.4
CAIR
2010 2015 2020
53.7 58.6 58.C
59.4 64.5 63.3
61.0 72.0 72.7
53.9 60.4 62.0
52.9 49.6 48.C
83.3 88.9 88.5
77.4 84.7 83.1
71.7 72.3 70.5
57.0 56.2 56.6
54.6 57.5 57.0
49.8 47.5 46.9
64.1 65.6 65.4
97.3 99.1 99.5
61.3 64.5 64.3
Option 1
2010 2015 2020
53.9 58.7 58.1
59.1 64.9 63.4
61.3 72.1 72.9
54.1 60.5 62.3
53.0 49.6 48.2
83.3 89.1 88.7
77.5 84.8 83.1
71.8 72.3 70.7
57.1 56.2 56.8
54.6 57.5 57.2
49.8 47.5 47.1
64.2 65.6 65.1
97.3 99.1 99.7
61.3 64.5 64.5
Change
from
CAIR
2020
0.2%
0.1%
0.2%
0.5%
0.5%
0.4%
0.0%
0.3%
0.3%
0.3%
0.5%
-0.4%
0.3%
0.2%
Sorbent Sensitivity
Power Region
ECAR(l)
ERCOT (2)
MAAC (3)
MAIN (4)
MAPP (5)
NY (6)
NE(7)
FRCC (8)
STV (9)
SPP (10)
PNW (11)
RM(12)
CALI (13)
National
Primary States Included
OH, MI, IN, KY, WV, PA
rx
PA, NJ, MD, DC, DE
[L, MO, WI
MN, IA, SD, ND, NE
NY
VT, NH, ME, MA, CT, RI
FL
VA, NC, SC, GA, AL, MS, TN, AR, LA
KS, OK, MO
WA, OR, ID
MT, WY, CO, UT, NM, AZ, NV, ID
CA
Contiguous Lower 48 States
2000
57.4
65.1
80.4
61.2
57.4
104.3
89.9
67.9
59.3
59.3
45.9
64.1
94.7
66.C
Base Case
2010 2015 2020
51.7 55.2 56.1
57.9 64.4 62.6
59.3 69.4 72.2
52.6 57.8 61.0
52.8 49.3 47.6
82.8 87.9 88.1
77.4 83.9 82.8
71.2 71.3 69.5
56.2 55.1 55.3
54.2 57.0 56.7
49.6 47.4 46.9
63.9 65.2 64.7
97.1 98.9 99.3
60.3 63.1 63.4
CAIR
2010 2015 2020
53.7 58.6 58.0
59.4 64.5 63.3
61.0 72.0 72.7
53.9 60.4 62.0
52.9 49.6 48.0
83.3 88.9 88.5
77.4 84.7 83.1
71.7 72.3 70.5
57.0 56.2 56.6
54.6 57.5 57.0
49.8 47.5 46.9
64.1 65.6 65.4
97.3 99.1 99.5
61.3 64.5 64.3
Sensitivity
2010 2015 2020
53.8 58.6 58.0
59.2 64.9 63.4
61.1 72.0 72.9
54.0 60.5 62.3
52.9 49.6 48.0
83.3 89.1 88.8
77.5 85.0 83.0
71.8 72.3 70.7
57.1 56.2 56.7
54.6 57.5 57.2
49.8 47.5 47.2
64.1 65.6 65.3
97.2 99.1 99.7
61.3 64.5 64.4
Change
from
CAIR
2020
0.1%
0.2%
0.2%
0.4%
0.1%
0.3%
-0.1%
0.2%
0.3%
0.3%
0.5%
-0.2%
0.22%
0.2%
Source: Retail Electricity Price Model run by EPA. 2000 prices from EIA's AEO 2003.
7.15 Sensitivity Analysis on Assumptions for Natural Gas Prices and Electricity Growth
Sensitivity analyses were performed using projections from the 2004 Annual Energy
Outlook produced by the Energy Information Administration (EIA). EPA used EIA estimates
for the difference between natural gas prices and coal prices, which we have short-handed as
"EIA natural gas prices," as well as EIA's projection of electricity growth. These particular
assumptions involve considering the higher differential between minemouth coal and wellhead
natural gas prices. For the years 2010, 2015, and 2020, there was a higher differential of $0.25
mmBtu, $0.42 mmBtu, and $0.38 mmBtu, respectively. The electricity growth was changed to
match EIA's growth of 1.8 percent a year rather than EPA's growth of 1.6 percent.
Nationwide emissions of Hg, SO2, and NOx using EIA assumptions are presented in
Tables 7-26 and 7-27. Mercury emissions profiles with EIA assumptions are similar and lower
than emissions with EPA assumptions. Lower Hg emissions for EIA assumptions can be
attributed to the building of new and cleaner coal-fired capacity.
7-23
-------
Total annual costs and present value costs of CAMR incremental to CAIR with EIA
assumptions are in Table 7-28. The costs of C AMR with EIA assumptions for natural gas prices
and electricity growth in 2010 and 2015 are only slightly different from costs of CAIR without
those assumptions and can be attributed to the building of new and cleaner coal-fired capacity
that leads to lower overall costs (see Tables 7-28 and 7-29). As demand continues to grow, coal-
fired generation continues to increase and requires the use of additional scrubbers. Although
more pollution controls are installed using EIA assumptions, dispatch changes lead to the use of
more efficient generation. The power sector is less inclined to use gas as a compliance option in
the region because of the higher operating cost. Once the power sector passes the point where
there is no longer excess gas capacity in the marketplace (as currently exists), new coal-fired
capacity is the logical choice to meet demand.
Coal-fired generation under CAMR increases using EIA assumptions for natural gas
prices and electricity growth. Table 7-31 shows the generation mix with EIA assumptions. Coal
production patterns change slightly and production for all three major coal-producing regions is
higher, because coal-fired generation is a cheaper source of electricity than natural gas in most
parts of the country with the higher EIA prices, even as more pollution controls are added to
coal-fired generation and used to meet the additional electricity demand (see Table 7-32).
Electricity prices are not greatly altered with EIA assumptions for natural gas and
electricity growth (see Table 7-33). Average electricity prices are projected to be lower than
current levels (2000) using both EPA and EIA assumptions for natural gas and electricity
growth.
Table 7-26. Projected Emissions of Hg for the New Base Case (CAIR) and CAMR with
EPA and EIA Assumptions for Natural Gas Prices and Electric Growth (Tons)
Old Base Case
New Base
Case: CAIR
Option 1
EPA Assumptions
EIA Assumptions
EPA Assumptions
EIA Assumptions
EPA Assumptions
EIA Assumptions
2010
46.6
47.5
38.0
38.3
31.3
31.5
2015
45.0
47.0
34.4
35.2
27.9
28.5
2020
46.2
47.8
34.0
35.4
24.3
23.5
Note: The emissions data presented here are EPA modeling results.
Table 7-27. Projected Nationwide Emissions of SO2 and NOX under the New Base Case
(CAIR) and CAMR with EPA and EIA Assumptions for Natural Gas and Electric Growth
(Million Tons)
7-24
-------
Old Base Case with EPA Assumptions
New Base Case (CAIR) with EPA
Assumptions
Option 1 with EPA Assumptions
Old Base Case with EIA Assumptions
New Base Case (CAIR) with EIA
Assumptions
Option 1 with EIA Assumptions
2010
9.7
6.1
6.1
9.7
6.1
6.1
SO2
2015
8.9
5.0
4.9
8.8
5.0
4.9
2020
8.6
4.3
4.2
8.6
4.0
4.3
NOx
2010
3.6
2.5
2.4
3.7
2.4
2.4
2015
3.7
2.2
2.2
3.8
2.1
2.2
2020
3.7
2.2
2.2
3.8
2.2
2.2
Source: Integrated Planning Model run by EPA.
Table 7-28. Annualized Cost and Present Value Cost Incremental to the New Base Case
(CAIR) with EPA and EIA Assumptions for Natural Gas Prices and Electric Growth
(Billion $1999)
Option 1- EPA Assumptions
Option 1 - EIA Assumptions
2010
$0.16
$0.16
2015
$0.10
$0.21
2020
$0.75
$0.53
Present value (2007-2025)
$3.9
$3.1
Note: Annual incremental costs of CAIR with EPA assumptions are $2.4 billion in 2010, $3.6 billion in 2015, and $4.4 billion
in 2020, present value (2007-2025) is $41.1 billion. Annual incremental costs of CAIR with EIA assumptions are $2.6 billion
in 2010, $3.4 billion in 2015, and $4.1 billion in 2020, present value (2007-2025) is $42.9 billion.
Note: Numbers rounded to the nearest tenth million for annualized cost.
Source: Integrated Planning Model run by EPA.
Table 7-29. Marginal Cost of SO2 and NOX Reductions under the New Base Case (CAIR)
and CAMR with EPA and EIA Assumptions for Natural Gas Prices and Electric Growth
($/ton, in $1999)
2010
2015
2020
New Base
Case: CAIR
Option 1
SO2
NOx
SO2
NOx
Hg
EPA Assumptions
EIA Assumptions
EPA Assumptions
EIA Assumptions
EPA Assumptions
EIA Assumptions
EPA Assumptions
EIA Assumptions
EPA Assumptions
EIA Assumptions
$800
$800
$1,300
$1,400
$700
$800
$1,200
$1,200
$23,200
$26,400
$1,000
$1,200
$1,600
$1,700
$900
$1,000
$1,500
$1,600
$30,100
$34,200
$1,300
$1,500
$1,600
$1,700
$1,200
$1,300
$1,200
$1,300
$39,000
$44,400
Source: Integrated Planning Model run by EPA.
7-25
-------
Table 7-30. Pollution Controls under the New Base Case (CAIR) with EPA and EIA
Assumptions for Natural Gas and Electricity Growth (GWs)
New Base Case:
CAIR
Option 1
Technology
FGD
SCR
FGD
SCR
ACI
2010
146
125
146
126
2
EPA Assumptions
2015
177
151
179
153
3
2020
198
153
199
156
13
EIA
2010
157
134
155
137
3
Assumptions
2015 2020
185 209
161 162
187 203
160 162
4 26
Note: Retrofits include existing scrubbers and SCR as well as additional retrofits for the Title IV Acid Rain Program, the
NOX SIP call, NSR settlements, and various state rules.
Source: Integrated Planning Model run by EPA.
Table 7-31. Generation Mix under the New Base Case (CAIR) and CAMR with EPA and
EIA Assumptions for Natural Gas and Electric Growth (Thousand GWhs)
Old Base
Case
New Base
Case:
CAIR
Option 1
Fuel
Coal
Oil/Natural Gas
Other
Total
Coal
Oil/Natural Gas
Other
Total
Coal
Oil/Natural Gas
Other
Total
2010
2,198
111
1,223
4,198
2,165
807
1,217
4,190
2,160
812
1,216
4,188
EPA Assumptions
2015
2,242
1,026
1,235
4,503
2,197
1,069
1,232
4,498
2,194
1,072
1,233
4,499
2020
2,410
1,221
1,218
4,850
2,384
1,247
1,217
4,848
2,365
1,265
1,217
4,847
EIA Assumptions
2010
2,243
902
1,224
4,369
2,228
916
1,223
4,367
2,221
922
1,222
4,366
2015
2,638
867
1,235
4,739
2,632
871
1,234
4,738
2,616
887
1,235
4,738
2020
3,048
873
1,224
5,145
3,045
874
1,221
5,141
3,014
904
1,219
5,138
Note: Numbers may not add due to rounding.
Source: Integrated Planning Model run by EPA.
Table 7-32. Coal Production for the Electric Power Sector under the New Base Case
(CAIR) and CAMR with EPA and EIA Assumptions for Natural Gas and Electricity
Growth (Million Tons)
Supply
Area 2000 2003
EPA Assumptions
2010 2015 2020
EIA Assumptions
2010 2015 2020
7-26
-------
Old Base
Case
New
Base
Case:
CAIR
Option 1
Appalachia
Interior
West
National
Appalachia
Interior
West
National
Appalachia
Interior
West
National
299
131
475
905
299
131
475
905
299
131
475
905
275
135
526
936
275
135
526
936
275
135
526
936
325
161
603
1,089
306
164
607
1,077
303
169
589
1,061
315
162
631
1,109
310
193
579
1,082
310
194
568
1,071
301
173
714
1,188
331
219
607
1,156
330
224
572
1,127
328
161
626
1,115
320
174
614
1,109
317
179
595
1,091
341
182
748
1,271
367
207
676
1,250
377
209
639
1,225
340
247
840
1,428
390
260
765
1,415
396
269
706
1,371
Source: 2000 and 2003 data are from EIA. All projections are from the Integrated Planning Model run by EPA.
Table 7-33. Retail Electricity Prices by NERC Region for the Base Case (No Further
Controls), CAIR, and CAMR with EPA and EIA Assumptions for Natural Gas and
Electricity Growth (Mills/kWh) ($1999)
EPA Assumptions for Natural Gas and Electricity Growth
Power Region
ECAR(l)
ERCOT (2)
MAAC (3)
MAIN (4)
MAPP (5)
NY (6)
NE(7)
FRCC (8)
STV (9)
SPP(IO)
PNW(ll)
RM(12)
CALI(13)
National
Primary States Included
OH, MI, IN, KY, WV, PA
rx
PA, NJ, MD, DC, DE
1L, MO, WI
MN, IA, SD, ND, NE
NY
VT, NH, ME, MA, CT, RI
FL
VA, NC, SC, GA, AL, MS, TN, AR, LA
KS, OK, MO
WA, OR, ID
MT, WY, CO, UT, NM, AZ, NV, ID
CA
Contiguous Lower 48 States
2000
57.4
65.1
80.4
61.2
57.4
104.3
89.9
67.9
59.3
59.3
45.9
64.1
94.7
66.0
Base Case
2010 2015 2020
51.7 55.2 56.1
57.9 64.4 62.6
59.3 69.4 72.2
52.6 57.8 61.0
52.8 49.3 47.6
82.8 87.9 88.1
77.4 83.9 82.8
71.2 71.3 69.5
56.2 55.1 55.3
54.2 57.0 56.7
49.6 47.4 46.9
63.9 65.2 64.7
97.1 98.9 99.3
60.3 63.1 63.4
CAIR
2010 2015 2020
53.8 58.5 58.0
59.3 64.6 63.3
61.2 71.7 72.8
54.0 60.3 62.0
52.9 49.6 48.0
83.3 88.8 88.4
77.5 84.7 83.0
71.7 72.3 70.5
57.0 56.2 56.6
54.6 57.5 57.0
49.8 47.5 46.9
64.1 65.6 65.4
97.3 99.1 99.5
61.3 64.4 64.3
Option 1
2010 2015 2020
53.9 58.7 58.1
59.1 64.9 63.4
61.3 72.1 72.9
54.1 60.5 62.3
53.0 49.6 48.2
83.3 89.1 88.7
77.5 84.8 83.1
71.8 72.3 70.7
57.1 56.2 56.8
54.6 57.5 57.2
49.8 47.5 47.1
64.2 65.6 65.1
97.3 99.1 99.7
61.3 64.5 64.5
Change
from
CAIR
2020
0.2%
0.1%
0.2%
0.5%
0.5%
0.4%
0.2%
0.3%
0.3%
0.3%
0.5%
-0.4%
0.3%
0.2%
EIA Assumptions for Natural Gas and Electricity Growth
Power Region
ECAR(l)
ERCOT (2)
MAAC (3)
MAIN (4)
MAPP (5)
NY (6)
NE(7)
FRCC (8)
STV (9)
SPP(IO)
PNW(ll)
RM(12)
CALI(13)
Primary States Included
OH, MI, IN, KY, WV, PA
rx
PA, NJ, MD, DC, DE
IL, MO, WI
MN, IA, SD, ND, NE
NY
VT, NH, ME, MA, CT, RI
FL
VA, NC, SC, GA, AL, MS, TN, AR, LA
KS, OK, MO
WA, OR, ID
MT, WY, CO, UT, NM, AZ, NV, ID
CA
2000
57.4
65.1
80.4
61.2
57.4
104.3
89.9
67.9
59.3
59.3
45.9
64.1
94.7
Base Case
2010 2015 202C
53.5 59.8 57.1
63.3 66.0 64.4
63.1 74.7 72.8
54.9 63.8 62.4
52.9 49.6 48.1
89.0 91.3 87.8
85.1 85.5 81.2
72.5 74.6 73.7
57.1 57.1 57.1
56.2 59.5 57.9
50.4 50.0 49.9
66.0 67.9 66.6
99.5 101.4 101.8
CAIR
2010 2015 202C
55.3 61.5 58.8
63.6 66.6 65.0
64.0 75.4 73.7
55.9 65.2 63.3
53.1 49.9 48.6
89.1 91.9 88.8
84.7 85.9 81.8
73.3 75.3 74.3
57.8 58.3 58.6
56.7 59.7 58.1
50.7 50.2 49.9
66.3 68.0 66.4
99.6 101.5 101.8
Option 1
2010 2015 2020
55.5 61.7 59.2
63.5 66.9 65.2
64.0 75.6 73.7
56.0 65.2 63.5
53.1 50.0 48.9
89.1 91.7 89.0
84.6 86.0 82.5
73.4 75.5 74.4
57.8 58.2 58.8
56.7 59.9 58.7
50.3 49.9 49.5
66.3 68.2 66.9
99.9 101.8 102.0
Change
from
CAIR
2020
0.6%
0.3%
0.0%
0.3%
0.6%
0.2%
0.8%
0.0%
0.4%
1.0%
-0.7%
0.8%
0.2%
7-27
-------
National [Contiguous Lower 48 States I 66.CJ 62.8 66.1 64.9| 63.5 67.0 6S.g| 63.6 67.1 66. l| 0.5%
Source: Retail Electricity Price Model run by EPA. 2000 prices from EIA's AEO 2003.
7.16 Small Entity Impacts
The Regulatory Flexibility Act (5 U.S.C. § 601 et seq.), as amended by the Small
Business Regulatory Enforcement Fairness Act (Public Law No. 104-121), provides that
whenever an agency is required to publish a general notice of proposed rulemaking, it must
prepare and make available an initial regulatory flexibility analysis, unless it certifies that the
proposed rule, if promulgated, will not have "a significant economic impact on a substantial
number of small entities" (5 U.S.C. § 605[b]). Small entities include small businesses, small
organizations, and small governmental jurisdictions.
For the purposes of assessing the impacts of CAMR on small entities, a small entity is
defined as:
(1) A small business according to the Small Business Administration size standards
by the North American Industry Classification System (NAICS) category of the
owning entity. The range of small business size standards for electric utilities is 4
billion kilowatt-hours of production or less;
(2) a small government jurisdiction that is a government of a city, county, town,
district, or special district with a population of less than 50,000; and
(3) a small organization that is any not-for-profit enterprise that is independently
owned and operated and is not dominant in its field.
Table 7-34 lists entities potentially affected by this proposed rule with applicable NAICS
code.
Table 7-34. Potentially Regulated Categories and Entities'
Category
Industry
Federal
Government
State/Local/
Tribal
Government
NAICS
Code"
221112
221112°
221112"
921150
Examples of Potentially Regulated Entities
Coal-fired electric utility steam generating units.
Coal-fired electric utility steam generating units owned by the
federal government.
Coal-fired electric utility steam generating units owned by
municipalities.
Coal-fired electric utility steam generating units in Indian Country.
Include NAICS categories for source categories that own and operate electric generating units only.
North American Industry Classification System.
Federal, state, or local government-owned and operated establishments are classified according to the activity in
which they are engaged.
7-28
-------
Courts have interpreted the RFA to require a regulatory flexibility analysis only when
small entities will be subject to the requirements of the rule.12 In the January 30, 2004 Notice of
Proposed Rulemaking (NPR) EPA determined that the proposed rule would not have a
significant impact on a substantial number of small entities. However, to provide additional
information to States and affected sources, EPA conduct a general analysis of the potential
economic impact of CAMR on small entities.
EPA examined the potential economic impacts to small entities associated with this
rulemaking based on assumptions of how the affected states will implement control measures to
meet their NOX and SO2 budgets under the Clean Air Interstate Rule (CAIR) and their Hg
budgets for EGUs under CAMR. Under CAMR, States have the option of either participating in
an EPA-run trading program, or implementing their Hg budget as a strict cap on Hg emissions
from EGUs. This analysis assumes that all affected States in the CAIR region choose to meet
their CAIR budgets by controlling EGUs only, and that all States participate in the nationwide
Hg cap-and-trade program. This analysis does not examine potential indirect economic impacts
associated with CAIR or CAMR, such as employment effects in industries providing fuel and
pollution control equipment, or the potential effects of electricity price increases on industries
and households. Because CAMR is implemented in conjunction with CAIR, the costs of
CAMR are measured incrementally to the costs of CAIR alone.
This analysis presents the annualize cost of CAMR for the year 2020, which is two years
into the second phase of the Hg cap-and-trade program, and for which the Hg emission cap is 15
tons. An important caveat to note in considering the results presented in this section is (as
discussed earlier in this chapter) that EPA assumes no development in control technologies over
the course of the Hg cap-and-trade program. In reality, Hg emissions control is a fast moving
area with new developments nearly monthly. Actual costs may be lower than those presented
since modeling assumes no improvements in the cost of mercury control technology, while in
reality, control costs are expected to improve over time. As a result, this the projected costs of
the Hg cap-and-trade program for 2020 presented in this analysis most certainly overstate the
impact of the rule on small entities during the second phase of the program. At the same time,
however, the marginal cost projected for mercury control in 2020 may also overstate the cost-
savings that entities selling allowances may experience under the rule. Finally, it should be
noted that during the first phase of the program, the fact that the cap is equal to co-benefits under
CAIR should limit the impact of CAMR on small entities.
7.16.1 Identification of Small Entities
EPA used EGRID data as a basis for compiling the list of potentially affected small
entities. EGRID is EPA's Emissions & Generation Resource Integrated Database, which
contains emissions and resource mix data for virtually every power plant and company that
generates electricity in the United States.13 The data set contains detailed ownership and
12 See Michigan v. EPA. 213 F.3d 663, 668-69 (D.C. Cir. 2000), cert, den. 121 S.Ct. 225, 149 L.Ed.2d 135 (2001).
An agency's certification need consider the rule's impact only on entities subject to the rule.
13 eGRID is available at http://www.epa.gov/cleanenergy/egrid/download.htm.
7-29
-------
corporate affiliation information. For plants burning coal as the primary fuel, plant-level boiler
and generator capacity, heat input, generation, and emissions data were aggregated by owner and
then parent company. Entities with more than 4 billion kWh of annual electricity generation
were removed from the list, as were municipal-owned entities serving a population greater than
50,000. Finally, for cooperatives, investor-owned utilities, and subdivisions that generate less
than 4 billion kWh of electricity annually but may be part of a large entity, additional research on
power sales, operating revenues, and other business activities was performed to make a final
determination regarding size. Because the rule does not affect units with a generating capacity
of 25 MW or less, small entities that do not own at least one coal-fired generating unit with a
capacity greater than 25 MW were dropped from the data set. According to EPA's analysis,
approximately 35 small entities were exempted by this provision. EPA identified a total of 81
potentially affected small entities, out of a possible 116. The number of potentially affected
small entities by ownership type, and summary of projected impacts, is listed in Table 7-35.
Table 7-35. Projected Impact of CAMR on Small Entities
ECU
Ownership
Type
Cooperative
Investor-
Owned Utility
Municipal
Subdivision
Other
Total
Number of
Potentially
Affected
Entities
21
2
48
8
1
80
Total Net
Compliance Cost
in 2020
Incremental to
CAIR ($1999
millions)
8.5
6.4
15.2
6.3
-0.003
36.5
Number of Small
Entities with
Compliance Costs
>1% of Generation
Revenues in 2020
7
2
28
5
0
42
Number of Small
Entities with
Compliance Costs
>3% of Generation
Revenues in 2020
1
0
11
2
0
14
Note: The total number of potentially affected entities in this table excludes the 35 entities that have been dropped
because they will not be affected by CAMR. Also, the total number of entities with costs greater than
1 percent or 3 percent of revenues includes only entities experiencing positive costs.
Source: IPM and TRUM analysis
7.16.2 Overview of Analysis and Results
This section presents the methodology and results for estimating the impact of CAMR to
small entities in 2020 based on the following endpoints:
• annual economic impacts of CAMR on small entities and
• ratio of small entity impacts to revenues from electricity generation.
7.16.2.1 Methodology for Estimating Impacts of CAMR on Small Entities
7-30
-------
An entity can comply with CAMR through some combination of the following: installing
retrofit technologies, purchasing allowances, switching to a lower Hg fuel, or reducing emissions
through a reduction in generation. Additionally, units with more allowances than needed can sell
these allowances on the market. The chosen compliance strategy will be primarily a function of
the unit's marginal control costs and its position relative to the marginal control costs of other
units. Because CAMR will be implemented in conjunction with CAIR, units affected by both
rules will attempt to minimize their cost of compliance over both rules, by considering Hg, SO2,
and NOX control strategies simultaneously.
To attempt to account for each potential control strategy over the combined rules, EPA
estimates compliance costs as follows:
^Compliance ~ " ^Operaling+RetroJit ~"~ " ^Fuel ~"~ " ^Allowances ~"~ " ^Transaction ~ " "- (°-l)
where C represents a component of cost as labeled, and A R represents the retail value of
foregone electricity generation.
In reality, compliance choices and market conditions can combine such that an entity
may actually experience a savings in any of the individual components of cost. Under CAIR and
CAMR, for example, EPA projects that the price of low-sulfur coal will fall as many units install
scrubbers and switch away from low-sulfur coal to cheaper bituminous coal, such that many
entities actually experience a reduction in fuel costs relative to the base caes as a result of lower
prices due to the demand shift. Similarly, although some units will forgo some level of
electricity generation (and thus revenues) to comply, this impact will be lessened on these
entities by the projected increase in electricity prices under CAIR and CAMR as well as
reductions in fuel costs, and those not reducing generation levels will see an increase in
electricity revenues. Elsewhere, unscrubbed units burning low-sulfur coal might find it most
economical to install mercury-specific controls such as ACI, and sell their surplus of Hg
allowances on the market. Because this analysis evaluates the total costs along each of the four
compliance strategies laid out above for each entity, it inevitably captures savings or gains such
as those described. As a result, what we describe as cost is really more of a measure of the net
economic impact of the rule on small entities.
For this analysis, EPA used IPM-parsed output to estimate net compliance costs at the
unit level. These impacts were then summed for each small entity, adjusting for ownership
share. Net impact estimates were based on the following: operating and retrofit costs, sale or
purchase of allowances, and the change in fuel costs or electricity generation revenues under
CAMR relative to CAIR. These individual components of compliance cost were estimated as
follows:
(1) Operating and retrofit costs: Using the IPM-parsed output for the base case,
CAIR, and CAMR (available in the docket), EPA identified units that install
control technology under CAIR and CAMR and the technology installed. The
equations for calculating retrofit costs were adopted from EPA's Technology
Retrofit and Updating Model (TRUM). The model calculates the capital cost (in
$/MW); the fixed operation and maintenance (O&M) cost (in $/MW-year); the
7-31
-------
variable O&M cost (in $/MWh); and the total annualized retrofit cost for units
projected to install FGD, SCR, SNCR, or ACL
(2) Sale or purchase of allowances: EPA estimated the value of initial SO2, NOX,
and Hg allowance holdings. For SO2, units were assumed to retain their Phase II
allowance allocations as determined under EPA's 1998 reallocation of Acid Rain
allowances, adjusted to reflect the 50 percent reduction in 2010 and 65 percent
reduction in 2015 under CAIR. Because of the resources involved in compiling
allowance-holding data, the value of banked SO2 allowances was not considered
in this analysis. The implication of this is that the annual net purchase of
allowances may be overstated for some units. For NOX, the state emission
budgets were assumed to be apportioned to units on a heat-input basis. Each unit
was assumed to receive a share of the state NOX emission budget equal to its share
of the total state heat input for that year in the base case. This is a simplification
of what is included in the model rule, which proposes allocating NOX allowances
based on heat input from 1999-2002.14 However, states can ultimately decide
how to allocate NOX allowances. For Hg, unit allocations were the same as those
listed in the March 16, 2004 Supplemental Notice of Proposed Rulemaking.
To estimate the value of allowances holdings, allocated NOx and SO2 allowances
were subtracted from projected emissions, and the difference was then multiplied
by the allowance prices projected by IPM for 2020. Units were assumed to
purchase or sell allowances to exactly cover their projected NOX and SO2
emissions under CAIR + CAMR. For Hg, units that did not have allowances
sufficient to cover projected 2020 emissions were projected to withdraw
allowances from their respective Hg allowance banks if available, or else
purchase the required amount of allowances. Units holding 2020 allowances in
excess of projected 2020 emissions were projected to sell these excess
allowances. The estimation of the size of a unit's mercury allowance bank is
discussed further below.
(3) Fuel costs: Fuel costs were estimated by multiplying fuel input (MMBtu) by
region and fuel-type-adjusted fuel prices ($/MMBtu) from TRUM. The change in
fuel expenditures under CAMR was then estimated by taking the difference in
fuel costs between CAMR and CAIR.
(4) Value of electricity generated: EPA estimated electricity generation by first
estimating unit capacity factor and maximum fuel capacity. Unit capacity factor
is estimated by dividing fuel input (MMBtu) by maximum fuel capacity
(MMBtu). The maximum fuel capacity was estimated by multiplying capacity
(MW) * 8,760 operating hours * heat rate (MMBtu/MWh). The value of
electricity generated is then estimated by multiplying capacity (MW)*capacity
factor*8,760*regional-adjusted retail electricity price ($/MWh).
14 A similar approach was used in regulatory impact analyses for the 126 FIP and NO, SIP Call.
7-32
-------
As discussed later in this analysis, many small entities projected to be affected by
CAMR do not have to operate in a competitive market environment and thus
should be able to pass compliance costs on to consumers. To somewhat account
for this, we incorporated the projected regional-adjusted retail electricity price
calculated under CAMR in our estimation of generation revenue under CAMR.
(5) Administrative costs: Because most affected units are already monitored as a
result of other regulatory requirements, EPA considered the primary
administrative cost to be transaction costs related to purchasing or selling
allowances. EPA assumed that transaction costs were equal to 1.5 percent of the
total absolute value of a unit's allowances. This assumption is based on market
research by ICF Consulting.
(6) Value of the Mercury Bank: EPA's economic analysis of CAMR suggests that a
significant bank of approximately 70 tons of Hg allowances will be built up
during the first phase of the cap-and-trade program. Sources will be relying
heavily on this bank for compliance during the second phase of the program.
While not all sources will have banked allowances during the first phase of the
program, many sources will be able to draw from this bank during the second
phase and avoid or limit Hg allowance purchases. EPA estimated the size of the
bank by comparing projected emissions for the years 2010-2019 with allocations
for those years. This estimate assumed that small entity sources with surplus
allowances in those years would bank those allowances rather than sell them on
the market, and would draw from this bank in any year that they were short
allowances. EPA estimated the cost of using banked allowances by taking the
average cost of Hg control in the first phase of the program discounted to 2020,
multiplied by the number of banked allowances used. Finally, any surplus
allowances remaining in the small entity banks in 2020 were valued at the 2020
Hg allowance price.
7.16.2.2 Results
The potential impacts of CAMR on small entities are summarized in Table 7-35. All
costs are presented in $1999. EPA estimated the incremental annualized net compliance cost to
small entities to relative to CAIR to be approximately $37 million in 2020. This cost is driven
largely by mercury allowance purchases and additional retrofits relative to CAIR. The costs to
small entities in 2020 are limited, however, by the ability of approximately 30 of the 81 small
entities to sell surplus 2020 and/or banked allowances in 2020.
EPA does not project that any coal-fired generation would be uneconomic to maintain
relative to CAIR. This finding suggests that the extent of CAMR's adverse economic impacts
beyond CAIR on small entities is limited.
EPA further assessed the economic and financial impacts of the rule using the ratio of
compliance costs to the value of revenues from electricity generation, focusing in particular on
entities for which this measure is greater than 1 percent. Although this metric is commonly used
in EPA impact analyses, it makes the most sense when as a general matter an analysis is looking
7-33
-------
at small businesses that operate in competitive environments. However, small businesses in the
electric power industry often operate in a price-regulated environment where they are able to
recover expenses through rate increases. Given this, EPA considers the 1 percent measure in this
case a crude measure of the price increases these small entities will be asking of rate
commissions or making at publicly owned companies.
Of the 80 small entities considered in this analysis, and 116 total small entities in the
affected region 42 were projected to have compliance costs greater than 1% of revenues, while
14 were projected to have compliance costs greater than 3% of revenues. As was emphasized
earlier, this result is largely due to the magnitude of the projected marginal Hg control cost in
2020. A marginal cost similar to what was projected in the sensitivity analysis discussed earlier
in this chapter would eliminate significant impacts. Furthermore, the majority of small entities
in this analysis operate in a competitive market and thus should be able to recover their costs of
complying with CAMR. It should also be emphasized that under CAMR, states, through their
choice of Hg allowance allocation methodologies, can potentially mitigate adverse affects of
CAIR on small entities.
The distribution across entities of economic impacts as a share of base case revenue is
summarized in Table 7-36. Although the distributions of economic impacts on each ownership
type are in general fairly tight. Entities with the lowest negative net impacts are those that have
complied with the Hg rule without additional retrofits, and have a number of surplus Hg
allowances for sale. On average, the impact of the rule on small entities is less than 1% of
electricity generation revenues.
Table 7-36. Summary of Distribution of Economic Impacts of CAIR on Small Entities
ECU Ownership Type
Cooperative
Investor-owned utility
Municipal
Subdivision
Other
All
Capacity-Weighted
Average Economic
Impacts as a % of
Generation Revenues
0.52 %
1.94%
1.21 %
1.54%
-0.09 %
0.96 %
Min
-6.30 %
1.48%
-5.30 %
-0.52 %
-0.09 %
-6.30 %
Max
4.7 %
2.22 %
6.39 %
3.31%
-0.09 %
6.4%
Source: IPM and TRUM analysis
In the cases where entities are projected to experience positive net impacts that are a high
percentage of revenues, these entities generally have a shortage of Hg allowances and must
7-34
-------
purchase them on the market at the projected 2020 price. Many of these entities also reduce
generation slightly, and thus generation revenues, relative to CAIR alone.
The separate components of annualized costs to small entities under CAIR and CAIR +
CAMR are summarized in Table 7-37. Under CAMR, allowance purchases, driven largely by
the marginal cost projected for Hg in 2020, as well as additional retrofits, are the most significant
components of compliance cost for small entities in 2020. Also, fuel costs under for all groups
with the exception of lOUs increase relative to CAIR, largely because of an increased demand
for bituminous coal and the resulting higher bituminous coal price relative to CAIR. Retrofit
and operating costs for subdivisions, municipals, and cooperatives increase significantly, largely
because of the installation of FGD, SCR and ACL Finally, all groups with the exception of
lOUs experience an increase in electricity revenues relative to CAIR alone. This increase is
largely driven by increases in the retail price of electricity relative to CAIR alone, although a few
units are projected to increase generation under CAIR + CAMR. The two lOUs in this analysis
experience an increased revenue loss that results largely from generation reductions relative to
CAIR in 2020.
Table 7-37. Incremental Annualized Costs under CAMR relative to CAIR, Summarized by
Ownership Group and Cost Category ($1,000,000)
ECU
Ownership
Type
Cooperative
IOU
Municipal
Subdivision
Other
Retrofit +
Operating
Cost
4.9
-0.1
6.2
5.3
0
Net Purchase
of Allowances
2.9
4.1
10.3
0.8
0
Fuel Cost
2
-0.6
6.3
0.4
0
Lost
Electricity
Revenue
-1.3
3
-7.6
-0.2
-0.1
Administrative
Cost
0.1
0.1
0.1
0.1
0.001
Note: Numbers may not add to totals in Table 7-35 due to rounding.
Source: IPM and TRUM analysis.
7.16.3 Summary of Small Entity Impacts
While EPA has certified, based on earlier analysis that was summarized in the January
30, 2004 NPR, that CAMR will not have a significant impact on a substantial number of small
entities, this analysis has been conducted to provide additional understanding of the nature of
potential impacts, and additional information to the states as they propose plants to meet the
emissions budgets set by this rulemaking.
EPA projects an incremental impact on small entities relative to CAIR of approximately
$37 million relative to CAIR. EPA also projects that no additional small entity coal capacity
will be uneconomic to maintain under CAMR relative to what was projected to be uneconomic
7-35
-------
to maintain under CAIR, which is the new base case. This finding suggests that the incremental
impact of CAMR on small entities is limited.
Furthermore, of the Slsmall entities potentially affected, and the 116 small entities with
in the country with coal units included in EPA's modeling, 42 may experience compliance costs
in excess of 1 percent of revenues, while 14 are projected to experience compliance costs in
excess of 3 percent of revenues, based on our assumptions of how the affected states implement
control measures to meet their emissions budgets as set forth in this rulemaking. As is discussed
earlier in this analysis, the finding of a significant impact to some entities during the second
phase of the program is largely a product of the marginal cost projected for Hg control in 2020.
In reality, control costs of Hg are expected to be lower by 2020, such that allowance prices
would be reduced, and significant impacts unlikely. Further, the majority of these small entities
operate in cost-of-service markets where they should be able to pass on their costs of compliance
to rate-payers.
Two other points should be considered when evaluating the impact of CAMR,
specifically, and cap-and-trade programs more generally, on small entities. First, under CAIR,
the cap-and-trade program is designed such that states determine how Hg allowances are to be
allocated across units. States electing to participate in the Hg cap-and-trade program could
allocate allowances in a manner that would mitigate any potential disadvantage faced by small
entities. Further, States that chose to implement their State budget as a strict cap could provide
some level of exemption to sources owned by small entities, and require greater reductions from
other sources. Finally, it should be noted that, the use of a cap-and-trade program in general
will limit impacts on small entities relative to a less flexible command-and-control program.
7.17 Unfunded Mandates Reform Act (UMRA) Analysis
Title II of the UMRA of 1995 (Public Law 104-4)(UMRA) establishes requirements for
federal agencies to assess the effects of their regulatory actions on state, local, and Tribal
governments and the private sector. Under Section 202 of the UMRA, 2 U.S.C. 1532, EPA
generally must prepare a written statement, including a cost-benefit analysis, for any proposed or
final rule that "includes any Federal mandate that may result in the expenditure by State, local,
and Tribal governments, in the aggregate, or by the private sector, of $100,000,000 or more ... in
any one year." A "Federal mandate" is defined under Section 421(6), 2 U.S.C. 658(6), to
include a "Federal intergovernmental mandate" and a "Federal private sector mandate." A
"Federal intergovernmental mandate," in turn, is defined to include a regulation that "would
impose an enforceable duty upon State, Local, or Tribal governments," Section 421(5)(A)(i), 2
U.S.C. 658(5)(A)(i), except for, among other things, a duty that is "a condition of Federal
assistance," Section 421(5)(A)(i)(I). A "Federal private sector mandate" includes a regulation
that "would impose an enforceable duty upon the private sector," with certain exceptions,
Section 421(7)(A), 2 U.S.C. 658(7)(A).
Before promulgating an EPA rule for which a written statement is needed under Section
202 of the UMRA, Section 205, 2 U.S.C. 1535, of the UMRA generally requires EPA to identify
and consider a reasonable number of regulatory alternatives and adopt the least costly, most
cost-effective, or least burdensome alternative that achieves the objectives of the rule.
7-36
-------
In the NPR, EPA concluded that the proposed Hg MACT contained a Federal Mandate
that may result in expenditures of $100 million or more for State, local, and Tribal governments
in aggregate, or the private sector in any one year. For that reason, EPA prepared a written
statement for the NPR consistent with the requirements of Section 202 of the UMRA. In today's
final rule, EPA is not directly establishing any regulatory requirements that may significantly or
uniquely affect small governments, including Tribal governments. Thus, under CAMR, EPA is
not obligated to develop under Section 203 of the UMRA a small government agency plan.
Furthermore, in a manner consistent with the intergovernmental consultation provisions of
Section 204 of the UMRA, EPA carried out consultations with the governmental entities affected
by this rule.
EPA analyzed the economic impacts of the final CAMR. This analysis does not examine
potential indirect economic impacts associated with CAIR, such as employment effects in
industries providing fuel and pollution control equipment, or the potential effects of electricity
price increases on industries and households.
This analysis presents the annualize cost of CAMR for the year 2020, which is two years
into the second phase of the Hg cap-and-trade program, and for which the Hg emission cap is 15
tons. An important caveat to note in considering the results presented in this section is (as
discussed earlier in this chapter) that EPA assumes no development in control technologies over
the course of the Hg cap-and-trade program. In reality, Hg emissions control is a fast moving
area with new developments nearly monthly. Actual costs may be lower than those presented
since modeling assumes no improvements in the cost of mercury control technology, while in
reality, control costs are expected to improve over time. As a result, this the projected costs of
the Hg cap-and-trade program for 2020 presented in this analysis most certainly overstate the
impact of the rule on government-owned entities during the second phase of the program. At the
same time, however, the marginal cost projected for mercury control in 2020 may also overstate
the cost-savings that entities selling allowances may experience under the rule. Finally, it should
be noted that during the first phase of the program, the fact that the cap is equal to co-benefits
under CAIR should limit the impact of CAMR on government entities.
7.17.1 Identification of Government-Owned Entities
Using eGRID data, EPA identified state- and municipality-owned utilities and
subdivisions. EPA then used IPM-parsed output to associate these plants with individual
generating units. Entities that did not own at least one unit with a generating capacity of greater
than 25 MW were omitted from the analysis because of their exemption from the rule. This
exempts 37 entities owned by state or local governments. Thus, EPA identified 88 state and
municipality-owned utilities that are potentially affected by CAIR, out of a possible 125, which
are summarized in Table 7-38.
Table 7-38. Summary of Potential Impacts on Government Entities under CAIR
7-37
-------
ECU Ownership Type
Subdivision
State
Municipal
Total
Potentially
Affected
Entities
8
10
70
88
Net Compliance
Cost in 2020
Incremental to
CAIR ($1999
millions
6.5
9.2
32.2
47.9
Number of
Government Entities
with Compliance
Costs >1% of
Generation Revenues
5
4
35
44
Number of
Government
Entities with
Compliance Costs
>3% of Generation
Revenues
2
0
12
14
Note: The total number of potentially affected entities in this table excludes the 37 entities that have been dropped
because they will not be affected by CAMR. Also, the total number of entities with costs greater than
1 percent or 3 percent of revenues includes only entities experiencing positive costs.
Source: IPM and TRUM analysis
7.17.2 Overview of Analysis and Results
After identifying potentially affected government entities, EPA estimated the impact of
CAMR + CAIR, relative to CAIR alone, in 2020 based on the following:
• total impacts of compliance on government entities and
• ratio of small entity impacts to revenues from electricity generation.
The financial burden to owners of EGUs under CAMR is composed of compliance and
administrative costs. This section outlines the compliance and administrative costs for the 88
potentially affected government-owned units identified in EPA modeling.
7.17.2.1
Methodology for Estimating Impacts of CAMR on Government Entities
The primary burden on state and municipal governments that operate utilities under
CAMR is the cost of installing control technology on units to meet their Hg emission budget or
the cost of purchasing allowances. An entity can comply with CAMR through some
combination of the following: installing retrofit technologies, purchasing allowances, switching
to a lower Hg fuel, or reducing emissions through a reduction in generation. Additionally, units
with more allowances than needed can sell these allowances on the market. The chosen
compliance strategy will be primarily a function of the unit's marginal control costs and its
position relative to the marginal control costs of other units. Because CAMR will be
implemented in conjunction with CAIR, units affected by both rules will attempt to minimize
their cost of compliance over both rules, by considering Hg, SO2, and NOX control strategies
simultaneously.
To attempt to account for each potential control strategy over the combined rules, EPA
estimates compliance costs as follows:
Compliance
CFuel + A CAl!owances + A CTr
— A R
(8.2)
7-38
-------
where C represents a component of cost as labeled, and A R represents the retail value of
foregone electricity generation.
In reality, compliance choices and market conditions can combine such that an entity
may actually experience a savings in any of the individual components of cost. Under CAIR and
CAMR, for example, EPA projects that the price of low-sulfur coal will fall as many units install
scrubbers and switch away from low-sulfur coal to cheaper bituminous coal, such that many
entities actually experience a reduction in fuel costs relative to the base caes as a result of lower
prices due to the demand shift. Similarly, although some units will forgo some level of
electricity generation (and thus revenues) to comply, this impact will be lessened on these
entities by the projected increase in electricity prices under CAIR and CAMR as well as
reductions in fuel costs, and those not reducing generation levels will see an increase in
electricity revenues. Elsewhere, unscrubbed units burning low-sulfur coal might find it most
economical to install mercury-specific controls such as ACI, and sell their surplus of Hg
allowances on the market. Because this analysis evaluates the total costs along each of the four
compliance strategies laid out above for each entity, it inevitably captures savings or gains such
as those described. As a result, what we describe as cost is really more of a measure of the net
economic impact of the rule on small entities.
In this analysis, EPA used IPM-parsed output for the base case, CAIR, and CAMR to
estimate net compliance cost at the unit level. These costs were then summed for each
government entity, adjusting for ownership share. Compliance cost estimates were based on the
following: operating and retrofit costs, sale or purchase of allowances, and the change in fuel
costs or electricity generation revenues under CAMR relative to CAIR. These components of
compliance cost were estimated as follows:
(1) Operating and retrofit costs: Using the IPM-parsed output for the base case,
CAIR, and CAMR (available in the docket), EPA identified units that install
control technology under CAIR and CAMR and the technology installed. The
equations for calculating retrofit costs were adopted from EPA's Technology
Retrofit and Updating Model (TRUM). The model calculates the capital cost (in
$/MW); the fixed operation and maintenance (O&M) cost (in $/MW-year); the
variable O&M cost (in $/MWh); and the total annualized retrofit cost for units
projected to install FGD, SCR, SNCR, or ACI.
(2) Sale or purchase of allowances: EPA estimated the value of initial SO2, NOX,
and Hg allowance holdings. For SO2, units were assumed to retain their Phase II
allowance allocations as determined under EPA's 1998 reallocation of Acid Rain
allowances, adjusted to reflect the 50 percent reduction in 2010 and 65 percent
reduction in 2015 under CAIR. Because of the resources involved in compiling
allowance-holding data, the value of banked SO2 allowances was not considered
in this analysis. The implication of this is that the annual net purchase of
allowances may be overstated for some units. For NOX, the state emission
budgets were assumed to be apportioned to units on a heat-input basis. Each unit
was assumed to receive a share of the state NOX emission budget equal to its share
of the total state heat input for that year in the base case. This is a simplification
of what is included in the model rule, which proposes allocating NOX allowances
7-39
-------
based on heat input from 1999-2002.15 However, states can ultimately decide
how to allocate NOX allowances. For Hg, unit allocations were the same as those
listed in the March 16, 2004 Supplemental Notice of Proposed Rulemaking.
To estimate the value of allowances holdings, allocated NOx and SO2 allowances
were subtracted from projected emissions, and the difference was then multiplied
by the allowance prices projected by IPM for 2020. Units were assumed to
purchase or sell allowances to exactly cover their projected NOX and SO2
emissions under CAIR + CAMR. For Hg, units that did not have allowances
sufficient to cover projected 2020 emissions were projected to withdraw
allowances from their respective Hg allowance banks if available, or else
purchase the required amount of allowances. Units holding 2020 allowances in
excess of projected 2020 emissions were projected to sell these excess
allowances. The estimation of the size of a unit's mercury allowance bank is
discussed further below.
(3) Fuel costs: Fuel costs were estimated by multiplying fuel input (MMBtu) by
region and fuel-type-adjusted fuel prices ($/MMBtu) from TRUM. The change in
fuel expenditures under CAMR was then estimated by taking the difference in
fuel costs between CAMR and CAIR.
(4) Value of electricity generated: EPA estimated electricity generation by first
estimating unit capacity factor and maximum fuel capacity. Unit capacity factor
is estimated by dividing fuel input (MMBtu) by maximum fuel capacity
(MMBtu). The maximum fuel capacity was estimated by multiplying capacity
(MW) * 8,760 operating hours * heat rate (MMBtu/MWh). The value of
electricity generated is then estimated by multiplying capacity (MW)*capacity
factor*8,760*regional-adjusted retail electricity price ($/MWh).
As discussed later in this analysis, most government entities projected to be
affected by CAMR do not have to operate in a competitive market environment
and thus should be able to pass compliance costs on to consumers. To somewhat
account for this, we incorporated the projected regional-adjusted retail electricity
price calculated under CAMR in our estimation of generation revenue under
CAMR.
(5) Administrative costs: Because most affected units are already monitored as a
result of other regulatory requirements, EPA considered the primary
administrative cost to be transaction costs related to purchasing or selling
allowances. EPA assumed that transaction costs were equal to 1.5 percent of the
total absolute value of a unit's allowances. This assumption is based on market
research by ICF Consulting.
15 A similar approach was used in regulatory impact analyses for the 126 F1P and NOX SIP Call.
7-40
-------
(6) Value of the Mercury Bank: EPA's economic analysis of CAMR suggests that a
significant bank of approximately 70 tons of Hg allowances will be built up
during the first phase of the cap-and-trade program. Sources will be relying
heavily on this bank for compliance during the second phase of the program.
While not all sources will have banked allowances during the first phase of the
program, many sources will be able to draw from this bank during the second
phase and avoid or limit Hg allowance purchases. EPA estimated the size of the
bank by comparing projected emissions for the years 2010-2019 with allocations
for those years. This estimate assumed that state and local government-owned
sources with surplus allowances in those years would bank those allowances
rather than sell them on the market, and would draw from this bank in any year
that they were short allowances. EPA estimated the cost of using banked
allowances by taking the average cost of Hg control in the first phase of the
program discounted to 2020, multiplied by the number of banked allowances
used. Finally, any surplus allowances remaining in the government entity banks
in 2020 were valued at the 2020 Hg allowance price.
7.77.2.2 Results
A summary of economic impacts on government-owned entities is presented in
Table 7-38. According to EPA's analysis, the total net economic impact on government-owned
entities (state- and municipality-owned utilities and subdivisions) is expected to be
approximately $48 million in 2020. This cost is driven largely by mercury allowance purchases
and additional retrofits relative to CAIR. The costs to government entities in 2020 are limited,
however, by the projection that 33 of the 88 entities sell surplus and/or banked allowances in
2020. In the absence of banked allowances, costs to these entities in 2020 would be greater.
EPA does not project that any coal-fired generation would be uneconomic to maintain
relative to CAIR. This finding suggests that the extent of CAMR's adverse economic impacts
beyond CAIR on small entities is limited.
As was done for the small entities analysis, EPA further assessed the economic and
financial impacts of the rule using the ratio of compliance costs to the value of revenues from
electricity generation in the base case, also focusing specifically on entities for which this
measure is greater than 1 percent. EPA projects that 44 government entities will have
compliance costs greater than 1 percent of revenues from electricity generation in 2020, and 12
entities are projected to have compliance costs greater than 3 percent of revenues. Entities that
are projected to experience negative compliance costs under CAMR are not included in those
totals. This approach is more indicative of a significant impact when an analysis is looking at
entities operating in a competitive market environment. Government-owned entities do not
operate in a competitive market environment and therefore will be able to recover expenses
under CAIR and CAMR through rate increases. Given this, EPA considers the 1 percent
measure in this case a crude measure of the extent to which rate increases will be made at
publicly owned companies.
7-41
-------
The distribution across entities of economic impacts as a share of base case revenue is
summarized in Table 7-39. For state-owned entities and subdivisions, the maximum economic
impact as a share of base case revenues is approximately 3 percent. A few municipality-owned
entities experience economic impacts that are significantly higher than the capacity-weighted
average for this group. In the cases where entities are projected to experience positive net costs
that are a high percentage of revenues, these entities do not find it economic to retrofit and are
unable to switch to a lower-sulfur coal. Thus, these entities comply primarily through the
purchase of allowances and reductions in generation. Overall, the capacity-weighted average
impact of the rule as a share of revenues is well under 1%.
Table 7-39. Distribution of Economic Impacts on Government Entities under CAMR
ECU Ownership Type
Sub-division
State
Municipal
All
Capacity-
Weighted Average
Economic Impacts
as a % of
Generation
Revenues
1.50%
0.30 %
0.38 %
0.40 %
Min
-0.52 %
-0.96 %
-16.55 %
-16.55 %
Max
3.31 %
2.88 %
6.39 %
6.39 %
Source: IPM and TRUM analysis
Additionally, a few municipal entities are projected to experience negative net costs that
are a high percentage of base case generation revenues. These entities have units that are able to
switch to a cheaper, lower-sulfur coal to comply with CAIR and are able to maintain or increase
generation levels, thus increasing revenues. Further, entities in regions for which we project
large electricity price increases relative to other regions tend to be among those at the lower end
of the distribution.
The various components of annualized incremental cost under CAIR to each group of
government entities are summarized in Table 7-40. Under CAMR, the most significant
components of control costs for these entities are allowance purchases, driven largely by the
marginal cost projected for Hg in 2020, as well as additional retrofits. Also, the increased
demand for bituminous coal and the resulting higher bituminous coal price relative to CAIR
leads to an increase in fuel costs for all groups. Retrofit and operating costs for all groups
increase relative to CAIR alone, because of the installation of ACI, as well as some additional
FGD and SCR. Finally, both states and municipals are projected to experience an increase in
electricity generation revenues relative to CAIR alone, while subdivisions are projected to
experience a slight additional drop in revenues relative to CAIR alone. Increased generation
revenues are largely a result of slight increases in the retail price of electricity in most regions
under CAMR, although some facilities are projected to increase generation. Subdivisions
experience a loss in generation revenues because of a net decrease in electricity generation
relative to CAIR that is not offset by the increase in electricity prices.
7-42
-------
Table 7-40. Incremental Annualized Costs under CAMR Relative to CAIR Summarized
by Ownership Group and Cost Category ($1,000,000)
ECU Ownership Type
Retrofit +
Operating
Cost
Net Purchase of
Allowances
Fuel Cost
Lost
Electricity
Revenue
Administrative
Cost
Subdivision 5.3 1.0 0.4 0.2 0.1
State 8.5 7.2 2.0 -8.6 0.2
Municipal 7.6 17.5 9.0 -2.1 0.3
Note: Numbers may not add to totals in Table 7-38 due to rounding.
Source: IPM and TRUM analysis.
7.77.3 Summary of Government Entity Impacts
EPA examined the potential economic impacts on state and municipality-owned entities
associated with this rulemaking based on assumptions of how the affected states will implement
control measures to meet their emissions. These impacts have been calculated to provide
additional understanding of the nature of potential impacts and additional information to the
states as they create State plans to meet the Hg emission budgets set by this rulemaking.
According to EPA's analysis, the total net economic impact on government-owned
entities is expected to be approximately $48 million in 2020. These costs are driven largely by
the purchase of Hg allowances and the cost of additional retrofits under the combination of
CAIR and CAMR. EPA projects that no additional government entity capacity will be
uneconomic to maintain under CAMR relative to what was projected to be uneconomic to
maintain under CAIR. This suggests that the incremental impact of CAMR on small entities
relative to CAIR alone is limited.
Of the 88 government entities considered in this analysis and the 125 government entities
that are included in EPA's modeling, 44 are projected to experience compliance costs in excess
of 1 percent of electricity generation revenues in 2020, and 14 of these are projected to
experience compliance costs in excess of 3% of generation revenues, may in 2015, based on our
assumptions of how the affected states implement control measures to meet their emissions
budgets as set forth in this rulemaking. As is discussed earlier in this analysis, the finding of a
significant impact to some entities during the second phase of the program is largely a product of
the marginal cost projected for Hg control in 2020. In reality, control costs of Hg are expected
to be lower by 2020, such that allowance prices would be reduced, and significant impacts
unlikely. Further, government entities operate in cost-of-service markets where they should be
able to pass on their costs of compliance to rate-payers. The above points aside, potential
adverse impacts of CAMR on state- and municipality-owned entities could be limited by the fact
that the cap-and-trade program is designed such that states determine how Hg allowances are to
be allocated across units. A state that wishes to mitigate the impact of the rule on state- or
municipality-owned entities might choose to allocate Hg allowances in a manner that is
favorable to these entities. Finally, in general, the use of cap-and-trade programs in general will
limit impacts on entities owned by small governments relative to a less flexible
command-and-control program.
7-43
-------
EPA has determined that this rule may result in expenditures of more than $100 million
to the private sector in any single year. EPA believes that the final rule represents the least
costly, most cost-effective approach to achieve the air quality goals of this rule. The costs and
benefits associated with the final rule are discussed throughout this RIA.
7-44
-------
7.18 List of IPM Runs in Support of CAMR
A list of the IPM runs that were used in the various analyses done in support of the final
CAMR is provided. Model output from each of the IPM runs listed in this memo is available in
the CAMR docket and also on EPA's Web site at www.epa.gov/airmarkets/epa-ipm.
Table 7-41. Listing of Runs from the Integrated Planning Model Used in Analyses Done in
Support of the CAMR Final Rule Analyses
Run Name
Base Case 2004
CAIR 2004_Analysis
CAIR 2004_Final
CAMR_Option 1
CAMR_Option 2
CAMR_Option 3
CAMR_Sorbent Sensitivity_Option 1
Base Case 2004_EIA
CAIR 2004_EIA
CAMR 2004_EIA
Run Description
Base case model run, which includes the national Title IV SO2
cap-and-trade program; NOX SIP Call regional ozone season
cap-and-trade program; and state-specific programs in
Connecticut, Illinois, Maine, Massachusetts, Minnesota,
Missouri, New Hampshire, New York, North Carolina,
Oregon, Texas, and Wisconsin. This run represents conditions
without the proposed CAIR.
CAIR control strategy used for much of the analytical work for the
final CAIR (includes AR/DE/NJ for annual controls and no ozone
season cap and is the IPM run used for air quality modeling)
Final CAIR policy (includes annual and ozone season caps for the
States who contribute to PM2.5 and/or ozone nonattainment), used in
Hg cost modeling
Final CAMR control strategy
CAMR option with Hg caps of 38 tons in 2010 and 15 tons in 2015
CAMR option with Hg caps of 38 tons in 2010, 24 tons in 2015, and
15 tons in 2018
CAMR run with second ACI control option in 2013 using advanced
sorbents
Base Case run with EIA assumptions for the difference between
natural gas prices and coal prices, as well as EIA's projection of
electricity growth
CAIR run with EIA assumptions for the difference between natural
gas prices and coal prices, as well as EIA's projection of electricity
growth
CAMR run with EIA assumptions for the difference between natural
gas prices and coal prices, as well as EIA's projection of electricity
growth
EPA base case parsed for year 2010
EPA base case parsed for year 2015
EPA base case parsed for year 2020
EPA CAIR parsed for year 2020
EPA CAMR_Option 1 parsed for year 2020
EPA CAMR_Option 2 parsed for year 2020
EPA C AMR_Option 3 parsed for year 2020
7-45
-------
SECTION 8 AIR QUALITY MODELING: CHANGES IN HG DEPOSITION TO U.S.
WATERBODIES 8J.
8.1 Emissions Inventories and Estimated Emissions Reductions 8-2
8.2 Model, Domain, Configuration, Inputs, Application 8-5
8.2.1 Air Quality Model 8-5
8.2.2 Modeling Domain 8-6
8.2.3 Time Periods Modeled for Mercury Deposition 8-6
8.2.4 Model Inputs 8-7
8.3 CMAQ Model Performance Evaluation 8-8
8.4 Mercury Deposition Results 8-9
8.5 Summary of Findings: HUC Level Deposition Analysis 8-14
8.6 References 8-17
Tables
Table 8-1. Summary of Emissions Sources for 2001 and 2020 Mercury Emissions
Inventories 8-3
Table 8-2. Summary of Mercury Emissions by Species: 2001 and 2020 (with CAIR)
Baselines 8-3
Table 8-3. Summary of Changes in Mercury Emissions Associated with CAMR Control
Option 1: 2020 8-4
Table 8-4. Summary of Changes in Mercury Emissions Associated with CAMR Control
Option 2: 2020 8-4
Table 8-5. CMAQ Performance Statistics for Mercury Wet Deposition: 2001 8-9
Table 8-6. Summary Statistics of Total Mercury Depositions (ug/m2) by Modeling
Scenario 8-14
Table 8-7. Summary Statistics of Utility Attributable Deposition (ug/m2) by Modeling
Scenario 8-16
Figures
Figure 8-1. CMAQ Modeling Domain 8-6
Figure 8-2. Base Case Total Mercury Deposition: 2001 8-10
Figure 8-3. Decrease in Total Mercury Deposition with Power Plant Zero-Out
Simulation: 2001 8-11
Figure 8-4. Change in Total Mercury Deposition for All Sources: 2020 (with CAIR)
Relative to 2001 8-12
Figure 8-5. Total Mercury Deposition: 2020 (with CAIR) 8-12
Figure 8-6. Change in Mercury Depositions from Power Plants Due to CAMR
Option 1: 2020 8-13
Figure 8-7. Change in Mercury Deposition from Power Plants Due to CAMR
Option 2: 2020 8-13
Figure 8-8. Cumulative Distribution of Total Mercury Deposition (ug/m2) Fat HUC-8 Level by
Modeling Scenario 8-15
Figure 8-9. Cumulative Distribution of Utility Attributable Mercury Deposition at HUC-8 Level
by Model Scenario 8-16
Figure 8-10. Cumulative Distribution of Percent Deposition (ug/m2) Attributable to Utilities at
HUC-8 Level by Modeling Scenario 8-17
-------
SECTION 8
AIR QUALITY MODELING: CHANGES IN HG DEPOSITION TO U.S.
WATERBODIES
This section summarizes the emissions inventories and air quality modeling that serve as
the inputs to the benefits analysis for the Clean Air Mercury Rule (CAMR). EPA used a
sophisticated photochemical air quality model to predict the levels of mercury deposition for a
2001 base year and a 2020 baseline reflecting co-control of mercury from implementation of the
Clean Air Interstate Rule (CAIR) as well as two control options for CAMR. The estimated
changes in mercury deposition associated with the control options were then combined with fish
tissue data for use in estimating health and welfare effects. In addition, utility attributable
deposition of mercury was estimated based on zero-out modeling for both the 2001 and 2020
baselines.
The 1997 Mercury Study Report to Congress noted that "a single air quality model which
was capable of model both the local as well as regional fate of mercury was not identified." In
fact, at that time such a model did not exist. Thus, the modeling approach for this report
employed two models: 1) the Regional Lagrangian Model of Air Pollution (RELMAP) to
address regional-scale atmospheric transport, and 2) the Industrial Source Code model (ISC3) to
address local-scale analyses (i.e., within 50 km of source). This approach also required
assumptions to be made about the background concentrations of mercury that were uniformly
added to the regional component and the use of "model plants" to represent typical sources for
the local-scale transport. At this time, the Agency would have significant concerns about using
the ISC3 model for assessments of Hg deposition associated with CAMR. The Agency will later
this year promulgate the American Meteorological Society/Environmental Protection Agency
Regulatory MODel (AERMOD) that will replace ISC3 as the recommended and preferred model
for use in regulatory permit modeling assessments. This model contains the Argonne National
Laboratory (ANL) versions of the wet and dry deposition algorithm which contain refinements
beyond the ISC3 model and are considered more robust through extensive testing and evaluation.
The ISC3 outputs for wet and dry deposition were never fully tested and verified for use in
regulatory applications.
The Agency views the application of a more robust and sophisticated modeling approach
as critical and required for assessing the Hg deposition associated with CAMR because of the
density and properties of Hg and its complex transport and reactions in the atmosphere. The
Community Multiscale Air Quality (CMAQ) modeling system best meets our requirements and
the recommendations of the Report to Congress for a 'single air quality model" to address Hg
deposition. CMAQ is a three-dimensional grid-based Eulerian air quality model designed to
estimate pollutant concentrations and depositions over large spatial scales (e.g., over the
contiguous United States). Because it accounts for spatial and temporal variations as well as
differences in the reactivity of Hg emissions, CMAQ is the best available model for evaluating
the impacts of the CAMR on U.S. mercury depositions. This model appropriately accounts for
the atmospheric reactions of specific Hg emissions and their significance in the levels of
deposition as shown through our results here for CAMR. In addition, the boundary and initial
species concentrations are provided by a three-dimensional global atmospheric chemistry and
8-1
-------
transport model, i.e., Harvard's GEOS-CHEM model. The model simulations are performed
based on plant-specific emissions of Hg by species as provided by the Integrated Planning Model
(IPM).
Section 8.1 provides a summary of the emissions inventories that were modeled for this
rule. Section 8.2 summarizes the model, domain, configuration, inputs, and application. Section
8.3 summarizes the model performance. Section 8.4 summarizes the results of estimating
mercury depositions for the 2001 and 2020 scenarios modeled. Section 8.5 summarizes the
findings at water bodies for the scenarios modeled and Section 8.6 provides the key references
for this analysis.
8.1 Emissions Inventories and Estimated Emissions Reductions
The CAMR Emissions Inventory Technical Support Document (TSD) discusses the
development of the 2001 and 2020 emissions inventories for input to the air quality modeling of
this final rule in greater detail. Table 8-1 provides the emission sources and the basis for current
and future-year inventories, while Table 8-2 summarizes the mercury emissions by species from
utilities, also known as Electric Generating Units (EGUs), and other sources that were used in
modeling of mercury deposition.
As Table 8-2 demonstrates, a total of almost 115 tons of mercury were emitted across all
sources in 2001. EGUs emitted a total of 48.6 tons, or 42.3 percent of mercury emissions across
all sources during this base year. Almost 21 tons of the most readily deposited form of mercury,
i.e., reactive gaseous mercury (RGM), were emitted by these utilities and therefore comprised
42.4 percent of their mercury emissions.
The 2020 baseline emissions shown in Table 8-2 accounts for increases in economic
activity and population growth between 2001 and 2020 that lead to increased production in the
utility and manufacturing sectors and hence increases in emissions over time, as well as the
implementation of regulatory policies from MACT standards (primarily on non-EGU sources)
and the CAIR controls (as applied to EGUs in the eastern U.S.) which decreases emissions over
this time period. Total mercury emissions in 2020 are roughly 87 tons, reflecting a net reduction
of almost 28 tons (or 24 percent) from 2001 levels. As shown, the 2020 baseline with CAIR
shows net reductions in mercury emissions for EGUs of 14.2 tons or a 29.1 percent reduction
from 2001 levels. Utility emissions are expected to account for 39.5 percent of total mercury
emissions in 2020, which is only slightly lower than their share in 2001. However, the
reductions associated with CAIR co-control show a large reduction of 61.8 percent in their
emissions of reactive gaseous mercury relative to their 2001 level of emissions, i.e., 20.58 tons in
2001 to only 7.87 tons in 2020.
8-2
-------
Table 8-1. Summary of Emissions Sources for 2001 and 2020 Mercury Emissions
Inventories
Sector
Emissions Source 2001 Base Year
2020 Base Case Projections
Utilities -
Electric
Generating
Units (ECU)
Non-EGU
point sources
Power industry
electric generating
units (EGUs)
Non-Utility Point
Non-point
sources
All other stationary
sources inventoried
at the county level
1999 National Integrated Planning Model (IPM) reflecting growth in
Emission Inventory Btu demand as well as regulatory policies
(NEI) data implemented through 2020, such as the Clean Air
Interstate Rule
1999 NEI, with (1) Department of Energy (DOE) fuel use projections,
medical waste (2) Regional Economic Models, Inc. (REMI) Policy
incinerator sources Insight® model, (3) decreases to REMI results based
replaced with draft on trade associations, Bureau of Labor Statistics
2002 NEI (BLS) projections and Bureau of Economic Analysis
(BEA) historical growth from 1987 to 2002, (4)
Maximum Achievable Control Technology category
growth and control assumptions
1999 NEI, with same as above
medical waste
incinerator sources
replaced with draft
2002 NEI
"This table documents only the sources of data for the U.S. inventory. The sources of data used for Canada and Mexico are explained in the
technical support memorandum and were held constant from the base year to the future years.
Table 8-2. Summary of Mercury Emissions by Species: 2001 and 2020 (with CAIR)
Baselines
Emissions Source
Mercury Emissions Species (tons)
Elemental
Reactive Gaseous
Particulate
Total Mercury
Emissions (tons)
2001 Base Year
EGUs
Non-EGU Point
Non-point
Total, All Sources
26.26
37.85
5.05
69.16
20.58
13.33
1.53
35.44
1.73
7.60
0.96
10.29
48.57
58.78
7.54
114.89
2020 (with CAIR) Baseline
EGUs
Non-EGU Point
Non-point
Total, All Sources
25.72
28.03
5.69
59.44
7.87
10.37
1.30
19.54
0.83
6.61
0.77
8.21
34.42
45.01
7.76
87.19
Table 8-3 shows the reductions in mercury emissions associated with the CAMR Control
Option 1 in 2020. The 2020 EGU emissions are reduced by approximately 10 tons to a total of
25 tons, representing all percent reduction from total baseline emissions in 2020 (with CAIR),
or a 27 percent reduction from the EGU sector alone. Under CAMR Control Option 2
8-3
-------
(Table 8-4), EGU emissions are further reduced by an additional 4 tons to a total of roughly 21
tons. This represents a 16 percent reduction from total emissions from the 2020 baseline (with
CAIR), or a 39 percent reduction from the EGU sector alone.
Table 8-3. Summary of Changes in Mercury Emissions Associated with CAMR Control
Option 1: 2020
Emissions Source
EGUs
Non-EGU Point
Non-point
Total, All Sources
Change in Mercury Emissions Species (tons)
Elemental
8.07
(31.4%)
n/a
n/a
8.07
(13.6%)
Reactive Gaseous
1.30
(16.5%)
n/a
n/a
1.30
(6.7%)
Participate
0.00
(0.0%)
n/a
n/a
0.00
(0.0%)
Total Change in
Mercury
Emissions (tons)
9.37
(27.2%)
n/a
n/a
9.37
(10.7%)
Note: n/a is not applicable.
Table 8-4. Summary of Changes in Mercury Emissions Associated with CAMR Control
Option 2: 2020
Emissions Source
EGUs
Non-EGU Point
Non-point
Total, All Sources
Change in Mercury Emissions Species (tons)
Elemental
11.39
(44.3%)
n/a
n/a
11.39
(19.2%)
Reactive Gaseous
2.16
(27.4%)
n/a
n/a
2.16
(11.1%)
Particulate
0.04
(4.8%)
n/a
n/a
0.04
(0.5%)
Total Change in
Mercury
Emissions (tons)
13.59
(39.5%)
n/a
n/a
13.59
(15.6%)
Note: n/a is not applicable.
In comparison to current mercury emissions (i.e., the 2001 base year scenario), the CAIR
and CAMR Option 1 achieve a total reduction in EGU emissions of approximately 24 tons (48
percent), while CAIR and CAMR Option 2 achieve a total reduction in EGU emissions of
approximately 28 tons (57 percent).
8-4
-------
8.2 Model, Domain, Configuration, Inputs, Application
This section summarizes the methods for and results of estimating mercury depositions
for 2001 and 2020 base cases and control scenarios for the purposes of the benefits analysis. The
mercury deposition changes were estimated using national-scale applications of the Community
Multi-Scale Air Quality (CMAQ) model. In Section 8.2.1, we describe the estimation of mercury
depositions using CMAQ.
8.2.1 Air Quality Model
We use the emissions inputs summarized above with a national-scale application of the
Community Multi-scale Air Quality (CMAQ) modeling system to estimate mercury depositions
in the contiguous United States. CMAQ is a three-dimensional grid-based Eulerian air quality
model designed to estimate pollutant concentrations and depositions over large spatial scales
(e.g., over the contiguous United States). Because it accounts for spatial and temporal variations
as well as differences in the reactivity of emissions, CMAQ is useful for evaluating the impacts
of the CAMR on U.S. mercury depositions. Our analysis applies the modeling system to the
entire United States for six emissions scenarios: a 2001 base year, a 2001 base year with utility
mercury emissions zeroed-out, a 2020 projection with CAIR incorporated, a 2020 projection
with CAIR incorporated and utility mercury emissions zeroed-out, a 2020 projection with CAIR
and control option 1 incorporated, a 2020 projection with CAIR and control option 2
incorporated.
The CMAQ version 4.3 was employed for this CAMR modeling analysis (Byun and
Schere, 2004, Bullock and Brehme 2002). This version reflects updates in a number of areas to
improve performance and address comments from its peer review. The updates in mercury
chemistry used for CAMR from that described in (Bullock and Brehme 2002) are as follows:
(1) the elemental mercury (HgO) reaction with H2O2 assumes the formation of 100 percent
reactive gaseous mercury (RGM) rather than 100 percent particulate mercury (HgP), (2) the HgO
reaction with ozone assumes the formation of 50 percent RGM and 50 percent HgP rather than
100 percent HgP, (3) the HgO reaction with OH assumes the formation of 50 percent RGM and
50 percent HgP rather than 100 percent HgP, and (4) the rate constant for the HgO + OH reaction
was lowered from 8.7 to 7.7 xlO"14cm3molecules"1s"1. CMAQ simulates every hour of every day
of the year and, thus, requires a variety of input files that contain information pertaining to the
modeling domain and simulation period. These include hourly emissions estimates and
meteorological data in every grid cell, as well as a set of pollutant concentrations to initialize the
model and to specify concentrations along the modeling domain boundaries. These initial and
boundary concentrations were obtained from output of a global chemistry model. We use the
model predictions in a relative sense by first determining the ratio of mercury deposition
predictions. The calculated relative change is then combined with the corresponding fish tissue
concentration data to project fish tissue concentrations for the future case scenarios. The
following sections provide a more detailed discussion of the modeling and a summary of the
results.
8-5
-------
8.2.2 Modeling Domain
As shown in Figure 8-1, the modeling domain encompasses the lower 48 states and
extends from 126 degrees west longitude to 66 degrees west longitude and from 24 degrees north
latitude to 52 degrees north latitude. The modeling domain is segmented into rectangular blocks
referred to as grid cells. The model actually predicts pollutant concentrations for each of these
grid cells. For this application the horizontal domain consisted of 16,576 grid cells that are
roughly 36 km by 36 km. In addition, the modeling domain contains 14 vertical layers with the
top of the modeling domain at about 16,200 meters, or 100 millibar. The height of the surface
layer is 38 meters.
Figure 8-1. CMAQ Modeling Domain
8.2.3 Time Periods Modeled for Mercury Deposition
CMAQ was run for a full year for each of the six CAMR emissions scenarios modeled.
The overall model run time for completing an annual simulation was reduced by dividing the
year into two six-month periods which were run in parallel on different computer processors.
That is, the annual simulation was performed as two separate six month model runs. One run
was for January through June and the other run was for July through December. Each six-month
run included a 10-day ramp-up (i.e., "spin-up") period designed to minimize the influence of the
initial concentration fields (i.e., initial conditions) used at the start of the model run. The
development of initial condition concentrations is described in Section 8.2.4 below. The ramp-
up periods used for the CAMR CMAQ applications are as follows:
8-6
-------
• First six-month ramp-up period is December 22-31, 2000
• Second six-month ramp-up period is June 21 - 30, 2001
Model predictions from these ramp-up periods were discarded and not used in analyses of the
modeling results. The meteorological conditions, initial conditions and boundary conditions were
held constant for each of the emissions scenarios modeled and are described below in section
8.2.4.
8.2.4 Model Inputs
CMAQ requires a variety of input files that contain information pertaining to the
modeling domain and simulation period. These include gridded, hourly emissions estimates and
meteorological data as well as initial and boundary conditions. Separate emissions inventories
were prepared for the 2001 base year and each of the future-year base cases and control
scenarios. All other inputs were specified for the 2001 base year model application and
remained unchanged for each future-year modeling scenario.
CMAQ requires detailed emissions inventories containing temporally allocated emissions
for each grid cell in the modeling domain for each species being simulated. The previously
described annual emission inventories were processed into model-ready inputs through the
emissions processing system. Details of the processing of emissions are provided in the Clean
Air Mercury Rule Emissions Inventory Technical Support Document (EPA, 2005).
Meteorological data, such as temperature, wind, stability parameters, and atmospheric
moisture contents influence the formation, transport, and removal of air pollution. The CMAQ
model requires a specific suite of meteorological input files in order to simulate these physical
and chemical processes. For the CAMR CMAQ modeling, meteorological input files were
derived from a simulation of the Pennsylvania State University / National Center for
Atmospheric Research Mesoscale Model (Grell et al, 1994) for the entire year of 2001. This
model, commonly referred to as MM5, is a limited-area, nonhydrostatic, terrain-following
system that solves for the full set of physical and thermodynamic equations which govern
atmospheric motions. For this analysis, version 3.6.1 of MM5 was used.
National modeling, such as the CAMR annual mercury modeling, requires the
prescription of boundary conditions (BC's) to account for the influx of pollutants and precursors
from the upwind source areas outside the modeling domain. A scientifically sound approach to
estimate incoming pollutant concentration associated with intercontinental transport is to use a
global chemistry model to provide the dynamic BC's for the national model simulation. For the
CAMR mercury modeling, we used the predictions from a three-dimensional global atmospheric
chemistry and transport model, the GEOS-CHEM model (Yamatosca B., 2004) developed at
Harvard University to provide the lateral boundary and initial species concentrations. The lateral
boundary species concentrations varied with height and time (every 3 hours). Terrain elevations
and land use information were obtained from the U.S. Geological Survey database at 10 km
resolution and aggregated to the roughly 36 km horizontal resolution used for this CMAQ
application.
8.3 CMAQ Model Performance Evaluation
8-7
-------
At this point in time, it is difficult to assess model performance for total mercury
deposition. Scientist currently believe through analysis of very limited measurements that wet
and dry deposition are approximately equal in magnitude. There currently is no measurement
network to evaluate the performance of models in estimating dry deposition of mercury. Thus,
we are not able to evaluate the performance of air quality models in predicting dry deposition,
which is thought to be roughly half of total mercury deposition. There is a network of mercury
wet deposition monitors, which are scattered throughout remote locations in the United States
and Canada, mostly in the east. Thus, model predictions of wet deposition can be evaluated by a
monitoring network.
An operational model performance evaluation for mercury wet deposition for 2001 was
performed to estimate the ability of the CMAQ modeling system to replicate base-year wet
depositions of mercury. The wet deposition evaluation principally comprises statistical
assessments of model versus observed pairs that were matched in time and space on a seasonal
and annual basis. The statistics are presented separately for the entire domain, the East, and the
West (using the 100th meridian to divide the eastern and western United States). These statistics
on model performance along with an annual observed versus predicted performance scatter plot
can be found in the Clean Air Mercury Rule Emissions Inventory and Air Quality Modeling
Technical Support Document.
For mercury wet deposition, this evaluation includes comparisons of model predictions to
the corresponding measurements from the Mercury Deposition Network (MDN). The principal
evaluation statistics used to evaluate CMAQ performance are the fractional bias and fractional
error. Fractional bias is defined as:
2 N (Pred* - Obs')
FBIAS = — - - ^ - ^ * 100
where: N = the number of measurement sites
Pred = model predicted deposition at site x over time t (i.e. Annual)
Obs = observed deposition at site x over time t
Fractional bias is a useful model performance indicator because it has the advantage of equally
weighting positive and negative bias estimates. Fractional error is similar to fractional bias
except the absolute value of the difference is used so that the error is always positive. Fractional
error is defined as:
2 N \Pred' - Obs'\
= - J- - ^ - ^
FERROR = - - - ^ - * 100
The fractional bias and fractional error statistics were calculated using the predicted-observed
pairs for the full year of 2001 and for each season, separately. These metrics were calculated
annually and seasonally for all available MDN sites in 2001. Only sites where data was
8-8
-------
available more than half the weeks in a season were utilized for the seasonal performance
evaluation and only sites that had four seasons meeting this data completeness requirement were
utilized for the annual performance evaluation. There were 52 MDN sites in 2001 that meet the
annual data completeness requirements, of those sites 48 were located in the east and 4 were
located in the west. Fractional bias for cases where the model underpredicts by a factor of 2
would be -67 and for cases where the model overpredicts by a factor of 2 would be + 67 percent.
The results in Table 8-5 shows that averaged annually over all MDN monitoring sites, CMAQ
underestimates mercury wet deposition with a fractional bias of approximately -23 percent. This
underprediction bias is well within a factor of 2. The 4 MDN sites in the west do not provide an
adequate or representative basis for inferring model performance.
Table 8-5. CMAQ Performance Statistics for Mercury Wet Deposition: 2001
Area
Entire Domain
East
West
#MDN
Sites
52
48
4
Fractional Bias (%)
-23.2
-27.0
21.7
Fractional Error (%)
30.2
30.2
30.5
8.4 Mercury Deposition Results
Maps showing the mercury deposition results are provided below. The annual total
modeled mercury deposition for the 2001 base case is shown in Figure 8-2. The reduction in
total mercury deposition that would result if all US power plant mercury emissions were zeroed-
out in 2001 is shown in Figure 8-3. The change in 2001 total mercury deposition in 2020 with
CAIR is shown in figure 8-4. The total mercury deposition for 2020 with CAIR is shown in
Figure 8-5. The decrease in 2020 with CAIR when all US power plant emissions are zeroed-out
is shown in Figure 8-6. The change in 2020 CAIR total mercury depositions with CAMR Option
1 is shown in Figure 8-7. The change in 2020 CAIR total mercury depositions with CAMR
Option 2 is shown in Figure 8-8. It can be seen in Figures 8.3 and 8.4 that the implementation of
CAIR and other minor non-utility mercury emissions decreases in 2020 result in a similar
reduction in total mercury deposition as completely eliminating power plant mercury emissions.
The main cause of this result is that CAIR results in a very large decrease in reactive gaseous
mercury (RGM) emissions from Power Plants through the implementation of scrubber control
technology (see Table 8-2 ). RGM is the most readily deposited form of mercury. It can be seen
in Figures 8-7 and 8-8 that the implementation of CAMR Option 1 and CAMR Option 2 results
in some scattered total mercury deposition reductions beyond CAIR in 2020, but for the most
part these reductions are not very significant compared to those obtained by CAIR. Most of the
mercury emissions reductions from CAMR are in the form of elemental mercury (HgO). This
form of mercury is not readily deposited, but enters the global pool of mercury. Thus, CAMR
will result in a reduction of the transport of mercury to other places in the world.
8-9
-------
O.OOO 1
ug/m2
148
January 1,0 O:OO:00
Min- 3.348 at (33,19), Max- 133.229 at (21.84)
Figure 8-2. Base Case Total Mercury Deposition: 2001
8-10
-------
0.000
ug/m2
148
January 1,0 0:00:00
Min- 0.000 at (1,88). Max- 33.589 at (118,64)
Figure 8-3. Decrease in Total Mercury Deposition with Power Plant Zero-Out Simulation:
2001
8-11
-------
16.000112
14.000
12.000
10.000
8.000
6.000
4.000
2.000
0.000 1
ug/m2
148
January 1,00:00:00
Min- -30.130 at (21,84). Max- 43.963 at (98,50)
Figure 8-4. Change in Total Mercury Deposition for All Sources: 2020 (with CAIR)
Relative to 2001
0.000
ug/m2
148
January 1,0 0:00 :OO
Min- 3.261 at (71,95). Max= 163.359 at (21.84)
Figure 8-5. Total Mercury Deposition: 2020 (with CAIR)
8-12
-------
o.ooo
ug/mZ
January 1,O O:OO:OO
Min- -1.24O at (128,60), Max- 3.408 at (124,56)
Figure 8-6. Change in Mercury Depositions from Power Plants Due to CAMR Option 1:
2020
8.000112
7.000
6.000
5.000
4.0OO
3.000
2.OOO
1.000
0.000
ug/m2
148
January 1,0 0:00:00
Min- -1.122 at (128.60). Max- 9.518 at (124,56)
Figure 8-7. Change in Mercury Deposition from Power Plants Due to CAMR Option 2:
2020
8-13
-------
8.5 Summary of Findings: HUC Level Deposition Analysis
The cumulative distribution of Hydrologic Unit Code (HUC) level depositions across
watersheds are provided in Table 8-6 and Figure 8-8. The cumulative percentage of HUCs that
have deposition less than the value on the x-axis for each of the six modeled scenarios are shown
in Figure 8-8. For example, 90 percent of the HUCs have depositions below 22.16 ug/m2 in the
2001 base case. For the 2020 CAIR plus CAMR Option 1 scenario, 90 percent of the HUCs
have depositions below 19.48 ug/m2.
Table 8-6. Summary Statistics of Total Mercury Depositions (ug/m2) by Modeling Scenario
Statistics
Minimum
Maximum
50* percentile
90th percentile
99th percentile
2001
Base
Case
6.994
54.54
15.92
22.16
32.35
2001 Utility Hg
Zero-Out
6.942
54.38
14.60
19.48
27.20
2020 CAIR
6.078
62.76
14.59
19.46
29.15
2020 Utility Hg
Zero-Out
5.898
62.72
13.92
19.04
28.93
2020 CAIR
&CAMR
Option 1
6.075
62.76
14.44
19.37
28.96
2020 CAIR
&CAMR
Option 2
6.075
62.75
14.39
19.33
28.95
8-14
-------
Cumulative Distributions of Total Hg Deposition
100.00%
90.00%
80.00%
70.00%
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61
Deposition (units)
Figure 8-8. Cumulative Distribution of Total Mercury Deposition (ug/m2) Fat HUC-8
Level by Modeling Scenario
The cumulative distribution of Hydrologic Unit Code (HUC) level depositions
attributable to utilities are provided in Table 8-7 and Figure 8-9. The cumulative percentage of
HUCs that have deposition less than the value on the x-axis for 4 of the modeled scenarios are
shown in Figure 8-9. For example, 90 percent of the HUCs have depositions attributable to
utilities below 4.08 ug/m2 in the 2001 base case. For the 2020 CAIR plus CAMR Option 1
scenario, 90 percent of the HUCs have depositions attributable to utilities below 1.16 ug/m2.
CAIR shifts the distribution of utility attributable deposition significantly, resulting in a 75
percent reduction in the 99th percentile of utility attributable deposition, and a 20 percent
reduction in the 50th percentile. CAMR Option 1 and Option 2 results in an additional reduction
in 2020 utility attributable deposition in the 99th percentile of 15 and 20 percent, respectively.
At the 50th percentile, CAMR Option 1 and Option 2 result in an additional reduction of 2020
utility attributable deposition of 16 and 29 percent, respectively. As can be seen in Figure 8-10,
CAIR also shifts the distribution of percentage of HUCs with deposition attributable to utilities.
In the 2001 base case, 10 percent of HUCs had greater than 20 percent of deposition attributable
to utilities. In the 2020 with CAIR scenario, 10 percent of HUCs had greater than 10 percent of
deposition attributable to utilities. In the 2020 CAIR plus CAMR Option 1 scenario, 10 percent
of HUCs had greater than 7 percent of deposition attributable to utilities.
8-15
-------
Table 8-7. Summary Statistics of Utility Attributable Deposition (ug/m2) by Modeling
Scenario
Statistics
Minimum
Maximum
50th percentile
90* percentile
99th percentile
2001 Base Case
0.00
19.71
0.39
4.08
10.15
2020 CAIR
0.00
4.03
0.31
1.38
2.56
2020 CAIR &
CAMR Option 1
0.00
3.85
0.26
1.16
2.17
2020 CAIR &
CAMR Option 2
0.00
3.80
0.22
0.99
2.04
Cumulative Distributions of Utility Attributable Hg Deposition Across HUC-8 Units
100%
95%
90%
85%
g 75%
g 70%
| 65%
0
60%
55%
50%
-2001 Utility Attributable
-2020 Utility Attributable
-CAMR1 Utility Attributable
-CAMR2 Utility Attributable
9 10 11 12 13 14 15 16 17 18 19 20
Deposition
Figure 8-9. Cumulative Distribution of Utility Attributable Mercury Deposition at HUC-8
Level by Model Scenario
8-16
-------
Cumulative Distribution of Percent of Deposition Attributable to Utilities
1% 4% 7% 10% 13% 16% 19% 22% 25% 28% 31% 34% 37% 40% 43% 46% 49% 52% 55%
Percent of Total Deposition
Figure 8-10. Cumulative Distribution of Percent Deposition (ug/m2) Attributable to
Utilities at HUC-8 Level by Modeling Scenario
8.6 References
Bullock, R. and Brehme, K., "Atmospheric Mercury Simulation using the CMAQ Model:
Formulation, Description, and Analysis of Wet Deposition Results", Atmospheric
Environment 36, 2135-2146,2002.
Byun, D., and K.L. Schere. March 2004. "Review of the Governing Equations, Computational
Algorithms, and Other Components of the Models-3 Community Multiscale Air Quality
(CMAQ) Modeling System." Submitted to the Journal of Applied Mechanics Reviews.
Grell, G., J. Dudhia, and D. Stauffer, 1994: A Description of the Fifth-Generation Penn
State/NCAIR Mesoscale Model (MM5), NCAIR/TN-398+STR., 138 pp, National Center
for Atmospheric Research, Boulder CO.
U.S. Environmental Protection Agency (EPA). 2005. Clean Air Mercury Rule Emission
Inventory Technical Support Document. Office of Air Quality Planning and Standards.
Research Triangle Park, NC.
Yantosca, B., 2004. GEOS-CHEMv7-01-02 User's Guide, Atmospheric Chemistry Modeling
Group, Harvard University, Cambridge, MA, October 15, 2004.
8-17
-------
SECTION 9 ANALYSIS OF THE DOSE-RESPONSE RELATIONSHIP BETWEEN
MATERNAL MERCURY BODY BURDEN AND CHILDHOOD IQ
-------
SECTION 9
ANALYSIS OF THE DOSE-RESPONSE RELATIONSHIP BETWEEN MATERNAL
MERCURY BODY BURDEN AND CHILDHOOD IQ
9.1 Introduction
In considering possible health endpoints for quantification and monetization in this
analysis, EPA reviewed the scientific literature on the health effects of mercury, including the
Toxicological Effects of Methylmercury, published by the National Research Council (NRC) in
2000 (NRC 2000).
Epidemiological studies of prenatal mercury exposure conducted in the Faroe Islands
(Grandjean et al. 1997), New Zealand (Kjellstrom et al. 1989, Crump et al. 1998), and the
Seychelles Islands (Davidson et al. 1998, Myers et al. 2003) have examined neurodevelopmental
outcomes through the administration of tests of cognitive functioning. Each of these studies
included some but not all of the following tests: full-scale IQ, performance IQ, problem solving,
social and adaptive behavior, language functions, motor skills, attention, memory and other
functions. The NRC reviewed the studies and determined that "Each of the studies was well
designed and carefully conducted, and each examined prenatal MeHg [methylmercury]
exposures within the range of the general U.S. population exposures" (NRC 2000).
EPA held a neurotoxicology workshop with several of the NRC panel members in
November 2002. Participants were asked about which studies should be considered in
generating dose-response functions for developmental neurotoxicity. Participants were also
asked about endpoints to consider for monetization and they suggested looking at neurological
tests that might lead to changes in IQ or other neurodevelopmental impacts.
EPA has chosen to focus on quantification of intelligence quotient (IQ) decrements
associated with prenatal mercury exposure as the initial endpoint for quantification and valuation
of mercury health benefits. Reasons for this initial focus on IQ include the availability of
thoroughly-reviewed, high-quality epidemiological studies assessing IQ or related cognitive
outcomes suitable for IQ estimation, and the availability of well-established methods and data
for economic valuation of avoided IQ deficits, as applied in EPA's previous benefits analyses for
childhood lead exposure. Bellinger (2005) provides a more detailed discussion of the use of IQ
as the focus of benefits analysis.
The focus of this analysis was to identify the appropriate dose-response coefficients from
the Faroe Islands, New Zealand, and Seychelles studies, and to devise a statistical approach for
combining those coefficients to provide an integrated estimate of the IQ dose-response
coefficient.
EPA is using a linear model that goes through the origin to fit population-level
dose-response relationships to the pooled data from the three studies. The application of a linear
model should not be interpreted to suggest that any of the three studies used have data showing
health effects from methylmercury exposure at or below the RfD. The RfD is an estimate (with
9-1
-------
uncertainty spanning perhaps an order of magnitude) of a daily exposure to the human
population (including sensitive subgroups) that is likely to be without an appreciable risk of
deleterious effects during a lifetime (EPA 2002). EPA believes that exposures at or below the
RfD are unlikely to be associated with appreciable risk of deleterious effects. It is important to
note, however, that the RfD does not define an exposure level corresponding to zero risk;
mercury exposure near or below the RfD could pose a very low level of risk which EPA deems
to be non-appreciable. It is also important to note that the RfD does not define a bright line,
above which individuals are at risk of adverse effect. The regulation is focused on the reduction
of exposures and the associated health benefits that would accrue to people currently exposed at
levels above the RfD due solely to power plants.
Use of a linear model that goes through the origin, rather than one that reflects a
threshold effect is technically more simple and practical. It associates an increment of IQ benefit
with a given reduction in exposure. A linear model allows us to estimate the benefits of
reductions in exposure due to power plants without a complete assessment of other sources of
exposure. Other models would require information on the joint distribution of exposure from
power plants and other sources to estimate the benefits of reducing the exposure due to power
plants, which would require much more precise information about consumption patterns.
9.2 Epidemiological Studies of Mercury and Neurodevelopmental Effects
The IQ dose-response analysis uses data from three major prospective studies
investigating potential neurotoxicity of low-level, chronic mercury exposure: the New Zealand
study, the Seychelles Child Development Study, and the Faroe Islands study.
In assembling the New Zealand sample, Kjellstrom et al. (1989) ascertained the fish
consumption of 10,930 of 16,293 pregnant women in the study area. They identified 935 women
who reportedly consumed fish at least 3 times per week. Hair samples were obtained from these
women, and 73 were found to have a hair mercury level of 6 parts per million (ppm) or greater.
In this group, the mean was 8.3 ppm, with a range of 6 to 86 ppm, although only one woman had
a level greater than 20 ppm. Each woman with 6 ppm hair mercury or greater was matched to 3
controls - one with hair mercury between 3-6 ppm, one with hair mercury less than 3 ppm and
high fish consumption, and one with hair mercury less than 3 ppm and low fish consumption.
Ethnic group, age, smoking, residence time in New Zealand, and child sex were also used to
select controls. The final study group included 237 children, including 57 fully-matched sets of
4 children. Although children were assessed at 4 and 6 years of age, only the data collected at
the older age is considered in this analysis, as the reliability and validity of neurodevelopmental
testing generally increases with child age. Table 9-1 lists the tests administered at the 6 year
evaluation, and indicates the general functional domain each is considered to assess.
9-2
-------
Table 9-1. Neurobehavioral Tests Administered at the 6-Year Evaluations in the New
Zealand Study
Test
Primary Domain Assessed
Wechsler Intelligence Scale for Children-Revised
McCarthy Scales of Children's Abilities
Test of Language Development
Peabody Picture Vocabulary Test
Clay Reading Diagnostic Survey
Burt Word Recognition Test
Key Math Diagnostic Arithmetic Test
Everts Behavioral Rating Scale
General intelligence
General development
General verbal skills
Receptive language
Reading
Single word reading
General math skills
Behavior disorders
The Faroe Islands investigators assembled a birth cohort of 1,353 newborns recruited
from 3 hospitals over a 21 month period in 1986-1987. In 1,022 women, two biomarkers of
prenatal mercury exposure were collected: cord-blood mercury, and maternal hair mercury at
delivery. Neurodevelopmental assessments of 917 children were conducted at age 7 (Grandjean
et al. 1997). For these 917 children, the geometric mean concentration of mercury in cord-blood
was 22.6 parts per billion (ppb) (interquartile range 13.1 - 40.5 ppb, full range 0.9 - 351 ppb).
The geometric mean concentration of mercury in maternal hair was 4.2 ppm (interquartile range:
2.5-7.7 ppm, full range 0.2 - 39.1 ppm) (Budtz-Jorgensen et al. 2004a). Neurodevelopmental
assessments of the children were conducted at age 7 years (Grandjean et al. 1997). Table 9-2
lists each test administered and the corresponding general functional domain.
Table 9-2. Neurobehavioral Tests Administered at the 7-Year Evaluations in the Faroe
Islands Study
Test
Primary Domain Assessed
Wechsler Intelligence Scale for Children-Revised
(selected subtests)
Digit span
Similarities
Block Design
Bender-Gestalt Test
California Verbal Learning Test-Children
Boston Naming Test
Tactual Performance Test
Neurobehavioral Evaluation System (NES) (selected tests)
Finger tapping
Hand-eye coordination
Continuous performance test
Profile of Mood States
Child Behavior Checklist (selected items)
Short-term memory
Abstract verbal reasoning
Constructional praxis
Visual-motor integration
Verbal learning and memory
Confrontational naming
Nonverbal memory
Motor speed
Hand-eye coordination
Vigilance
Mood
Behavior disorders
9-3
-------
In assembling the Seychelles Child Development Study sample, investigators obtained
hair samples from 779 pregnant women and ultimately enrolled a study sample consisting of 740
newborns. The mean maternal hair mercury level was 6.8 ppm (range 0.9-25.8 ppm) (Davidson
et al. 1998). Neurodevelopmental assessments were conducted when the children were 6.5, 19,
29, and 66 months, and at 9 years. The mean maternal hair mercury level for the 643 children
who participated in the assessment at age 9 years was 6.9 ppm (standard deviation 4.5 ppm)
(Myers et al. 2003). Table 9-3 lists the tests administered at this age and the corresponding
general functional domain.
Table 9-3. Neurobehavioral Tests Administered at the 9-Year Evaluations in the
Seychelles Islands Study
Test
Primary Domain Assessed
Wechsler Intelligence Scale for Children-Third Edition
California Verbal Learning Test-Children
Boston Naming Test
Finger tapping
Continuous performance test
Developmental Test of Visual-Motor Integration
Bruininks-Oseretsky Test of Motor Proficiency (selected subtests)
Grooved Pegboard
Trail-Making Test
Woodcock-Johnson Tests of Achievement (selected subtests)
Letter-Word Identification
Applied Math
Wide Range Assessment of Memory and Learning
Design Memory subtest
Haptic Discrimination Test
Child Behavior Checklist
Connors' Hyperactivity Index
General intelligence
Verbal learning and memory
Confrontational naming
Motor speed
Vigilance
Visual-motor integration
Gross and fine motor skills
Manual dexterity
Visual tracking and executive function
Single word reading
Quantitative problem-solving
Visual memory
Cross-modal integration
Behavioral disorders
ADHD screener
9.3 Statistical Analysis
A statistical analysis was conducted to integrate data from the three studies to produce a
single estimate of the IQ dose-response relationship. Details of the analysis, including statistical
model formulation, selection of input values, results and sensitivity analysis are reported in Ryan
(2005) and are summarized below.
Data available for this analysis consisted of dose-response coefficients estimated by the
investigators for each of the three studies. These coefficients express a central estimate of the
average reduction in children's scores in tests of IQ (or other tests of cognitive performance) for
a one unit change in the mercury body burden of the mother during pregnancy.
A Bayesian hierarchical statistical model was used to estimate the integrated dose-
response coefficient. This is similar to the approach used by the NRC panel to calculate a
9-4
-------
benchmark dose value integrating data from all three studies (NRC 2000). A more technical
description of these same methods has been provided by Coull et al. (2004). The model makes
use of dose-response coefficients for IQ, and also incorporates coefficients for other cognitive
tests conducted in the studies, in an effort to obtain more robust estimates of the IQ relationship
that account for within-study (endpoint-to-endpoint) variability as well as variability across
studies. As compared with a model that uses only the IQ dose-response coefficients and their
variances, this approach makes use of more data from the studies to better characterize the
variability in estimation of the integrated IQ coefficient.
The key parameter inputs to the statistical model are the estimated IQ dose-response
coefficients for each study. The Wechsler Intelligence Scales for Children (WISC) is a standard
test of childhood IQ that was used in each of the three studies. The version of the test
administered in the Seychelles Islands (WISC-III) was different from that used in New Zealand
and the Faroe Islands (both WISC-R). As part of the standardization of the WISC-III, however,
both versions were administered to approximately 200 children. The correlation between the
Full-Scale IQ scores for the two versions of the WISC was 0.89; thus the WISC-R and WISC-III
appear to measure the same constructs and generate scores with similar dispersion (Wechsler
1991).
For the New Zealand study, full-scale IQ dose-response coefficients were reported in
Crump et al. (1998). In Table III of this paper, two coefficients for full scale IQ are reported:
one with the complete cohort, and the other for which one very influential observation (with
unusually high maternal hair mercury) was excluded. The NRC Committee on the Toxicological
Effects of Methylmercury reviewed the influence of the one outlier on the model outcome in
comparison to a model without this outlier, and determined that exclusion of the outlier was
reasonable and appropriate (NRC, 2000). In keeping with the conclusions by the NRC
committee, this analysis uses the coefficient from the regression in which the outlier child was
excluded: an IQ change of-0.53 IQ points (95% confidence interval -1.1, 0.069) for each ppm
of mercury in maternal hair.
For the Seychelles study, a 2003 paper reports results for IQ tests administered at age 9
(Myers et al. 2003). This analysis uses the coefficient from Table 9-2 of this study: an IQ
change of-0.13 IQ points (95% confidence interval -0.33, 0.07) for each ppm of mercury in
maternal hair.
The WISC-R includes 10 core subtests and 3 supplementary subtests. For the Faroes
study, the investigators did not administer the complete version of the WISC-R because of their
conclusion that a methylmercury-associated deficit in a broad measure such as Full-Scale IQ
provides relatively little insight into the specific nature of methylmercury's neuropsychological
effects on children. The Faroes investigators did administer three of the WISC-R subtests to the
children in their study (Similarities, Block Design, and Digit Span). Thus, to include data from
the Faroe Islands in this integrated assessment of prenatal mercury exposure on childhood IQ, it
was necessary to estimate a Faroe Islands dose-response coefficient for full-scale IQ from the
three available subtests.
Information on correlations between WISC-R subtest scores and WISC-R full-scale IQ
scores is available to assess the validity of a full-scale IQ estimated from the subtests.
9-5
-------
Similarities and Block Design are core subtests of the WISC-R, and Digit Span is a
supplementary subtest. The WISC-R was standardized on a nationally-representative sample of
U.S. children ages 6 to 16 years. Based on subtest scores, Sattler (1988) identified the pair, triad,
quartet, etc. of subtests that provides the most valid estimate of full-scale IQ. Only the 10 core
subtests were considered in this exercise. Of the 45 possible combinations of 2 core subtests
(i.e., 10 subtests taken 2 at a time), the combination of Similarities and Block Design, the two
core subtests administered in the Faroe Islands study, ranked 3rd in the magnitude of the validity
coefficient (0.885). The top-ranked combination was Vocabulary and Block Design (0.906). The
combination ranked 2nd was Information and Block Design (0.888). It is reasonable to expect
that taking into account Digit Span scores, the supplementary subtest administered, will increase
the validity coefficient. The results of this exercise indicate that combining the scores of the
Faroese children on Similarities, Block Design and Digit Span will provide valid estimates of
their full-scale IQ scores.
In support of this integrated IQ dose-response analysis, the Faroes research team
conducted further analysis to estimate a full-scale IQ dose-response coefficient based on the data
for the three subtests. This new Faroes analysis makes use of a structural equation model similar
to that described in Budtz-Jorgensen et al. (2002), and is reported in Budtz-Jorgensen et al.
(2005). As with other reports on results of the Faroe Islands study, the full-scale IQ coefficient
for the Faroes data is reported using cord blood mercury as the marker for exposure; results from
New Zealand and the Seychelles are presented in terms of maternal hair mercury. The Faroes
coefficient was converted to terms of hair mercury, using the reported median maternal haircord
blood mercury ratio for the Faroes cohort of approximately 200 (Budtz-Jorgensen et al. 2004a).
After conversion of the Faroes estimate from terms of cord blood mercury to hair mercury, the
estimated Faroes coefficient is an IQ change of-0.12 IQ points (95% confidence interval -0.24, -
0.01) for each ppm of mercury in maternal hair (Ryan 2005).
The IQ dose-response estimates for each of the three studies, along with 95% confidence
intervals, are shown in Table 9-4 and Figure 9-1.
Table 9-4. Relationship Between Maternal Mercury Body Burden and IQ in Three
Studies: IQ Decrement per ppm of Maternal Hair Mercury
Study
New Zealand
Seychelles
Faroe Islands
Ryan (2005) Integrative
Analysis - Main Case
Regression Coefficient
(95% Confidence Interval)
-0.53
(-1.1,0.069)
-0.13
(-0.33, 0.07)
-0.12
(-0.24, -0.01)
-0.13
(-0.28, -0.03)
Notes
Reported in Table III of Crump (1998); outlier
child omitted.
Reported in Table 2 of Myers (2003).
Reported in Ryan (2005), based on structural
equation modeling of three IQ subtests by
Budtz-Jorgensen et al. (2005).
see text and Ryan (2005)
9-6
-------
d ~
o
o
"wlr
*®
o *-? "
<**<
«— ~"
I
1
N
a
'_ I _
1 }" ""
o
o
i
Z Secyehelles Faroes
Figure 9-1. 95% Confidence Intervals for Full Scale IQ from the New Zealand, Seychelles
and Faroes Studies
Coefficients for other cognitive outcomes used in the model were obtained from the same
sources as the IQ dose-response coefficients (Crump et al. 1998, Myers et al. 2003, and Budtz-
Jorgensen et al. 2005). Tests included in the model included: California Verbal Learning Test
(Faroes and Seychelles); Boston Naming Test (Faroes and Seychelles); the Wide-Range
Assessment of Memory and Learning (WRAML) and Visual-Motor Integration (VMI)
(Seychelles only); Bender Visual Motor Gestalt Test (Faroes only); and the Test of Language
Development - Spoken Language (TOLD-SL), WISC performance IQ, and McCarthy Scales of
Children's Abilities perceptual performance scale (New Zealand only). Criteria for selection of
these outcomes and details on how they were used in the model are described in the report on the
statistical analysis (Ryan 2005).
The statistical analysis produced a dose-response relationship, integrating data from all
three studies, with a central estimate of an IQ change of-0.13 IQ points (95% confidence
interval -0.28, -0.03) for every ppm of mercury in maternal hair. This central estimate is close to
the values for the Faroes and Seychelles studies, suggesting relatively little influence on the
9-7
-------
integrated value from the larger coefficient estimated in the New Zealand study. The smaller
influence of the New Zealand coefficient is due to the smaller size of the cohort in this study, as
well as the greater uncertainty in the central estimate of the dose-response coefficient in this
study, as depicted in Figure 9-1.
Several sensitivity analyses were conducted, reflecting the following variations in the
inputs to the model:
• Use of only the IQ dose-response coefficients, without the coefficients for the
additional cognitive endpoints;
Use of an alternate hainblood mercury ratio for calculation of the Faroes
coefficients in hair mercury terms;
Use of the New Zealand coefficients that include one highly influential
observation; and
• Use of an alternate interpretation of the Faroes structural equation model outputs
for the IQ dose-response coefficient.
These sensitivity analyses found very consistent results, with central estimates all in the
approximate range of-0.10 to -0.25 IQ points for each ppm of mercury in maternal hair.
The results consistently suggested a significant association between mercury and IQ, with lower
confidence limits ranging from about -0.2 to -0.5, and upper confidence limits between -0.02 and
-0.04. Details of the sensitivity analyses are presented in the statistical analysis report (Ryan
2005).
9.4 Strengths and Limitations of the IQ Dose-Response Analysis
This analysis has produced, for the first time, an estimate of the relationship between
maternal mercury body burdens during pregnancy and childhood IQs that incorporates data from
all three epidemiologic studies judged by the NRC to be of high quality and suitable for risk
assessment. The statistical approach makes use of all the available data (including information
on results for related tests of cognitive function), and can be used to produce population-based
estimates of a health outcome that can be readily monetized for use in benefit-cost analysis'.
1 There is limited evidence directly linking IQ and methylmercury exposure in the three large epidemiological
studies that were evaluated by the NAS and EPA. Based on its evaluation of the three studies, EPA believes that
children who are prenatally exposed to low concentrations of methylmercury may be at increased risk of poor
performance on neurobehavioral tests, such as those measuring attention, fine motor function, language skills,
visual-spatial abilities (like drawing), and verbal memory. For this analysis, EPA is adopting IQ as a surrogate for
the neurobehavioral endpoints that NAS and EPA relied upon for the RiD.
In the Faroes Island Study, a full scale IQ evaluation was not conducted. However, two core subtests were evaluated
(Similarities and Block design) and one supplementary test was conducted (Digit Span). The Similarities and Block
Design tests are reported to be well correlated with the full WISC-R battery (0.885, see Bellinger (2005)), but how
the Digit Span test relates is not reported. In the EPA analysis, we assume that it relates similarly. In the Faroes
study, performance scores on the Similarities and Block Design tests were not shown to be statistically related to
9-8
-------
There are several aspects of IQ as a metric for neurodevelopmental effects in this benefit-
cost analysis that are important to recognize.
Full-Scale IQ is a composite index that averages a child's performance across many
functional domains, providing a good overall picture of cognitive health. An extensive body of
data documents the predictive validity of full-scale IQ, as measured at school-age, and late
outcomes such as academic and occupational success (Neisser et al. 1996). In addition, methods
are readily available for valuing shifts in IQ and thus conducting a benefits analysis of
interventions that shift the IQ distribution in a population. Methods for monetization of the other
tests administered in the three studies have not been developed.
It is important to recognize, however, that full-scale IQ might not be the cognitive
endpoint that is most sensitive to prenatal mercury exposure. Significant inverse associations
were found, in both the New Zealand and Faroe Islands studies, between prenatal mercury levels
and neurobehavioral endpoints other than IQ. If the effects of mercury are highly focal, affecting
only specific cognitive functions, taking full-scale IQ as the primary endpoint for a benefits
analysis might underestimate the impacts. In averaging performance over diverse functions in
order to compute full-scale IQ, the specific effects of mercury on only certain of these functions
would be "diluted," and the estimated magnitude of the change in performance per unit change
in the mercury biomarker would be underestimated.
Moreover, it is well-known that there may be substantial deficits in cognitive well-being
even in individuals with normal or above average IQ. The criterion most frequently used to
identify children with learning disabilities for the purposes of assignment to special education
services is a discrepancy between IQ and achievement. Specifically, the child's achievement in
reading, math, or other academic areas is significantly lower than what would be expected, given
his or her full-scale IQ. Thus, there are deficits in cognitive functioning that are not captured by
IQ scores. For example, two of the most sensitive endpoints in the Faroe Islands study were the
Boston Naming Test, which assesses word retrieval, and the California Verbal Learning
Test-Children, which assesses the acquisition and retention of information presented verbally.
Depending on the severity of the deficits, a child who has deficits in either of these skills could
be at a considerable disadvantage in the classroom setting and at substantial educational risk.
Neither of these abilities is directly assessed by the WISC-R or WISC-III, however, and so do
not explicitly contribute to a child's IQ score.. Therefore, benefits calculations relying solely on
IQ decrements are likely to underestimate the benefits to cognitive functioning of reduced
mercury exposures. In additions, impacts on other neurological domains (such as motor skills
cord blood or maternal mercury levels; the Digit Span test did show a statistical relationship with cord blood
mercury.
Both the New Zealand and Seychelles study administered the WISC IQ test (WISC III in Seychelles, WISC
R in New Zealand). A reanalysis of the New Zealand data found a positive association, but it was not statistically
significant. No significant associations were seen in the Seychelles study. As displayed in Figure 5 of Ryan (2005),
the confidence intervals for full scale IQ in both these studies include zero. However, Ryan conducted an integrative
analysis, combining results from all three studies. When combined, the statistical power of the analysis increases.
While the size of the dose-response relationship declined relative to past studies with a statistically significant
finding, Ryan found a statistically significant relationship between IQ and mercury. The confidence interval did not
include zero.
9-9
-------
and attention/behavior) are not represented by IQ scores and thus are also excluded from the
benefits analysis.
As discussed above, the Faroe Islands study did not include testing for full-scale IQ. For
this analysis, an estimate of a dose-response coefficient for full-scale IQ was estimated using the
three subtests. While this extrapolation introduces some uncertainty, information has been
presented that demonstrates a high correlation between the subtests and full-scale IQ scores.
While the Seychelles and New Zealand studies use maternal hair mercury as the exposure
biomarker, the Faroe Islands study uses cord blood mercury. For purposes of the integrated
analysis, it was necessary to express results from all three studies in the same terms. Several
studies have examined the relationship between hair mercury and blood mercury, and have
reported hairblood ratios typically in the range of 200 to 300 (see ATSDR 1999, pages 249-252
for a review). However, these studies generally do not use cord blood mercury, which is the
exposure metric in the Faroes study. A recent analysis found that mercury concentrations in cord
blood are, on average, 70 percent higher than those in maternal blood (Stern and Smith 2003).
For conversion of Faroes data from cord blood mercury to maternal hair mercury, it was most
appropriate to use data specific to this population, indicating a median maternal haincord blood
mercury ratio of 200 (Budtz-Jorgensen et al. 2004a). An alternate ratio of 250 was examined in
sensitivity analysis, and resulted in an integrated dose-response coefficient that is reduced by
about 12 percent (central estimate of-0.131 vs. -0.115).
This analysis relies on use of summary statistics, i.e. dose-response coefficients and
associated variability statistics, for each of the three studies. Original data were not available for
this analysis. While a lack of original data is often cited as a problem for cross-study analyses,
its impact is lessened in this application for several reasons. All three studies had careful
epidemiological designs that measured a variety of important potential confounders such as
maternal age and education. All estimated dose-response coefficients were derived from well
documented regression models that adjusted for age, maternal education and other important
factors. Also, the National Research Council had asked the individual study investigators for
very specific details about the way in which their analyses had been done and had also asked for
additional analyses where necessary. Further, work by Dominici et al. (2000) took a similar
approach for hierarchical modeling of estimated dose-response coefficients extracted from
separate studies.
The major uncertainty concerning the New Zealand study is the strong influence of one
child in the study population with a particularly high maternal hair mercury level. Published
analyses of the New Zealand study presented results with data for this child both included and
excluded (Crump et al. 1998). In keeping with the conclusions of the NRC (2000), the
integrated dose-response analysis presented in this section made use of the dose-response
coefficients calculated with this child omitted. A sensitivity analysis using the New Zealand
coefficient with this child included results in an integrated dose-response coefficient that is
reduced by about 17 percent (central estimate of-0.131 vs. -0.108).
Some uncertainty is also associated with the Seychelles study due to the exclusion of
some members of the cohort from the data reported by Myers et al. (2003) and used as input to
this integrated dose-response analysis. The Seychelles researchers did not include a small
9-10
-------
number of outliers (defined as observations with model residuals exceeding 3 standard deviation
units), and no results are available for the full cohort. However, the authors report that "In all
cases, the association between prenatal MeHg exposure and the endpoint was the same,
irrespective of whether outliers were included" (Myers et al. 2003).
Finally, the integrated dose-response analysis assumes the exposures assigned to each
study subject are accurate representations of true exposure. In reality, there is likely to be some
discrepancy between measured and actual exposures, for example, due to variation in hair length.
Alternatively, the true exposure of interest may have been during the first trimester of pregnancy,
whereas exposures in maternal hair and cord blood measured at birth reflect exposures later in
pregnancy. Presence of exposure measurement error could introduce a bias in the results, most
likely towards the null (Budtz-Jorgensen et al. 2004b).
9.5 References
ATSDR (1999). Toxicological Profile for Mercury. Agency for Toxic Substances and Disease
Registry.
Bellinger DC (2005). Neurobehavioral Assessments Conducted in the New Zealand, Faroe
Islands, and Seychelles Islands Studies of Methylmercury Neurotoxicity in Children.
Report to the U.S. Environmental Protection Agency.
Budtz-Jorgensen E, Keiding N, Grandjean P, Weihe P (2002). Estimation of health effects of
prenatal methylmercury exposure using structural equation models. Environmental
Health,\(\):2.
Budtz-Jorgensen E, Grandjean P, Jorgensen P, Weihe P, Keiding N (2004a). Association
between mercury concentrations in blood and hair in methylmercury-exposed subjects at
different ages. Environmental Research, 95(3):385-93.
Budtz-Jorgensen E, Keiding N, Grandjean P (2004b). Effects of exposure imprecision on
estimation of the benchmark dose. Risk Analysis, 24(6): 1689-96.
Budtz-Jorgensen E, Debes F, Weihe P, Grandjean P (2005). Adverse mercury effects in 7 year-
old children expressed as loss in "IQ." Report to the U.S. Environmental Protection
Agency.
Coull BA, Mezzetti M, Ryan LM (2003). A Bayesian hierarchical model for risk assessment of
methylmercury. Journal of Agricultural, Biological & Environmental Statistics,
8(3):253-270.
Crump KS, Kjellstrom T, Shipp AM, Silvers A, Stewart A (1998). Influence of prenatal
mercury exposure upon scholastic and psychological test performance: Benchmark
analysis of aNew Zealand cohort. Risk Analysis, 18:701-713.
Davidson PW, Myers GJ, Cox C, Axtell C, Shamlaye C, Sloane-Reeves J, Cernichiari E,
Needham L, Choi A, Wang Y, Berlin M, Clarkson TW (1998). Effects of prenatal and
9-11
-------
postnatal methylmercury exposure from fish consumption on neurodevelopment:
outcomes at 66 months of age in the Seychelles Child Development Study. Journal of
the American Medical Association, 280(8):701-7.
Dominici F, Samet JM, Zeger SL (2000). Combining evidence on air pollution and daily
mortality from the 20 largest US cities: a hierarchical modeling strategy. Journal of the
Royal Statistical Society A, 163:263-284.
Environmental Protection Agency (EPA 2002). Mercury Neurotoxicity Workshop Notes;
available at:
Grandjean P, Weihe P, White RF, Debes F, Araki S, Yokoyama K, Murata K, Sorensen N, Dahl
R, Jorgensen PJ (1997). Cognitive deficit in 7-year-old children with prenatal exposure
to methylmercury. Neurotoxicology and Teratology, 19:417-428.
Kjellstrom T, Kennedy P, Wallis S, Stewart A, Friberg L, Lind B, et al. (1989). Physical and
mental development of children with prenatal exposure to mercury from fish. National
Swedish Environmental Protection Board Report No. 3642.
Myers GJ, Davidson PW, Cox, C, Shamlaye CF, Palumbo D, Cernichiari E, Sloane-Reeves J,
Wilding GE, Kost J, Huang LS, Clarkson TW (2003). Prenatal methylmercury exposure
from ocean fish consumption in the Seychelles child development study. Lancet,
361:1686-1692.
NRC (2000). Toxicological Effects of Methylmercury. National Research Council. Washington,
DC: National Academies Press.
Neisser U, Boodoo G, Bouchard TJ, et al. (1996). Intelligence: Knowns and unknowns.
American Psychologist, 51:77-101.
Ryan, LM (2005). Effects of Prenatal Methylmercury on Childhood IQ: A Synthesis of Three
Studies. Report to the U.S. Environmental Protection Agency.
Sattler JM (1988). Assessment of Children, 3rd Edition. San Diego: Jerome M. Sattler Publisher.
Stern, AH, Smith AE (2003). An assessment of the cord blood:maternal blood methylmercury
ratio: Implications for risk assessment. Environmental Health Perspectives, 111:1465-
1470.
Wechsler D (1991). W1SC-III Manual. San Antonio: The Psychological Corporation.
9-12
-------
SECTION 10 EXPOSURE MODELLING AND BENEFIT METHODOLOGY WITH AN
APPLICATION TO A NO-THRESHOLD MODEL 10-1
10.1 Introduction 10-1
10.1.1 Summary 10-2
10.1.2 Modeling Overview 10-6
10.1.3 Monetized Benefits: Results in Brief 10-8
10.1.4 Key Steps 10-11
10.2 Estimation of Mercury Levels in Freshwater Fish 10-12
10.3 Estimation of Exposed Populations and Fishing Behaviors 10-18
10.3.1 Primary Data Sources on Fishing Activity in the United States .... 10-18
10.3.2 Population Centroid Approach 10-24
10.3.3 Angler Destination Approach 10-36
10.4 Estimation of Mercury Exposures, IQ Decrements, and Lost Future Earnings
10-42
10.4.1 Modeling Approach for Estimating Individual Exposures 10-43
10.4.2 Modeling Approach for Estimating IQ Effects and Lost Earnings .. 10-45
10.5 Model Results: Estimated Benefits of Utility Mercury Emission Controls . 10-47
10.5.1 Results for the Population Centroid Approach 10-53
10.5.2 Results for the Angler Destination Approach 10-67
10.5.3 Comparison of Results from Two Approaches 10-89
10.5.4 Sensitivity Analysis of Alternative Dose-Response Functions 10-96
10.5.5 Distribution of Per-Capita IQ Changes for the Exposed Population (in
support of distributional equity analysis) 10-97
10.6 Analysis of Potentially High-Risk Subpopulations 10-103
10.6.1 Mercury Ingestion Estimates for Individuals in the Upper Range of the
Fish Consumption Distribution 10-104
10.6.2 Mercury Ingestion Estimates for Individuals in Low Income, High Fish
Consumption Households 10-110
10.6.3 Mercury Ingestion Estimates for Two Selected Ethnic Populations
10-112
10.6.4 Adaptation of the Population Centroid Approach to Estimate Exposed
Hmong and Chippewa Population 10-119
10.6.5 Sensitivity Analysis Examining the Economic Benefit Equity Issue in the
Context of High Fish Consuming (subsistence) Populations Including
Native Americans 10-129
10.7 Discussion and Qualification of Results: Assumptions, Limitations, and
Uncertainties 10-134
10.7.1 Mercury Concentration Estimates 10-135
10.7.2 Exposed Population Estimates 10-137
10.7.3 Matching of Exposed Populations to Mercury Concentrations .... 10-138
10.7.4 Fish Consumption Estimates 10-140
10.7.5 Modelling and Valuation of IQ Related Effects 10-141
10.7.6 Unquantified Benefits 10-142
10.8 References 10-144
Tables
Table 10-1 (a). Summary of Per Capita Changes in IQ Due to Mercury Exposure 10-4
-------
Table 10-l(b). Impacts of Mercury on High Fish Consuming Groups 10-6
Table 10-l(c). Summary of Total Benefits Associated with Modelled Avoided IQ Decrements in
Prenatally Exposed Children Due to Reduced Mercury Exposure from Freshwater
Recreational Angling 10-10
Table 10-2. Summary Statistics for Estimated Fish Tissue Mercury Concentrations (ppm) by
State: 2001 Base Case3 10-16
Table 10-3. HUC-Level Distribution of Mercury Sampling Sites and Estimated Fish Tissue
Concentrations: 2001 Base Case" 10-17
Table 10-4. Summary of Fishing Activity Levels by State in 2001 from NSFHWR 10-20
Table 10-5. Overview of Key Attributes of the Population Centroid and Angler Destination
Models 10-23
Table 10-6. Block Group Demographic Characteristics by State (in 2000): Data Used in
Population Centroid Approach 10-30
Table 10-7. Estimated Annual Number of Prenatally Exposed Children for Selected Lag Periods
from 2001: Population Centroid Approach 10-31
Table 10-8. Estimated Annual Number of Prenatally Exposed Children for Selected Lag Periods
from 2020: Population Centroid Approach 10-33
Table 10-9. Average Estimated Mercury Concentrations (ppm) in Freshwater Fish by Distance
Interval from Block Group Centroids: Base Case 2001 10-35
Table 10-10. State-Level Summary of Exposed Population Estimates: Angler Destination
Approach 10-42
Table 10-11. Effects of Emission Control Scenarios—Percent Reduction in Estimated Fish
Tissue Mercury Concentrations from 2001 Base Case8 10-49
Table 10-12. Effects of Emission Control Scenarios—Percent Reduction in Estimated Fish
Tissue Mercury Concentrations from 2020 Base Case with CAIR3 10-51
Table 10-13. Estimated Distribution of Mercury Ingestion by Distance Traveled to Fish:
Population Centroid Approach—2001 Base Case 10-54
Table 10-14. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and
Foregone Earnings: Population Centroid Approach—2001 Base Case3 10-55
Table 10-15. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and
Foregone Earnings: Population Centroid Approach—2020 Base Case with CAIR3
10-57
Table 10-16. 2020 Base Case with CAIR: Modelled Avoided Losses Relative to 2001 Base
Case (Applied to 2020 Demographics)—Population Centroid Approach31 b 10-61
Table 10-17. 2001 Utility Mercury Emissions Zero Out: Modelled Avoided Losses Relative to
2001 Base Case—Population Centroid Approach"-1' 10-63
Table 10-18. 2020 with CAIR Emissions Zero Out: Modelled Avoided Losses Relative to 2020
with CAIR Base Case—Population Centroid Approach3-b 10-65
Table 10-19. Estimated Benefits of 2020 CAMR Control Option 1: Relative to 2020 with
CAIR—Population Centroid Approach8-1" 10-68
Table 10-20. Estimated Benefits of 2020 CAMR Control Option 2: Relative to 2020 with
CAIR—Population Centroid Approach"-1' 10-70
Table 10-21. Summary of Annual Benefit Estimates: Population Centroid Approach3 ... 10-72
Table 10-22. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and
Foregone Earnings: Angler Destination Approach—2001 Base Case3 10-77
Table 10-23. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and
Foregone Earnings: Angler Destination Approach—2020 with CAIR3 10-81
-------
Table 10-24. 2020 Base Case with CAIR: Modelled Avoided Losses Relative to 2001 Base
Case Applied to 2020 Demographics—Angler Destination Approach8'b 10-83
Table 10-25. 2001 Utility Mercury Emissions Zero Out: Modelled Avoided Losses Relative to
2001 Base Case—Angler Destination Approach"-1" 10-85
Table 10-26. 2020 with CAIR Emissions Zero Out: Modelled Avoided Losses Relative to 2020
with CAIR Base Case—Angler Destination Approach8 b 10-87
Table 10-27. Estimated Benefits of 2020 With CAIR Control Option 1: Relative to 2020 with
CAIR—Angler Destination Approach8 b 10-90
Table 10-28. Estimated Benefits of 2020 With CAIR Control Option 2: Relative to 2020 with
CAIR—Angler Destination Approach8-15 10-92
Table 10-29. Summary of Annual Benefit Estimates: Angler Destination Approach 10-94
Table 10-30. Summary and Comparison of Annual Benefit Estimates: Population Centroid
Approach vs. Angler Destination Approach 10-95
Table 10-31. Summary and Comparison of Annual Benefit Estimates Under Alternative IQ
Dose-Response Assumptions: Population and Angler Destination Approach 10-96
Table 10-32. Summary of Estimated Mercury Exposures for Consumption-Based Subsistence
Population, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2001 Base Case8 10-105
Table 10-33. Summary of Estimated Mercury Exposures for Consumption-Based Subsistence
Population, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2020 with CAIR8 10-107
Table 10-34. Summary of Annual Benefit Estimates for Consumption-Based Subsistence
Population: Population Centroid Approach 10-109
Table 10-35. Summary of Estimated Mercury Exposures for Income-Based Subsistence
Population, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2001 Base Case8 10-114
Table 10-36. Summary of Estimated Mercury Exposures for Income-Based Subsistence
Population, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2020 with CAIR8 10-116
Table 10-37. Summary of Annual Benefit Estimates for Income-Based Subsistence Population:
Population Centroid Approach 10-118
Table 10-38. Block Group Demographics for Hmong and Chippewa Females, Aged 15 to 44 (in
2001) 10-120
Table 10-39. Estimated Annual Number of Prenatally Exposed Children from Special
Populations for Selected Lag Periods: Population Centroid Approach 10-124
Table 10-40. Summary of Estimated Mercury Exposures for Special Populations in 2001, with
Associated IQ Decrements and Foregone Earnings: Population Centroid
Approach—Base Case 2001 10-125
Table 10-41. Summary of Estimated Mercury Exposures for Special Populations in 2020, with
Associated IQ Decrements and Foregone Earnings: Population Centroid
Approach—Base Case 2020 with CAIR8 10-126
Table 10-42. Summary of Annual Benefit Estimates for Hmong Special Population: Population
Centroid Approach 10-127
Table 10-43. Summary of Annual Benefit Estimates for Chippewa Special Population:
Population Centroid Approach 10-128
Table 10-44. Results of the Sensitivity Analysis Examining Distributional Equity for Native
American (subsistence) Populations 10-133
-------
Table 10-45. Unqualified Health and Ecosystem Effects Associated with Exposure to Mercury
10-142
Figures
Figure 10-1. Locations of Lake Fish Tissue Mercury Sampling Sites Used in the Analysis
10-14
Figure 10-2. Locations of River Fish Tissue Mercury Sampling Sites Used in the Analysis
10-15
Figure 10-3. Flow Diagram for Population Centroid Approach 10-25
Figure 10-4. Population Centroid Approach: Linking Census Block Groups to Demographic
Data and Mercury Fish Tissue Samples 10-27
Figure 10-5. Flow Diagram for Angler Destination Approach 10-37
Figure 10-6. Estimated Distribution of Lake-Fishing Days Across HUCs in 2001 10-40
Figure 10-7. Estimated Distribution of River-Fishing Days Across HUCs in 2001 10-41
Figure 10-8. Spatial Distribution of Estimated Average Daily Maternal Mercury Ingestion
Rates: Angler Destination Approach—2001 Base Case 10-74
Figure 10-9. Spatial Distribution of Estimated IQ Decrements per HUC: Angler Destination
Approach—2001 Base Case 10-75
Figure 10-10. Spatial Distribution of Estimated Percent Reduction in IQ Losses: Improvement
with 2001 Utility Emissions Zero Out Scenario (Zero Lag) 10-76
Figure 10-11. Distribution of Modelled Avoided IQ Decrements (Benefits) due to Mercury
Emissions Reductions: 2001 Utility Emissions Zero-Out Relative to 2001 Base Case;
Population Centroid Approach; Variable Consumption Rate 10-99
Figure 10-12. Cumulative Distribution of Modelled Avoided IQ Decrements (Benefits) due to
Mercury Emissions Reductions: 2001 Utility Emissions Zero-Out Relative to 2001 Base
Case; Population Centroid Approach; Variable Consumption Rate 10-99
Figure 10-13. Distribution of Modelled Avoided IQ Decrements (Benefits) due to Mercury
Emissions Reductions: CAMR Control Option 1 Relative to 2020 Base Case with CAIR;
Population Centroid Approach; Variable Consumption Rate 10-100
Figure 10-14. Cumulative Distribution of Modelled Avoided IQ Decrements (Benefits) due to
Mercury Emissions Reductions: CAMR Control Option 1 Relative to 2020 Base Case
with CAIR; Population Centroid Approach; Variable Consumption Rate 10-100
Figure 10-15. Distribution of Modelled Avoided IQ Decrements (Benefits) due to Mercury
Emissions Reductions: CAMR Control Option 2 Relative to 2020 Base Case with CAIR;
Population Centroid Approach; Variable Consumption Rate 10-101
Figure 10-16. Cumulative Distribution of Modelled Avoided IQ Decrements (Benefits) due to
Mercury Emissions Reductions: CAMR Control Option 2 Relative to 2020 Base Case
with CAIR; Population Centroid Approach; Variable Consumption Rate 10-101
Figure 10-17. Distribution of Modelled Avoided IQ Decrements (Benefits) due to Mercury
Emissions Reductions: 2020 Utility Emissions Zero-Out Relative to 2020 Base Case
with CAIR; Population Centroid Approach; Variable Consumption Rate 10-102
Figure 10-18. Cumulative Distribution of Modelled Avoided IQ Decrements (Benefits) due to
Mercury Emissions Reductions: 2020 Utility Emissions Zero-Out Relative to 2020 Base
Case with CAIR; Population Centroid Approach; Variable Consumption Rate ... 10-102
Figure 10-19. U.S. Census Tracts with Native American Populations 10-113
-------
SECTION 10
EXPOSURE MODELLING AND BENEFIT METHODOLOGY
WITH AN APPLICATION TO A NO-THRESHOLD MODEL
10.1 Introduction
In this section, we describe two exposure modeling approaches designed to provide an
estimate the underlying benefits analysis of reducing mercury emissions that may result from the
Clean Air Mercury Rule. Sections 10 and 11 together form the basis for our benefits
methodology and calculations. In this Section, we construct a scenario that reflects an upper-
bound on the number of people affected by mercury. In particular, the scenario incorporates an
assumption of no theshold. We estimate benefits with this assumption by deploying a very
disaggregated, spatially-rich model. This exercise provides very detailed results. In Section 11,
the model is simplified a bit by aggregating recreational fishers into descrete bins or categories,
making the analysis much more manageable. The analysis in Section 11 simulates exposure
scenarios under the assumption of thresholds. Two different thresholds are explored. The
threshold analysis gives "scaling factors" or benefits as a percent of the no threshold case
developed in this Section. Benefit estimates are then estimated by multiplying the scaling
factors by the benefits calculated in this Section. Hence, this Section forms the core analytic
underpinnings for the final benefit numbers that are derived and presented in Section 11.
In this Section, we quantify and monetize, to the extent feasible, benefits associated with
modelled avoided IQ deficits due to reduced exposure from the consumption of recreationally-
caught freshwater fish assuming there is no threshold in effects at low doses of mercury. The
analysis focuses on estimating changes in exposures to women of childbearing age because
adverse health effects in children have been linked to prenatal mercury exposures. In addition,
because mercury emissions in the U.S. predominantly affect the eastern-half of the country, the
analysis is also focused on affected populations in that part of the country. While the geographic
coverage and the exposed population largely reflect the areas impacted by the CAMR, it should
be noted that, as Section 4 discusses, this analysis focuses on freshwater exposures in the
eastern-half of the U.S., which will reduce the size of the exposed population considered for
analysis. This focus for the analysis is necessary because of limitations in modeling how
changes in mercury deposition will effect fish tissue concentrations for the other fish
consumption pathways discussed in Section 4 of this report and there is relatively little fish
tissue information for the Western-half of the U.S. As discussed in Section 8 the largest change
in power plant deposition associated with the recently finalized CAIR and CAMR program will
occur in the eastern-half of the U.S., so the unqualified benefits for the western-hald of the U.S.
is expected to be quite small. As is discussed in previous sections, we are unable to quantify
several categories of potential benefits, such as benefits from other health and ecological effects,
as well as commercial and recreationally-caught saltwater species. As metioned throughout the
report, power plant emission reductions under CAIR and CAMR will have a minimal effect on
exposure levels associated with these other consumption pathways. Our benefit assessment has
several known uncertainties and biases, which are discussed further in Section 10.7. While some
of these are downward biases and some are upward biases, taken together, the Agency believes
10-1
-------
that the benefits presented in this section likely underestimate the total benefits of reducing
mercury emissions from power plants due to the potential health effects and potentially exposed
populations that are not quantified in this analysis.
10.1.1 Summary
The basic methodology used in this Section is to project the change in IQ of a population
of children due to mercury exposure in utero. The exposure is based on consumption offish by
pregnant women. The mercury in the fish is due in part to atmospheric deposition of mercury
from power plants. A monetary value is placed on incremental loss of IQ by these children1.
The incremental reduction in exposure due to mercury emission reductions from power plants is
then applied to this methodology to calculate the improvement in IQ and the monetary value of
that improvement, attributable to the emissions reduction. The study examines only
consumption of freshwater fish, because our analysis indicates that these are the only fish
significantly impacted by U.S. power plants.
The analysis first examines impacts on the general population of children of freshwater
fishers. It then considers much smaller populations that consume greater amounts offish than
the general population, including subsistence fishers, certain Native Americans, and Asian
Americans.
With respect to impacts on the general population, two methods of approaching the
problem were used, a "Population Centroid" and an "Angler Destination" approach. These
approaches reflect different ways of estimating where freshwater fishers fish. Table 1 shows the
relative impacts on IQ deriving from different mercury emission rates and the two different
analytical approaches, for the average child in the general population. More detailed results are
presented in the body of this Section.
1 There is limited evidence directly linking IQ and methylmercury exposure in the three large epidemiological
studies that were evaluated by the NAS and EPA. Based on its evaluation of the three studies, EPA believes that
children who are prenatally exposed to low concentrations of methylmercury may be at increased risk of poor
performance on neurobehovioral tests, such as those measuring attention, fine motor function, language skills,
visual-spatial abilities (like drawing), and verbal memory. For this analysis, EPA is adopting IQ as a surrogate for
the neurobehavioral endpoints that NAS and EPA relied upon for the RfD.
In the Faroes Island Study, a full scale IQ evaluation was not conducted. However, two core subtests were evaluated
(similarities and block design) and one supplementary test was conducted (Digit Span). The similarities and block
design tests are reported to be well correlated with the full WISC-R battery (0.885 see Bellinger paper), but how the
Digit Span test relates is not reported. In the EPA analysis, we assume that it relates similarly. In the Faroes study,
performance scores on the similarities and block design tests were not shown to be statistically related to cord blood
or maternal mercury levels; the digit span test did show a statistical relationship with cord blood mercury.
Both the New Zealand and Seychelles study administered the WISC IQ test (WISC III in Seychelles, WISC R in
New Zealand). A reanalysis of the New Zealand data found a positive association, but it was not statistically
significant. No significant associations were seen in the Seychelles study. In the EPA analysis, the the confidence
intervals for full scale IQ in both these studies include zero. However, Ryan (2005) conducted an integrative
analysis, combining results from all three studies. When combined, the statistical power of the analysis increases.
While the size of the dose-reponse relationship declined relative to past studies with a statistically significant
finding, Ryan found a statistically significant relationship between IQ and mercury. The confidence interval did not
include zero.
10-2
-------
The data in Table 10-la show that both analytical approaches yield similar results. A
typical child of freshwater fishers lost approximately 0.06 - 0.07 IQ points due to mercury
exposure in 2001, depending on the analytical approach. Average IQ is, by definition, 100
points. Implementing CAIR would reduce this IQ loss by a little less than 0.007 to 0.009 IQ
points in 2020. Under Options 1 and 2 of the CAMR, this reduction would be increased by
0.0006 to 0.0009 IQ points. Total elimination of power plant emissions would have about the
same effect as CAIR Option 2. Focusing on the Population Centroid approach, it is seen that
CAIR reduces the 2001 mercury impact on IQ by 11.8%; CAIR plus CAMR Option 1 reduces
the impact by 12.7% (an additional 0.9%); and totally eliminating power plant mercury
emissions reduces the 2001 impact by 13.2% (another 0.5% beyond CAIR plus CAMR Option 1.
10-3
-------
Table 10-l(a). Summary of Per Capita Changes in IQ Due to Mercury Exposure
Measurement of IQ impact per capita (average impact over study
population)
IQ loss due to mercury exposure in 2001
IQ loss in 2020 under CAIR emission reductions
Avoided IQ loss due to CAIR in 2020, relative to 2001
Avoided IQ loss with no power plant emissions in 2020, relative to
2001
Avoided IQ loss w/ no power plant emissions in 2020, vs CAIR in 2020
IQ loss in 2020 w/ CAIR & Option 1 of CAMR
Avoided IQ loss due to CAIR & Option 1 , vs CAIR alone
IQ loss in 2020 w/ CAIR & Option 2 of CAMR
Avoided IQ loss due to CAIR & Option 2, vs CAIR alone
Approach
Population
Centroid
0.0621
0.0548
0.0073
0.0082
0.0009
0.0542
0.0006
0.0539
0.0009
Angler
Destination
0.069
0.060
0.0089
0.0090
0.0001
0.0594
0.0006
0.0591
0.0009
10-4
-------
In short, the overall impact of mercury on the IQ of children in the general population is
relatively small, less than one-thousanth of a normal IQ (Normal = 100). Implementing CAIR
dramatically reduces the contribution of power plants to this small projected mercury impact on
children, and CAMR Option 1 eliminates the majority of the remaining impact associated with
power plants.
We apply a value of about $8,800 (net present value) per IQ point improvement per
capita'. Thus, the value of CAMR Option 1 is equal to the number of exposed children x the
mean improvement in IQ (0.0006 points) x $8,8002, or $2.6 million. In the body of this Section,
this number is adjusted to reflect various "lag" periods, to reflect the amount of time required for
emission reductions to result in changes in fish mercury concentrations.
In addition to the analysis of the general US population, this Section also assesses the
benefit of CAMR on subsistence anglers, Native Americans, and Asian Americans, who
consume more fish than the general population. Table 2 presents results similar to Table 1, for
subsistence fishers and two Native American tribes.
As expected, a larger impact on IQ due to mercury was found for these smaller groups,
with an average IQ impact in 2001 from all mercury sources of 0.331 IQ points on children of
subsistence fishers, for example. Implementation of CAIR reduced this impact to 0.290 IQ
points in 2020. Application of CAMR Option 1 reduced impacts by an additional 0.0033 points
and complete elimination of emissions from US power plants reduced impacts by another 0.012
points, or down to 0.275 points. Hence, for this more sensitive group, CAIR again provides the
bulk of the reduction possible by controlling power plants, but in this case totally eliminating
power plant emissions can provide about a one-hundredth of a point of improvement in average
IQ, compared to CAMR Option 1.
For the Native American tribes (the Hmong and the Chippewa), current impacts on IQ
are estimated to be about 0.1 IQ point. CAIR provided no benefit to the Hmong, where power
plants contribute only about 6% of the impact associated with mercury consumption. For the
Chippewa, CAIR reduced impacts about 11%. The CAMR options reduced impacts another
0.8% or 1.5%, and total elimination of power plant emissions would contribute another 8.5%
reduction. In absolute terms, the effect of total elimination of power plant mercury emissions,
beyond CAMR Option 1, was projected to be about one-hundredth of an IQ point.
2 This value is based on foregone earnings over a lifetime discounted at 3 percent. The value per IQ point when
calculated at a 7 percent discount rate is $1580 per IQ point (1999$).
10-5
-------
Table 10-l(b). Impacts of Mercury on High Fish Consuming Groups
Measurement of IQ impact per capita
IQ loss due to Hg Exposure in 2001, mean per capita
IQ loss in 2020 under CAIR
Avoided IQ loss due to CAIR 2020, v 2001
Avoided IQ loss w/ no PP Hg 2020, v 2001
Avoided IQ loss w/ no PP Hg 2020, v CAIR 2020
IQ loss in 2020 w/ CAIR & CAMR Opt1
Avoided IQ loss due to CAIR & CAMR Opt 1, v CAIR
IQ loss in 2020 w/CAIR & CAMR Opt2
Avoided IQ loss due to CAIR & CAMR Opt2, v CAIR
Number of children in group
Subsistence
Fishers
0.3310
0.2900
0.0410
0.0154
0.2867
0.0033
0.2852
0.0048
22,302
Hmong
(MN, Wl)
0.1140
0.1140
(0.0007)
0.0069
0.1136
0.0004
0.1126
0.0014
553
Chippewa
(Ml, MN, Wl)
0.1340
0.1220
0.0150
0.0130
0.1210
0.0010
0.1200
0.0020
1,094
10.1.2 Modeling Overview
The mercury benefits model developed to support this analysis estimates the IQ
decrement for children of recreational freshwater fishers exposed prenatally to methylmercury
through maternal fish consumption. The model is designed to provide two types of benefits
results:
1. Total reductions in IQ decrement (and associated dollar values) for the entire
modeled population of prenatally exposed children; and
2. Distributional results in the form of per-capita reductions in IQ decrements for
each of the modeled children in the analysis population.
The first category of results (total IQ benefits) can be used to support a traditional cost-benefit
analysis comparing the total monetized benefits against total monetized costs. The second
category of results (distribution of per-capita IQ losses) can be used to examine the distributional
equity of IQ impacts across the study population of prenatally-exposed children (e.g., what is the
range of individual IQ changes across the study population and how many children are projected
to have IQ changes above specific levels of interest?).
In addition to generating benefits estimates for the children of recreational freshwater
fishers, the mercury benefits model also provides benefits estimates for several high-exposure
sub-populations including:
1. A high fish consumption rate study population that is defined as "subsistence"
fishers for the purposes of this study;
10-6
-------
2. A low-income high fish consumption study population (an alternative approach to
model subsistence fishers);
3. A Southeast Asian ethnic group with high freshwater fish consumption due to
cultural practices; and
4. A Native American population with high freshwater fish consumption due to
cultural practices.
These high-exposure scenarios are intended to provide coverage for special populations
potentially experiencing health impacts from the consumption of self-caught freshwater fish due
to increased consumption rates.
Because key factors in modeling freshwater fisher exposure (e.g., methylmercury fish
tissue concentrations, mercury deposition rates from power plants, the distribution of fishers and
fishing activity) can display significant spatial variability, the mercury benefits model has been
developed to provide adequate spatial resolution for a regulatory analysis of modeling fishing
activity and subsequent mercury exposure associated with CAMR. The mercury benefits model
has been developed to provide coverage for local- to regional-scale trends in fisher exposure
linked to more generalized spatial patterns of fishing activity, mercury fish tissue concentrations
and mercury deposition. The model also considers variability in the consumption rate of self-
caught freshwater fish by fishers, which is not necessary for a total (best estimate) prediction of
IQ impacts, but which is critical in modeling the distribution of per-capita IQ impacts across the
study population. By considering local- to regional-scale trends in patterns of recreational fisher
exposure as well as variability in fish consumption rates, this model provides a quantitative
assessment of the distribution and magnitude of exposures and IQ decrements across the fisher
study population (prenatally-exposed children).
In addition, because of the complexity in modeling fishing activity, two models of
freshwater fishing behavior have been developed for this analysis. One model (the "population
centroid" approach) represents a "push" model in that it focuses first on identifying where
recreational fishers live and then models their fishing behavior in the form of fishing trips out to
different distance rings (10, 20, 50 and 100 miles) from their home residences. This model is
applied at the US Census block group level, which results in exposure estimates being generated
for a relatively large number of polygons (165,000 block groups in the study area). The second
model (the "angler destination" approach), represents a "pull" model in that it focuses on
identifying where anglers fish and does not consider their residential location. This model is
applied at the watershed-level as identified by USGS 8-digit hydrologic unit code (HUC) and
assesses the distribution of recreational fishing activity across HUCs in the study area. Because
fishing activity (behavioral) data used in the HUC model does not include coverage for special
subpopulations evaluated in this analysis (e.g., Native Americans, Southeast Asian
subpopulations and economically disadvantaged subsistence subpopulations), these specialized
analyses were implemented using the population centroid approach. In addition, due to the
greater spatial precision of the population centroid model, relative to the HUC model, the
population centroid model was also used as the basis for generating distributional (per-capita IQ
impact) results for the fishers.
10-7
-------
The mercury benefits model generates health impact and valuation results by first
estimating the change in total mercury deposition over waterbodies within the 37 state study area
(mercury deposition is modeled using CMAQ).3 These changes in mercury deposition are
generated by comparing two air modeling scenarios (e.g., a control scenario versus a baseline
scenario for a particular simulation year). These changes in mercury deposition are then
translated into changes in methylmercury fish tissue concentrations based on the proportionality
assumption advanced in Mercury Maps (i.e., a incremental percent change in deposition
produces a matching percentage change in mercury fish tissue concentrations)4. Modeled
changes in methylmercury fish tissue concentrations can, in turn, be used together with the
fishing behavioral models described above, to predict changes in population-level mercury
exposure. These exposure changes can be translated through modeling into IQ reductions, which
can then be monetized using valuation functions based primarily on foregone (lost) earnings
resulting from reductions in IQ. Case studies of individual ecosystems (as presented in Section
3) show that the time necessary for aquatic systems to reach a new steady state after a reduction
in mercury deposition rates can be as short as 5 years or as long as 50 years or more. The
medium response scenarios also varied widely but were generally on the order of one to three
decades. Thus, benefits results generated for this analysis are reported using a range of lag
periods following regulatory implementation (e.g., 5, 10, 20, and 50 years). Based on the
response times from the case studies discussed above, we present a range of benefits based on
the 10 and 20 year lag as central estimates.5 We also provide results for the 5 and 50 years to
demonstrate how benefits would differ under potential shorter and longer lag periods. Modeling
of benefits for these different lags reflects the effects of economic discounting as well as
demographic growth in the exposed population.
10.1.3 Monetized Benefits: Results in Brief
The mercury benefits analysis generated two categories of results including total IQ
decrements and associated monetary (dollar) values for modeled populations and distributional
results in the form of per-capita IQ reduction estimates for the group of modeled individuals.
Total IQ decrement and valuation results were generated for the recreational fisher population as
well as for the four potentially high-risk sub-populations described above. Distributional results
(per-capita IQ decrements) were generated only for the recreational fisher population.
3 The 37 states included in the analysis are the following (plus the District of Columbia): Connecticut, Maine,
Massachusetts, New Hampshire, Rhode Island, Vermont, Delaware, Maryland, New Jersey, New York,
Pennsylvania, Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, Missouri, North Carolina,
South Carolina, Tennessee, Virginia, West Virginia, Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota,
Nebraska, North Dakota, Ohio, South Dakota, Wisconsin, Oklahoma, and Texas
4 There are several limitations with the Mercury Maps approach that are discussed fully in Section 3 of this report.
In particular, it applies only to waterbodies where air deposition is the primary source of mercury load on a system.
In Section 10.7.1.1, we estimate the number of areas (HUCs) that have non-air deposition to the waterbodies and
remove them from the analysis to determine the affect on total benefits.
5 A 30 year lag is also indicated by the case studies in Section 3, but are not provided in the benefit analysis in this
Section. EPA expects that results of the 30 year lag would not significantly differ from the 20 year lag presented in
this Section.
10-8
-------
The following conditions and emissions control scenarios were modeled for this RIA:
2001 Base Case
2001 Utility Emissions Zero-Out
2020 Base Case with CAIR
2020 Utility Emissions Zero-Out
2020 CAMR Control Option 1
2020 CAMR Control Option 2
Table 10-lc provides a summary of the total benefits estimated for each of these
emissions control scenarios.
10-9
-------
Table 10-l(c). Summary of Total Benefits Associated with Modelled Avoided IQ
Decrements in Prenatally Exposed Children Due to Reduced Mercury Exposure from
Freshwater Recreational Angling
Emission Control Scenario
2001 Zero Out of ECU Emissions
(Relative to 2001 Baseline Emissions)
-Using a 3% discount rate
-Using a 7% discount rate
2020 Base Case (with CAIR)
(Relative to 2001 Baseline Emissions)
-Using a 3% discount rate
-Using a 7% discount rate
2020 Zero Out of ECU Emissions
(Relative to 2020 Base Case with CAIR)
-Using a 3% discount rate
-Using a 7% discount rate
CAMR Option 1
(Relative to 2020 Base Case with CAIR)
-Using a 3% discount rate
-Using a 7% discount rate
CAMR Option 2
(Relative to 2020 Base Case with CAIR)
-Using a 3% discount rate
-Using a 7% discount rate
Combined Benefits of CAIR and
CAMR
(CAMR Option 1 + 2020 Base with
CAIR Relative to 2001 Base Case)
-Using a 3% discount rate
-Using a 7% discount rate
Range of Estimated Benefits Associated with Modelled Avoided IQ
Decrements Due to Mercury Exposure
(millions of 1999 dollars)*
Recreational
Freshwater
Angler6
$19.0 - $37.0
$ 8.9 -$20.2
$20.5 - $43.8
$ 9.6 -$30.0
$8.1 -$16.1
$3.8 -$11.0
$ 1.7 -$3.0
$ 0.8 - $ 2.0
$ 2.5 - $ 4.6
$1.2-$3.1
$22.2 - $46.8
$10.4 -$32.0
Subsistence
Anglers1
$ 4.9 - $ 6.7
$ 2.3 - $4.6
$ 4.9 - $7.0
$ 2.3 - $4.8
$ 2.2 - $2.7
$1.0 -$1.8
$ 0.5 - $ 0.6
$ 0.2 - $ 0.4
$ 0.7 - $ 0.9
$ 0.3 - $ 0.6
$ 5.5 - $7.6
$ 2.5 - $5.2
Native American
Case Study
Population"
$0.10 -$0.12
$0.05 - $0.08
$0.12-0.13
$0.6 - $0.9
approx. $0.10
$0.05 - $0.08
approx. $0.007
$0.005-$0.003
approx. $0.015
$0.007-$0.010
$0.011-$0.012
$0.050-$0.080
Asian American
Case Study
Population'
$0.047-$0.050
$0.021 -$0.034
$0.005-$0.007
approx. $0.003
$0.060-$0.064
$0.028-$0.043
approx. $0.003
$0.001 -$0.002
approx. $0.012
$0.006-$0.009
$0.008-$0.010
$0.023-$0.036
The value per IQ point used to calculcate total benefits presented in this table is S8800/IQ point and is based on a 3 percent discount
rate of net earnings over a lifetime. The value per IQ point at a 7 percent discount rate is $1580/ IQ point (1999$) according to
EPA(1997c).
The range of results presented for the Recreational Freshwater Angler of recreational anglers are based on potential outcomes from the
Angler Destination exposure modeling and the Population Centroid exposure modeling discussed in this chapter, as well as results
ranging in value from a 10 to 20 year lag. See Table 10-21 and Table 10-29 for the full matrix of potential results from the benefit
modeling of this population.
The range of results presented for Subsistence Anglers are based on potential outcomes from the Income-based modeling and the
Consumption-based modeling discussed in this chapter, as well as results ranging in value from a 10 to 20 year lag. See Table 10-33
for the full matrix of potential results from the benefit modeling of this population.
The range of results presented for the Native American case study of the Chippewa in Minnesota, Wisconsin, and Michigan are based
on potential outcomes from the Population Centroid exposure modeling discussed in this chapter, as well as results ranging in value
from a 10 to 20 year lag. See Table 10-42 for the full matris of potential results frm the benefit modeling of this population.
The range of results presented for the Asian American case study of the Hmong in Minnesota and Wisconsin are based on potential
outcomes from the Population Centroid exposure modeling discussed in this chapter, as well as results ranging in value from a 10 to
20 year lag. See Table 10-41.
10-10
-------
10.1.4 Key Steps
The process used for estimating mercury exposures is divided into three general steps:
1. Estimate mercury levels in freshwater fish across the eastern half of the United
States.
2. Estimate the size of the exposed populations of interest (i.e., prenatally exposed
children of freshwater anglers) and their spatial relation to mercury levels in
freshwater fish (i.e., location and methylmercury concentration in fish consumed
by mothers).
3. Estimate fish consumption and mercury ingestion rates for the mothers of
prenatally exposed children of freshwater anglers.
Based on these exposure estimates, it is then possible to estimate associated health
decrement and monetary losses. The process for quantifying these losses can be summarized as:
1. Estimate reductions in expected IQ levels for the exposed population
2. Estimate the expected value of foregone future earnings associated with the IQ
decrements.
To estimate the benefits of mercury emissions reductions, the preceding exposure assessment
and IQ valuation steps were conducted under two baseline (i.e., "base case") scenarios—one for
2001 and the other for 2020 (with CAIR)—and four emissions reduction scenarios. The benefits
associated with each of the emissions reduction scenarios, in particular the CAMR control
options, were then estimated as the difference (reduction) in the total value of IQ losses, going
from the relevant baseline scenario to conditions with the emissions reductions in place. These
steps and the results are described in detail in the following sections of this report.
Section 10.2 describes the data sources and methods used to estimate the spatial
distribution of mercury levels in freshwater fish across the eastern United States. It also
summarizes the estimates of mercury concentrations based on these methods.
Section 10.3 describes data sources and two discrete methods for estimating the number
of modelled prenatally exposed children and average levels of mercury in freshwater fish
consumed by their mothers. Results from applying the two methods are also reported and
compared.
Section 10.4 describes the data, modeling approach, and results for estimating levels of
mercury ingestion through consumption of noncommercial freshwater fish in the study area. The
modeling approach was applied with both of the methods described in Section 10.3 to estimate
distributions of mercury ingestion rates across the exposed population. This section also
describes methods for estimating IQ decrements and foregone future earnings resulting from the
estimated exposures.
10-11
-------
Section 10.5 summarizes results and provides quantitative estimates of the benefits
associated with the zero out scenario. Estimates of exposures, IQ decrements, and foregone
future earnings are reported and compared for both baseline conditions and for the zero out
scenario.
Section 10.6 examines the distributional implications of the regulation by describing how
the estimated changes in IQ effects are distributed across the exposed population.
Section 10.6 also adapts the methods described in Sections 10.3 and 10.4 to assess
mercury exposures, IQ effects, and benefits for potentially highly exposed subpopulations.
Three methods are explored for evaluating high mercury exposures, each of which focuses on
subpopulations with high expected rates of freshwater fish consumption. The first method
focuses on exposures among individuals in the top fifth percentile of freshwater fish
consumption rates. The second method implements an approach for examining exposures
among low-income individuals for whom self-caught freshwater fish may be an integral part of
their diet. The third method focuses on two specifically studied ethnic groups in the Great Lakes
area who have been found to consume relatively high rates of noncommercial freshwater
fish—the Hmong and the Chippewa Indians.
Section 10.7 discusses the main assumptions, limitations, and uncertainties associated
with the methods described in the report. It also describes the unqualified benefits associated
with reductions in mercury emissions.
10.2 Estimation of Mercury Levels in Freshwater Fish
To estimate mercury levels in consumed freshwater fish across the eastern half of the
United States, this analysis relied primarily on monitoring data (i.e., fish tissue samples drawn
from freshwater sites across the study area). A potential alternative to monitored data would be
to estimate mercury levels based on dynamic and localized fate and transport modeling;
however, models of this type present significant technical challenges when applied at
regional/national spatial scale.6
Data from the NLFA and NLFTS were used as inputs into the NDMMF7 model to
generate MeHg fish tissue concentrations normalized to the typical sizes of frequently targeted
fish species (largemouth bass, walleye, crappie, catfish, trout, and perch). All six species
concentrations were generated for each waterbody, then averaged to develop a representative
MeHg concentration for that waterbody at a given sample date.8 Where a single location was
6 An independent peer review of the benefits methodology indicated that ecosystem based fate and transport
modeling of bioaccumulation of MeHg in fish tissue for a national scale analysis would not be practical or even
feasible at this time.
7 A comprehensive evaluation of the performance of the NDMMF, and detailed description of the model are
available in Appendix E.
8 Details related to the selection of MeHg sample data, variability within the data, how the NDMMF was applied,
and overall concentrations is provided in ch. 5.
10-12
-------
sampled on multiple dates, the representative MeHg concentrations were averaged at that
location. The resulting estimates are summarized by state in Table 10-2.
After MeHg concentrations are computed for every sample location, the concentrations
are then linked to exposed populations using two exposure modeling methods. The angler
destination approach, described in section 10.1.2, evaluates exposure based on fishing pressure at
the HUC, thus, we estimate the average fish tissue concentration for lakes and rivers located in
each HUC in the study area. In the second approach, the population centroid approach discussed
in Section 10.1.2, concentrations are linked to populations based on trip travel distance rings.
More detail about these two approaches is provided later in this section.
The values reported in Table 10-2 are based on these estimated averages per sampling
location. Across all sampling locations the average (median) estimated mercury concentration in
fish tissue was 0.23 ppm (0.18 ppm) for lake sites and 0.25 ppm (0.19 ppm) for river sites.
Table 10-3 describes how the lake and river mercury concentration data (summarized in
Table 10-2) are distributed across the 1,362 HUCs in the study area. Over 60 percent of HUCs
are without lake estimates of mercury concentrations in fish, and almost 50 percent are without
river estimates. For the 512 HUCs with fish tissue concentration estimates for lakes, the average
HUC level concentrations range from 0.004 ppm to 1.5 ppm. Roughly 15 percent have average
mercury concentrations below 0.1 ppm.
The minimum and maximum mercury concentration in the lake fish are 0.004 ppm in
Illinois and 2.64 ppm in Pennsylvania. In the case of 707 HUCs with river estimates, the mean
mercury concentration in HUCs range from 0.004 ppm to 2.2 ppm. Roughly 25 percent HUCs
have average mercury concentrations below 0.1 ppm. The minimum and maximum mercury
concentration in the river fish are 0.0003 ppm in Louisiana and 3.3 ppm in Pennsylvania.
10-13
-------
era
n
r
o
rs
p
I
VI
O
35'
BT
H
Cfl
i
n
e
•^
5
"H.
5'
ora
CB
I
s.
5'
i^-
BP
-------
era
r
o
n
D
W>
O
«s
2
«
cr
t/3
65
3
CTQ
ff
n
a
5*
*+
sr
-------
Table 10-2. Summary Statistics for Estimated Fish Tissue Mercury Concentrations (ppm)
by State: 2001 Base Case8
Lake Sampling Sites
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
ND
NE
NH
NJ
NY
OH
OK
PA
SC
TN
TX
Number
56
50
20
3
91
65
18
31
15
3
9
52
6
7
92
179
309
8
7
84
16
17
56
10
66
100
68
19
39
44
41
Min
0.011
0.016
0.072
0.036
0.022
0.013
0.030
0.004
0.030
0.103
0.099
0.013
0.124
0.015
0.026
0.048
0.018
0.050
0.080
0.065
0.096
0.029
0.037
0.081
0.033
0.013
0.017
0.061
0.109
0.090
0.006
Mean
0.113
0.261
0.311
0.070
0.299
0.195
0.069
0.081
0.124
0.361
0.153
0.175
0.266
0.061
0.480
0.190
0.173
0.152
0.215
0.304
0.204
0.136
0.390
0.254
0.310
0.205
0.214
0.546
0.221
0.184
0.141
Max
0.575
0.943
1.511
0.102
1.105
0.674
0.168
0.358
0.311
0.649
0.268
1.489
0.394
0.177
1.020
0.624
1.022
0.350
0.299
1.210
0.377
0.391
1.425
0.753
0.947
1.037
1.164
2.640
0.497
0.505
0.363
Median
0.076
0.221
0.246
0.000
0.102
0.282
0.174
0.060
0.071
0.088
0.649
0.154
0.126
0.278
0.064
0.479
0.165
0.145
0.120
0.244
0.228
0.185
0.129
0.350
0.162
0.266
0.149
0.179
0.255
0.208
0.158
0.137
Number
104
174
38
5
80
98
258
45
41
109
9
51
44
1
35
52
67
216
7
18
341
4
127
76
64
118
504
66
60
162
21
57
River Sampling Sites
Min
0.012
0.009
0.030
0.070
0.002
0.027
0.002
0.030
0.005
0.030
0.023
0.039
0.000
0.578
0.007
0.134
0.066
0.014
0.044
0.082
0.028
0.168
0.001
0.059
0.068
0.037
0.010
0.016
0.020
0.123
0.100
0.004
Mean
0.157
0.340
0.215
0.109
0.102
0.374
0.282
0.131
0.097
0.186
0.150
0.170
0.238
0.578
0.046
0.496
0.181
0.164
0.162
0.387
0.249
0.325
0.135
0.348
0.337
0.320
0.252
0.223
0.403
0.493
0.194
0.182
Max
1.003
1.311
0.581
0.165
1.563
0.983
3.223
0.480
0.552
0.502
0.380
0.548
3.190
0.578
0.173
1.787
0.477
0.734
0.287
0.727
0.985
0.517
0.647
0.804
2.129
1.350
0.980
0.648
3.300
2.479
0.414
1.228
Median
0.094
0.295
0.194
0.090
0.072
0.353
0.156
0.110
0.076
0.173
0.140
0.159
0.152
0.578
0.030
0.453
0.172
0.143
0.171
0.343
0.203
0.307
0.111
0.298
0.207
0.251
0.240
0.179
0.170
0.375
0.170
0.147
(continued)
10-16
-------
Table 10-2. Summary Statistics for Estimated Fish Tissue Mercury Concentrations (ppm)
by State; 2001 Base Case (continued)
Lake Sampling Sites
State
VA
VT
WI
WV
AH
Number
18
31
187
6
1,823
Min
0.018
0.054
0.029
0.170
0.004
Mean
0.146
0.291
0.201
0.296
0.227
Max
0.470
0.641
0.900
0.400
2.640
Median
0.092
0.270
0.182
0.306
0.175
Number
38
22
204
45
3361
River Sampling Sites
Min
0.011
0.069
0.029
0.034
0.000
Mean
0.118
0.345
0.240
0.195
0.254
Max
1.900
0.945
0.897
0.848
3.300
Median
0.050
0.279
0.212
0.117
0.185
a For summary purposes, in this table the data are summarized at a state level; however, the data used in the analysis
were analyzed at a Census block group level or at a HUC level.
Table 10-3. HUC-Level Distribution of Mercury Sampling Sites and Estimated Fish Tissue
Concentrations: 2001 Base Casea
Lake
Number11
Percent
River
Number1"
Percent
Number of Samples in HUC
0 850
1-3 366
4-10 125
11-25 16
26+ 5
Average Concentration (in ppm) in
HUC
62.41
26.87
9.18
1.17
0.37
655
429
205
60
13
48.09
31.50
15.05
4.41
0.95
No data
0.00-0.05
0.06-0.10
0.11-0.19
0.20-0.29
0.30 and over
850
47
81
168
115
101
62.41
3.45
5.95
12.33
8.44
7.42
655
59
107
218
159
164
48.09
4.33
7.86
16.01
11.67
12.04
u.ju ana over mi i .*£ lot
"Based on average concentration across samples and species-specific estimates at each sampling station.
""Number of sampling stations.
10-17
-------
10.3 Estimation of Exposed Populations and Fishing Behaviors
Based on the spatial distribution of estimated mercury levels in fish, the next step in the
analysis is to estimate the average daily ingestion of mercury (g/day) through noncommercial
freshwater fish consumption (Hgl) for selected populations of interest. Because the primary
measurable health effect of concern — developmental neurological effects in children — occurs as
a result of in-utero exposures to mercury, the specific population of interest in this case is
prenatally exposed children. To identify and estimate the size of this exposed population, the
analysis focuses on women of childbearing age in freshwater angler households in the 37-state
study area. More specifically, it estimates the number of pregnant women in these households in
2001 and in other selected years.
Generally speaking, mercury ingestion for a population i in a given year can be
calculated as the product of three estimates, as shown in Eq. (10.1):
Ii = Ni*CHgi*Ci (Eq. 10.1)
where
N; = size of the exposed population of interest i (annual number of pregnant
women in freshwater angler households during the year),
CHgj = average concentration (u,g/g) of methyl mercury in noncommercial
freshwater fish filets consumed by population i, and
Q = average daily consumption rate (g/day) of noncommercial freshwater fish
by population i.
As described in more detail below, EPA applied two alternative approaches for defining
and estimating Nf and CHg( for the 37-state study area. Both approaches rely primarily on two
national-level data sources for characterizing freshwater angler populations and their fishing
behaviors. The discussion below begins by describing these two data sources, and then describes
how they are used in the two alternative approaches ( in combination with data from the Census
and other information sources) to estimate Ns and CHgj. Consumption rate estimates for
recreationally caught freshwater (C() are based primarily on recommendations in EPA's
Exposure Factors Handbook (EPA, 1997a) although several other sources were also considered
(see Section 10.1.3). These estimates are described in more detail in Section 10.1.4.
10.3.1 Primary Data Sources on Fishing Activity in the United States
Two main sources of national-level activity data are available and suitable for estimating
the size and spatial distribution of freshwater angler populations and activities:
• The National Survey of Fishing, Hunting, and Wildlife-Associated Recreation
(NSFHWR), maintained by the Department of the Interior (DOI) (DOI and DOC,
1992, 1997, 2002); and
The National Survey of Recreation and the Environment (NSRE 1994).
NSFHWR Angler Data. The NSFHWR, conducted by the U.S. Census Bureau about
every 5 years since 1955, includes data on the number and characteristics of participants as well
as time and money spent on hunting, fishing, and wildlife watching. The most recent version,
10-18
-------
the 20019 NSFHWR (DOI and DOC, 2002), collected data on 25,070 respondents in the
sportspersons (hunters and anglers) sample and 15,303 in the wildlife watchers sample, all over
the age of 15.
The freshwater fishing subset of the sportspersons survey is the most relevant to the
mercury analysis. Of the 25,070 respondents included in the 2001 sportspersons sample, 11,280
respondents participated in freshwater fishing. Using the weights provided in the NSFHWR
data, it is estimated that 28.4 million Americans participated in freshwater fishing in 2001,
spending approximately 467 million total days freshwater fishing.
The survey distinguishes between several types of freshwater fishing. Respondents were
asked the number of days spent fishing in the Great Lakes; in other ponds, lakes, or reservoirs;
and in rivers or streams. Table 10-4 summarizes the NSFHWR data on freshwater fishing by
state. It reports number of anglers and fishing days separately for lakes (including the Great
Lakes, ponds, reservoirs, etc.) and rivers (including rivers and streams), and it distinguishes
between states as the residence of the anglers and states as the destination for anglers (residents
and nonresidents).
Over 80 percent of freshwater anglers in the United States reside in our study area, and
they accounted for roughly 88 percent of lake-fishing days and 77 percent of river-fishing days
in the country in 2001. The states with the largest numbers of resident freshwater anglers in
2001 were Texas with 1.9 million anglers and Illinois and Minnesota, both with 1.3 million
anglers. Minnesota had the highest number of lake-fishing days (including both state residents
and nonresidents) in 2001 (25.1 million) followed by Texas and Wisconsin, with 22.6 million
and 18.2 million days, respectively. Pennsylvania, New York, and Florida had the largest
number of river-fishing days, each with over 6 million days in 2001.
The NSFHWR data also contain information on demographic characteristics of the
anglers, including gender, age, and marital status. For example, in the states in our study area,
roughly 17 percent of freshwater anglers were women of childbearing age. About 32 percent
were married male adults less than 45 years old. These two subpopulations are of interest for our
analysis because it is among these groups that we most expect to find either pregnant women
who consume their freshwater catch or men who share their catch with pregnant women.
9 The screening survey for the 2001 NSFHWR covered activities in 2000, but the follow-up and more in-depth
surveys focused on 2001.
10-19
-------
Table 10-4. Summary of Fishing Activity Levels by State in 2001 from NSFHWR
Fishing by State Residents
State
AL
AR
CT
DE
DC
FL
GA
IL
IN
IA
KS
KY
LA
ME
MD
MA
MI
MN
MS
MO
NE
NH
NJ
NY
NC
ND
OH
OK
PA
RI
SC
SD
TN
TX
Number of
Freshwater
Anglers"
572,604
546,099
236,880
47,496
13,921
1,153,514
953,119
1,302,368
729,603
511,674
420,418
609,959
552,769
180,475
327,679
335,587
970,174
1,294,333
409,808
976,151
264,223
133,489
345,726
864,957
710,251
137,839
1,246,262
679,571
1,008,107
46,446
511,862
144,382
763,484
1,882,755
Number of
Lake-Fishing
Days"*
6,237,724
9,805,445
2,413,792
318,592
0
12,036,042
10,286,143
16,091,265
12,043,601
4,891,453
5,405,143
9,774,419
5,961,407
2,318,947
1,640,229
4,496,619
15,373,711
23,908,920
6,832,990
9,840,152
2,726,954
1,824,024
4,729,331
11,839,200
9,470,233
2,291,206
16,197,914
11,384,571
8,856,695
509,566
6,951,092
1,807,212
11,159,872
24,014,324
Number of
River-Fishing
Days'
4,168,160
2,279,229
1,699,745
284,984
41,764
5,831,107
4,332,960
4,976,135
3,433,917
3,978,169
978,403
2,439,232
2,414,115
828,351
2,390,203
1,458,347
2,594,901
2,966,955
1,820,004
2,427,720
654,359
861,481
2,133,589
5,768,677
4,315,601
335,634
4,348,070
2,312,773
11,274,947
184,055
2,247,640
785,679
5,100,470
6,275,434
Fishing in Each State (by Residents and
Nonresidents)
Number of
Freshwater
Anglers
732,204
781,772
254,482
73,147
6,961
1,315,528
1,016,703
1,125,760
754,408
541,613
403,691
779,677
659,237
271,840
366,585
324,740
1,275,200
1,565,228
494,165
1,214,950
296,090
220,552
330,957
1,051,982
847,994
178,621
1,260,043
774,255
1,182,356
50,733
591,069
214,429
903,385
1,841,749
Number of
Lake-Fishing
Days'
6,838,700
10,576,329
2,320,467
315,203
0
12,332,411
9,761,303
11,866,587
10,985,097
4,183,232
4,910,572
10,288,090
6,117,510
2,774,182
1,461,502
4,028,656
16,562,948
25,139,287
7,441,168
10,848,890
2,586,388
2,004,541
4,290,623
12,560,010
9,163,218
1,936,036
15,098,754
11,193,434
9,124,678
429,931
6,997,418
2,200,846
11,118,719
22,573,278
Number of
River-Fishing
Days
4,078,525
2,716,241
1,404,884
329,185
6,961
6,386,844
3,525,712
4,105,517
2,942,734
3,539,145
795,508
2,553,575
2,523,430
1,005,643
2,912,591
1,098,457
2,933,216
2,859,056
1,538,864
2,697,573
656,657
1,036,094
1,476,570
6,417,990
4,346,851
316,964
3,979,234
2,225,586
10,664,975
222,018
2,249,877
1,044,079
5,375,402
5,419,444
(continued)
10-20
-------
Table 10-4. Summary of Fishing Activity Levels by State in 2001 from NSFHWR
(continued)
Fishing by State Residents
Fishing in Each State (by Residents and
Nonresidents)
State
VT
VA
WV
WI
All 50
States
plus
DC
Number of
Freshwater
Anglers"
99,564
652,561
260,343
948,912
28,438,814
Number of
Lake-Fishing
Days"'"
1,360,820
8,348,688
2,177,141
15,763,077
340,972,513
Number of
River-Fishing
Daysb
552,364
3,514,615
2,172,734
4,077,048
141,048,761
Number of
Freshwater
Anglers
171,420
721,301
317,632
1,349,553
28,438,814
Number of
Lake-Fishing
Days"
1,409,024
8,237,651
1,917,931
18,295,238
340,972,513
Number of
River-Fishing
Days
900,343
3,545,041
2,316,617
4,506,934
141,048,761
" Includes days fished in other states.
b Includes Great Lakes, lakes, ponds, and reservoirs.
Source: U.S. Department of the Interior (DOI), Fish and Wildlife Service and U.S. Department of Commerce,
Bureau of the Census. 2002. 2001 National Survey of Fishing, Hunting, and Wildlife-Associated
Recreation. Washington, DC: U.S. Government Printing Office.
NSRE Angler Data. The NSRE, formerly known as the National Recreation Survey
(NRS), is a nationally administered survey designed to assess outdoor recreation participation in
the United States and elicit information regarding people's opinions about their natural
environment. The NSRE sample of freshwater anglers is smaller than the NSFHWR sample, but
it is nonetheless a useful resource because it provides a wide variety of information about fishing
activities. Importantly, it includes relatively detailed information about the nature and location
of recent freshwater trips. Because the sampling procedure is designed to be representative,
inferences may be drawn as to the relative popularity of particular types of freshwater bodies
(e.g., lakes, rivers) among the general public and the average distance traveled to reach these
sites. Although most recently conducted in 2000, data from 1994 survey (NSRE 1994) are
currently best suited to support this analysis.
NSRE was conducted by the Survey Research Center at the University of Georgia.
Surveyors asked 16,000 individuals by telephone about their water-based recreation
activities—specifically boating, fishing, swimming, or viewing—during the previous year. The
survey elicited information from respondents about the most recent trip taken in each of the four
categories. Of particular interest for the mercury analysis is the information regarding fishing
trip destination for all respondents who fish.
In addition to information about the location of the last fishing trip destinations, the
NSRE contains information on the type of waterbody visited on the last trip. Waterbodies are
broken down into four categories: lakes, rivers and streams, wetlands, and coastal areas. Of the
3,257 respondents who had fished in the previous year, 1,698 indicated that their last trip was to
a lake, 694 to a river or stream, 5 to a wetland, and 720 to a coastal area. (Type of waterbody
visited was not available for 141 responses.)
10-21
-------
One of the main advantages of NSRE is that it includes geocoded data for reported
fishing destinations. To specify the location of the last fishing trip, respondents were asked to
provide the name of the waterbody, the nearest town to the waterbody, and an estimate of the
distance and direction from their home to the waterbody. Of the 2,520 freshwater destinations
reported in NSRE, HUC codes have been identified for 1,768 respondents/trips.
10.1.2.2 Estimation Approaches and Results for Exposed Populations and Fishing
Behaviors
To define and estimate exposed populations and their corresponding mercury exposures
through freshwater fishing in our study area, EPA has developed two distinct but related
approaches:
• The population centroid approach; and
• The angler destination approach.
Using two separate approaches, with respect to specific modeling assumptions, provides
useful insights into model uncertainty. These issues are discussed in more detail in Section 10.7.
The two approaches are similar in several respects. Most importantly, both approaches
define the main population of interest as prenatally exposed children in freshwater angler
households in the 37-state study area. They also both use estimates of mercury fish tissue
concentration estimates based on NLFA and NLFTS sampling data (as described in
Section 10.2). In addition, the two approaches use state-level data from the NSFHWR as an
essential input for estimating the size, characteristics, and behaviors of angler populations in the
study area.
However, there are also key differences in the way the two approaches use the mercury
sampling data, the NSFHWR, and other supporting data to estimate populations and exposures.
The population centroid approach focuses on the residential location of freshwater anglers, the
typical distances they travel to fish, and the distribution of mercury concentrations within these
travel distances. It uses Census information to define the location and demographic
characteristics of potentially exposed populations and uses data from the NSFHWR to estimate
the fraction of freshwater anglers with respect to each population and location. The second
approach—the angler destination approach—focuses on the fishing destination of anglers and
the distribution of mercury concentrations across these destinations. It defines destinations
according to standard watershed codes (eight-digit HUC) and uses information from both the
NSFHWR and NSRE to estimate levels of angler activity and fish consumption on a
watershed-by-watershed basis.
Below we describe each of the two approaches in more detail. We also summarize and
compare results from the two approaches. Key elements associated with each modeling
approach are presented in Table 10-5.
10-22
-------
Table 10-5. Overview of Key Attributes of the Population Centroid and Angler Destination
Models
Population Centroid Model
Angler Destination Model
Push" model based on first estimating the
number of recreational fisher within each US
Census block group and then predicting
fishing activity in the form of trip travel
distances out to different fishing trip travel
rings (10, 20, 50 and 100 mile rings).
Fish tissue samples are averaged within each
ring to provide exposure levels for fishers
assigned to a particular ring (averaged tissue
samples are also differentiated for rivers
versus lakes, and fishing activity within each
ring is separated into river- versus lake-
activity).
Fishing activity (i.e., trip travel distances) are
also differentiated according to income level
and urban/rural status of modeled recreational
fishers.
Model generates estimates of recreational
fisher exposure generated for 165,000 block
groups (significantly more spatial
differentiation compared with the angler
destination approach)
Total modeled population: 434,000
prenatally-exposed infants in 2001.
Model used to support the per-capita
distributional IQ impact analysis, due to its
greater spatial differentiation and use of more
detailed demographic data relative to the
angler destination model.
"Pull" model based on first determining the
relative attractiveness of individual
watersheds for recreational fishing activity
(this model does not consider the residential
location of fishers but instead, focuses on
modeling their fishing locations). Fishing
activity modeled at the 8-digit hydrologic unit
code (HUC) watershed level (these HUCs are
1,600 square miles on average, which is
significantly larger than the block groups
used in the population centroid model).
Predictive model for fishing activity in
individual HUCs consider range of factors
related to fishing activity (e.g., population
density in vicinity of HUC, HUC surface
area, total number of lake boundary and river
miles within the HUC).
Model generates separate recreational fisher
exposure estimates 1,360 HUCs (less than
l/10th of the number of block groups
modeled).
Total modeled population: 587,000
prenatally-exposed infants in 2001. Note, this
estimate is larger than the population
centroid's modeled population because the
angler destination model estimates exposures
for pregnant women as the combined effect of
(1) their own fishing activity and (2) adult
males bringing home and sharing fish they
catch.
10-23
-------
10.3,2 Population CentroidApproach.
This approach uses Census block groups to spatially separate and characterize potentially
affected populations in the study area. In the Census, block groups are generally defined to
contain between 600 and 3,000 people. A total of 503 block groups were identified in the study
area, and five distance intervals were mapped from the centroid of each block group. Then,
freshwater fishing trips and fish consumption are allocated to anglers in each group according to
data on typical travel distances for freshwater fishing in the United States. The flow diagram in
Figure 10-3 illustrates the main components of the approach, spatial scale of the data used to
estimate these components, and how these components are interrelated. For each selected block
group, the following steps were applied.
First, Census data were used to define the size, age, gender distribution, and income of
the population within each of the roughly 165,000 block groups in the study area.
Second, the size of the exposed population of interest (annual number of prenatally
exposed children in freshwater angler households) in each block group was estimated by
combining Census, Vital Statistics, and NSFHWR data. For each block group, this estimation
required:
1. Estimating the number of pregnant women (NP) living in the block group as
NP = NF*fs (Eq. 10.2)
where
NF = number of females aged 15 to 44 in (Census 2000) and
fs = state-level general fertility rate (average number of live births in an year per
1,000 women aged 15 to 44) (Hamilton, Sutton, and Ventura, 2003).
2. Estimating the annual number of prenatally exposed children in angler households
(NPA) as
NPA = NP*(NAS/NS) (Eq. 10.3)
where
NAS = state-level number of 15 years and older angler residents (NSFHWR) and
Ns =15 years and older adult population of states (Census).
Eq. (10.3) reflects an estimate of the number of pregnant women in each state who reside
in freshwater angler households (not necessarily the number who are freshwater anglers
themselves). To estimate this value (NPA), Eq. (10.3) implies that (1) the fraction of pregnant
women in a state who are in freshwater angler households is equal to the fraction of households
in the state that include freshwater anglers (i.e., pregnant women are no more or less likely than
the rest of the state population to live in households with freshwater anglers) and (2) the fraction
of households in the state that include freshwater anglers is equal to the fraction of adult
residents in the state who are freshwater anglers. The implications of these assumptions for the
results of the analysis are discussed in more detail in Section 10.7.
10-24
-------
o
to
a
TO
n
1
o
5'
OTQ
2
o
T3
5*
5'
P
n
P
3
T3
•a
I
rt
ET
Description
Number of females of
childbearing age
Fertility rate
Number of pregnant women
Number of angler residents (age > 15)
Population (age > 15)
Portion of population that are anglers
Number of prenatally exposed children
ally exposed children I
in angler households I
NPA in demographic group I,
distance j, and waterbody type k
Portion of freshwater fishing days by residents
to waterbody type k (k = lake (i) or river (r))
Portion of freshwater fishing trips by residents
in demographic group i (i = 1-4) to sites in
distance interval j (j = 0-10, >10-20, >25-50,
>50-100, or 100+ miles)
Portion of resident anglers in demographic
group i (i = 1-4)
Average Hg fish tissue concentration by
waterbody type (k) and distance (j)
Weighted average Hg concentration In
recreatJonally caught fish
Spatial Scale Data Source/ Computation
Block Group
State
Block Group
State
State
State
Block Group
Block Group
State
National
Block Group
Block Group
Block Group
Census
Vital Stats
Eq. (3.2)
NSFHWR
Census
Eq. (3.3)
Eq. (3.3)
Eq. (3.7)
NSFHWR
NSRE
Census
NLFWA/GIS
Eq.(3.6)
-------
To estimate NPA for years after 2001, it was assumed that state-level fertility rates (f.)
and angler participation rates (NAyNJ would remain constant; however, the number of women
of childbearing age in each block (NF) was increased based on county-level population growth
projections (Woods and Poole, 2001). In other words, for the period 2001-2025, the estimated
NPA for each block group was assumed to increase at the same rate as the projected annual
population growth rates for females 15-44 in their corresponding counties. Woods and Poole
(2001) provide population projections only for years up to 2025. For years beyond 2025, it was
assumed that NPA for each block group increase linearly at the same rate as the annual
population growth rate for NF in corresponding counties from 2024 to 2025. As discussed in
more detail in Section 10.1.4, estimates of exposed populations were developed for future years
to account for lagged effects between reductions in mercury emissions and reductions in mercury
concentrations in fish. Uncertainties resulting from these assumptions are discussed in more
detail in Section 10.7.
Third, average mercury concentrations in freshwater fish for angler households were
separately estimated for each block group. To estimate these averages, the available mercury
concentration data were separated according to (1) distance from each block group centre id
(separating them into five distance categories) and (2) waterbody type (lake or river). As shown
in Figure 10-4, a separate average mercury concentration was estimated for each waterbody type
and distance interval. These estimates were then applied to calculate a weighted average of these
mercury concentration estimates for each block group. To weight these estimates it was assumed
that the percentage offish consumed from each distance and waterbody category is equivalent to
the percentage of fishing trips to each distance and waterbody category. The procedure used to
estimate the distribution of trips across waterbody types and distance categories for each block
group is described below.
To approximate the percentage freshwater fishing trips from each block group to each
waterbody type (c, or cr), state-level averages were used. These averages were calculated for
each state, based on the portion of residents' freshwater angler days that are to each waterbody
type. In other words, these portions were calculated as
c, = Dls/(Dls + Dre) (Eq. 10.4)
cr = (l-c,) (Eq. 10.5)
where
D,s = number of lake-fishing days by state resident anglers (DOI, 2002) and
Drs = number of river-fishing days by state resident anglers (DOI, 2002).
These lake- and river-fishing day estimates from the NSFHWR are also summarized in
Table 10-4.
10-26
-------
100 Miles
Census Data:
Number of Females Aged 15-44
Percentage of Population with
Income >$50.000
Urban v. Rural Classification
NLFWA Data:
Average
River
Hg Cone.
50-100 miles
(CHgr4)
Average
River
Hg Cone.
20-50 miles
(CHg0)
Average
River
Hg Cone.
10-20 miles
Average
River
Hg Cone.
CMOmiles
(CHgrt)
Average
Lake
Hg Cone.
0-10miles
(CHg,,)
Average
Lake
Hg Cone.
10-20 miles
(CHgc)
Average
Lake
Hg Cone.
20-50 miles
(CHgH)
Average
Lake
Hg Cone.
50-100 miles
(CHgM)
Figure 10-4. Population Centroid Approach: Linking Census Block Groups to
Demographic Data and Mercury Fish Tissue Samples
Data from NSRE 1994 were used to approximate the percentage of freshwater fishing
trips to different distances from anglers' residential location. Five distance intervals were
defined as 0-10 miles, >10-20 miles, >20-50 miles, >50-100 miles, and 100+ miles. Based on
self-reported trip distance information from nearly 2,000 respondents (see Appendix E-l for
details), each of these distance categories was associated with roughly 20 percent of the reported
trips in the NSRE sample. Four distinct demographic groups were also found to have
significantly different average travel distances for freshwater fishing in the NSRE sample: urban
with above $50,000 income, rural with above $50,000 income, urban with below $50,000
income, and rural with below $50,000 income.10 The portion of trips for each demographic
group (i = 1 - 4) to each distance interval (j = 1 - 5) are defined as e^. The estimated values for
e;j are reported in Appendix E-3. Uncertainties associated with this trip travel distance analysis
are discussed in Section 10.7.
To estimate the portion of households in each demographic group (p; for i = 1 - 4), each
block group was categorized as either urban (including suburban) or rural. Census data on the
percentage of the population in each block group with household income less than $50,000 were
used to define the portion in the below $50,000 income groups.
10 An annual household income of $50,000 (in 2000 dollars) is close to the median value for both the NSRE sample
and the U.S. population.
10-27
-------
To estimate a weighted average mercury concentration for each block group (CHg), the
average mercury concentration in freshwater fish for each combination of waterbody (i.e, lake or
river) and distance interval (i.e., 10 = 2x5 combinations) was first estimated. This average
concentration is defined as CHgkj for k = lake or river and j = 1-5 distance interval." For each
block group, this approach assumes that the sample averages of CHg (from available NLFA and
NLFTS sampling data) for each distance-waterbody combination provide an reasonable estimate
of the true average mercury concentrations in the distance-waterbody combination.
The weighted average mercury concentration in freshwater fish for each block group was
then calculated as
(Eq. 10.6)
for
i = 1-4 demographic group,
j = 1-5 distance interval,
k = lake or river, and
Pi = percentage of block group households in demographic group i (Census).
To match exposed populations with corresponding average mercury concentrations in
fish, one method is to match the total NPA (annual number of prenatally exposed children in
angler households) from each block group with the corresponding weighted average mercury
concentration (CHg) for that block group. This method is equivalent to assuming that all
exposed individuals in the same block group and demographic group (i) are exposed to the same
average levels of mercury. This would occur, for example, if they all allocate trips to (or fish
consumption from) the different distance intervals and waterbody types according to the same
proportions, e^ and ck.
An alternative method for matching populations with mercury concentrations using the
population centroid approach is to assume that:
• Each exposed individual in a block group is associated with freshwater fishing in
a single distance interval and a single waterbody type (i.e., all the fish they
consume comes from the same distance and type of waterbody);
• The exposed populations in each block group (rather than just the fishing trips)
are distributed across the distance intervals and waterbody types according to the
estimated proportions e^ and c^.
In this case, as many as 20 separate exposed subpopulations can be defined for each block group:
p = NPA * Pi * efj * q (for all i, j, and k.) (Eq. 1 0.7)
11 For the 100+ miles category (j = 5) the mean mercury concentration for lake and river samples in the entire 37-
state study area (see Table 10-4) was used to define CHg5, and CHg5r respectively.
10-28
-------
Each subpopulation NPAijk can then be separately matched with the block group's average
mercury concentration for the corresponding distance and waterbody category (CHgjk).
Using the model described in Section 10.1.4 (below) to estimate mercury ingestion rates,
these two methods will produce the same estimates of total mercury ingestion for a block group
NPA. However, they will differ according to how individual ingestion rates are distributed
within the exposed population. The first method assumes constant ingestion rates across a block
group NPA, whereas the second method estimates separate ingestion rates for each
subpopulation (NPAjjk) in the block group.
Modeling separate exposure levels for each subpopulation (NPAijk) within each block
group is critical to characterizing the per-capita distribution of IQ changes within the study
population and does provide the basis for that assessment (along with consideration for
variability in fish consumption rates). However, providing this level of differentiation in
modeling individual exposure is not necessary for generating a mean (best estimate) IQ loss and
associated monetary value for the entire study population.
Summary of Results. This section summarizes some of the key results of applying the
steps outlined above. Table 10-6 reports state-level summaries of the Census block group data
used in the analysis. Nearly 165,000 block groups were identified in the 37-state study area.
The average number of women of childbearing age (16 to 44) per block group in 2000 was 281.
On average, there were approximately 60 percent of households with incomes above $50,000
and approximately 75 percent of block groups that were predominantly urban (or suburban).
Table 10-7 reports estimates of the annual number of prenatally exposed children in
selected years from 2001 to 2051 aggregated to the state level.12 For reasons discussed in more
detail below, the selected years correspond to lag periods of 5, 10, 20, and 50 years. We also
provide the estimate of prenatally exposed children in a base year of analysis (2001) for
comparison purposes. These estimates are derived from the Census demographic data reported
in Table 10-6 combined with state-level fertility rates, freshwater fishing participation rates, and
county-level population growth projections. The size of the exposed population of interest in
2001 was estimated to average over 11,000 individuals per state in the study area, with a total of
over 434,000 in the entire area.
Table 10-8 estimates the annual number of prenatally exposed children in selected years
and lag periods starting in 2020. We also provide the estimate of prenatally exposed children in
the base year (2020) for comparison purposes. In 2020, the size of the exposed population of
interest was estimated to be 482,000.
12 Although the results are reported at study area and the state levels, the analysis is conducted at the individual block
group level.
10-29
-------
Table 10-6. Block Group Demographic Characteristics by State (in 2000): Data Used in
Population Centroid Approach
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Study Area
Percentage
Number of Urban Block
Block Groups Groups*
3,329
2,135
2,620
433
502
9,112
4,788
2,634
9,843
4,798
2,299
3,157
3,509
5,053
3,678
1,143
8,450
4,082
4,540
2,148
5,271
630
1,591
874
6,510
15,079
9,354
2,901
10,387
821
2,859
688
4,014
14,463
4,749
530
4,388
1,588
164,950
55
54
88
99
83
87
69
58
85
69
69
53
73
92
84
38
74
68
67
51
58
45
69
59
94
85
78
66
78
92
61
45
64
81
70
36
69
46
75
Female Population, Aged
16—44, in Block Group
Min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
28
0
0
0
Mean
283
255
269
327
337
344
386
224
271
268
242
273
278
274
315
226
247
257
257
286
329
207
222
297
271
272
253
245
239
276
301
221
303
318
326
238
255
228
281
Max
2,591
1,752
3,718
2,243
2,899
6,355
4,239
2,618
3,960
3,559
2,773
3,207
2,962
2,678
4,212
960
3,452
1,882
2,004
2866
5,413
2,114
2,111
2,000
3,473
4,840
3,941
2,154
2,659
2,348
3,764
1,622
3,181
4,936
3,535
1,534
2,758
1,954
6,355
Percentage of Block Group
Households in Above
$50,000 Income Category
Min
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
15.05
0
0
0
Mean
71
73
46
57
53
64
63
64
55
62
63
70
70
50
48
65
56
54
67
73
65
70
62
52
45
55
62
71
63
58
67
70
68
63
55
62
58
75
60
Max
100
100
100
98
94
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
98.13
100
100
100
" Urban designation was assigned to a block group if 50 percent or greater of its population was categorized as
urban.
Source: 2000 Census.
10-30
-------
Table 10-7. Estimated Annual Number of Prenatally Exposed Children for Selected Lag Periods from 2001: Population
Centroid Approach
u>
Central Estimate of Fish Tissue Response Times
Base Year of Comparison
(2001)
Per Block
Group
Study
Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
Mean
2.63
3.0
4.6
1.6
0.5
1.7
2.3
4.2
3.2
2.8
2.8
3.6
3.3
3.1
1.1
1.6
2.1
2.0
5.6
3.7
S.D.
2.51
2.3
2.9
1.0
0.5
1.3
2.5
3.9
2.5
2.5
2.4
2.5
2.3
2.1
0.7
1.2
1.2
1.4
3.8
2.7
Total
Exposed
Population
434,059
9,948
9,832
4,134
236
836
21,218
20,255
8,448
27,670
13,640
8,246
10,322
10,709
5,663
5,881
2,378
17,028
22,806
16,804
10 Year Lag (20 11)
Per Block
Group
Mean
2.75
3.0
4.8
1.6
0.5
1.7
2.6
4.6
3.1
2.9
2.9
3.8
3.2
3.0
1.1
1.7
1.9
2.0
5.8
3.7
S.D.
2.80
2.4
3.2
1.0
0.4
1.3
2.9
4.7
2.6
2.7
2.6
2.8
2.3
2.1
0.7
1.3
1.1
1.5
4.3
2.8
Total
Exposed
Population
452,575
9,869
10,242
4,176
224
854
23,903
22,252
8,249
28,470
14,027
8,606
10,139
10,582
5,671
6,084
2,223
17,185
23,777
17,015
20 Year Lag (2021)
Per Block
Group
Mean
2.95
3.0
5.1
1.7
0.5
1.8
3.0
5.3
3.2
3.1
3.1
4.1
3.3
3.1
1.2
1.8
1.9
2.1
6.4
3.9
S.D.
3.22
2.5
3.7
1.1
0.4
1.3
3.4
5.8
2.8
2.9
2.8
3.2
2.5
2.3
0.8
1.5
1.1
1.6
5.2
3.2
Total
Exposed
Population
486,487
9,906
10,968
4,491
211
901
27,786
25,175
8,339
30,205
14,783
9,294
10,347
10,892
5,928
6,610
2,139
17,672
25,952
17,924
Alternative Estimate of Fish Tissue Response Times
5 Year Lag (2006)
Per Block
Group
Mean
2.69
3.0
4.7
1.6
0.5
1.7
2.5
4.4
3.2
2.8
2.9
3.7
3.2
3.0
1.1
1.6
2.0
2.0
5.7
3.7
S.D.
2.64
2.3
3.0
1.0
0.4
1.3
2.7
4.3
2.6
2.6
2.5
2.7
2.3
2.1
0.7
1.3
1.2
1.4
4.1
2.8
Total
Exposed
Population
442,938
9,907
10,033
4,152
228
850
22,532
21,306
8,326
27,963
13,799
8,410
10,211
10,580
5,686
6,017
2,324
17,186
23,286
16,906
50 Year Lag (2051)
Per Block
Group
Mean
3.84
3.4
6.8
2.4
0.7
2.2
4.6
7.8
3.4
3.8
3.8
5.4
3.8
3.3
1.6
2.4
1.9
2.5
8.1
4.9
S.D.
4.83
3.5
5.9
1.5
0.5
1.6
5.4
10.6
3.6
4.1
3.9
5.1
3.4
2.9
1.0
2.2
1.1
2.1
7.6
4.5
Total
Exposed
Population
632,017
11,230
14,557
6,229
286
1,109
41,941
37,558
9,054
37,881
18,037
12,289
12,041
11,677
7,846
8,742
2,124
20,816
32,953
22,289
(continued)
-------
Table 10-7. Estimated Annual Number of Prenatally Exposed Children for Selected Lag Periods from 2001: Population
Centroid Approach (continued)
Central Estimate of Fish Tissue Response Times
Base Year of Comparison
(2001)
Per Block
Group
State
MO
MS
NC
ND
NE
NH
5 NJ
& NY
OH
OK
PA
RI
SC
SD
TO
TX
VA
VT
Wl
WV
Mean
3.7
3.7
2.6
3.4
3.2
2.2
1.1
1.1
2.3
4.5
1.5
0.9
3.2
3.7
3.3
3.9
2.5
2.4
3.6
2.3
S.D.
2.7
2.3
2.1
3.1
1.9
1.6
0.7
0.9
1.7
3.0
0.9
0.6
2.6
3.1
2.6
3.4
2.0
1.5
2.4
1.3
Total
Exposed
Population
16,804
7,909
13,483
2,101
5,138
1,966
6,838
16,877
21,457
13,117
15,146
755
9,010
2,508
13,317
55,802
11,793
1,288
15,848
3,651
10 Year Lag (2011)
Per Block
Group
Mean
3.7
3.7
2.8
3.2
3.4
2.3
1.1
1.1
2.3
4.6
1.4
1.0
3.3
3.7
3.4
4.5
2.6
2.3
3.6
2.2
S.D.
2.8
2.5
2.5
3.0
2.1
1.6
0.8
1.0
1.7
3.1
1.0
0.6
2.7
3.2
2.8
4.1
2.3
1.5
2.5
1.3
Total
Exposed
Population
17,015
7,920
14,702
2,016
5,393
1,991
7,024
16,994
21,317
13,298
14,756
785
9,327
2,502
13,663
64,381
12,440
1,223
15,839
3,460
20 Year Lag (2021)
Per Block
Group
Mean
3.9
3.7
3.1
3.2
3.7
2.3
1.2
1.2
2.3
4.9
1.4
1.0
3.4
3.9
3.6
5.2
2.8
2.2
3.7
2.0
S.D.
3.2
2.7
3.0
3.1
2.4
1.7
0.9
1.0
1.8
3.4
1.0
0.6
3.0
3.5
3.2
4.9
2.6
1.5
2.6
1.3
Total
Exposed
Population
17,924
8,008
16,534
2,004
5,849
2,045
7,636
17,663
21,659
14,281
14,702
815
9,716
2,590
14,511
74,645
13,471
1,161
16,427
3,248
Alternative Estimate of Fish Tissue Response Times
5 Year Lag (2006)
Per Block
Group
Mean
3.7
3.7
2.7
3.3
3.3
2.3
1.1
1.1
2.3
4.5
1.4
0.9
3.2
3.7
3.4
4.2
2.6
2.4
3.6
2.2
S.D.
2.8
2.4
2.3
3.0
2.0
1.6
0.8
1.0
1.7
3.1
1.0
0.6
2.7
3.1
2.7
3.8
2.1
1.5
2.5
1.3
Total
Exposed
Population
16,906
7,886
14,066
2,050
5,256
2,004
6,927
16,937
21,356
13,126
14,965
773
9,222
2,500
13,510
59,828
12,158
1,270
15,884
3,518
SO Year Lag (2051)
Per Block
Group
Mean
4.9
4.3
4.8
3.4
4.8
3.0
1.6
1.4
2.7
6.1
1.6
1.3
4.2
4.4
4.8
7.6
3.8
2.3
4.3
2.0
S.D.
4.5
3.8
4.9
3.2
3.2
2.3
1.2
1.3
2.4
4.5
1.4
0.9
3.9
4.3
4.8
7.6
4.1
1.7
3.3
1.6
Total
Exposed
Population
22,289
9,208
25,075
2,138
7,635
2,625
10,130
21,400
25,190
17,713
16,862
1,066
12,031
2,941
19,133
108,932
17,946
1,242
18,932
3,160
-------
Table 10-8. Estimated Annual Number of Prenatally Exposed Children for Selected Lag Periods from 2020: Population
Centroid Approach
Central Estimate of Fish Tissue Response Times
Base Year of Comparison
(2020)
Per Block
Group
Study
Area
State
AL
AR
CT
DC
_- DE
o
w FL
W GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
Mean
2.92
3.0
5.1
1.7
0.5
1.8
3.0
5.2
3.2
3.0
3.1
4.1
3.3
3.1
1.2
1.8
1.9
2.1
6.3
3.9
S.D.
3.17
2.5
3.6
1.1
0.4
1.3
3.3
5.7
2.8
2.9
2.8
3.2
2.5
2.3
0.7
1.5
1.1
1.6
5.1
3.1
Total
Exposed
Population
481,987
9,871
10,877
4,437
211
894
27,339
24,796
8,325
29,953
14,665
9,222
10,303
10,857
5,867
6,533
2,137
17,539
25,696
17,805
10 Year Lag (2030)
Per Block
Group
Mean
3.21
3.1
5.6
1.9
0.5
1.9
3.5
6.0
3.2
3.3
3.3
4.4
3.4
3.2
1.3
2.0
1.9
2.2
6.9
4.2
S.D.
3.67
2.8
4.3
1.2
0.4
1.4
4.0
7.2
3.0
3.3
3.2
3.7
2.7
2.5
0.8
1.7
1.1
1.7
5.9
3.5
Total
Exposed
Population
528,721
10,275
11,978
5,005
227
966
31,950
28,805
8,551
32,470
15,769
10,073
10,808
11,120
6,499
7,253
2,133
18,610
28,010
19,123
20 Year Lag (2040)
Per Block
Group
Mean
3.51
3.2
6.2
2.1
0.6
2.1
4.0
6.9
3.3
3.6
3.5
4.9
3.6
3.2
1.4
2.2
1.9
2.3
7.4
4.5
S.D.
4.22
3.1
5.1
1.4
0.5
1.5
4.6
8.8
3.3
3.7
3.5
4.4
3.0
2.7
0.9
2.0
1.1
1.9
6.7
4.0
Total
Exposed
Population
577,910
10,730
13,206
5,588
255
1,034
36,708
32,973
8,791
35,047
16,849
11,128
11,395
11,385
7,140
7,962
2,129
19,660
30,364
20,630
Alternative Estimate of Fish Tissue Response Times
5 Year Lag (2025)
Per Block
Group
Mean
3.06
3.0
5.3
1.8
0.5
1.9
3.2
5.6
3.2
3.2
3.2
4.2
3.3
3.1
1.2
1.9
1.9
2.1
6.6
4.0
S.D.
3.41
2.6
4.0
1.2
0.4
1.4
3.7
6.4
2.9
3.1
3.0
3.4
2.6
2.4
0.8
1.6
1.1
1.6
5.5
3.3
Total
Exposed
Population
504,127
10,048
11,364
4,714
213
932
29,572
26,721
8,431
31,181
15,229
9,545
10,515
10,987
6,178
6,898
2,136
18,085
26,833
18,369
50 Year Lag (2070)
Per Block
Group
Mean
4.40
3.6
7.9
2.8
0.8
2.5
5.6
9.5
3.6
4.3
4.2
6.3
4.2
3.5
1.8
2.7
1.9
2.7
9.2
5.5
S.D.
5.93
4.2
7.5
1.8
0.7
1.8
6.7
13.7
4.1
4.8
4.7
6.5
4.0
3.3
1.2
2.7
1.1
2.4
9.2
5.4
Total
Exposed
Population
725,474
12,093
16,891
7,335
339
1,239
50,980
45,477
9,510
42,778
20,089
14,294
13,156
12,182
9,064
10,089
2,115
22,811
37,425
25,154
(continued)
-------
Table 10-8. Estimated Annual Number of Prenatally Exposed Children for Selected Lag Periods from 2020: Population
Centroid Approach (continued)
Central Estimate of Fish Tissue Response Times
Base Year of Comparison
(2020)
Per Block
Group
State
MS
NC
ND
NE
NH
NJ
5 NY
£ OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
Wl
WV
Mean
3.7
3.1
3.2
3.7
2.3
1.2
1.2
2.3
4.9
1.4
1.0
3.4
3.9
3.6
5.1
2.8
2.2
3.7
2.1
S.D.
2.7
2.9
3.1
2.4
1.7
0.9
1.0
1.8
3.4
1.0
0.6
2.9
3.5
3.2
4.8
2.6
1.5
2.6
1.3
Total
Exposed
Population
7,974
16,264
2,005
5,829
2,023
7,554
17,537
21,660
14,132
14,621
807
9,634
2,587
14,372
73,600
13,318
1,156
16,324
3,261
10 Year Lag (2030)
Per Block
Group
Mean
3.9
3.6
3.2
4.0
2.5
1.3
1.2
2.4
5.2
1.5
1.1
3.6
4.0
3.9
5.9
3.1
2.2
3.9
2.0
S.D.
3.0
3.5
3.1
2.6
1.8
1.0
1.1
2.0
3.7
1.1
0.7
3.2
3.7
3.7
5.7
3.1
1.5
2.8
1.4
Total
Exposed
Population
8,341
19,020
2,022
6,347
2,211
8,378
18,769
22,715
15,182
15,340
889
10,399
2,669
15,848
84,608
14,795
1,185
17,182
3,194
20 Year Lag (2040)
Per Block
Group
Mean
4.1
4.2
3.3
4.4
2.8
1.4
1.3
2.6
5.6
1.5
1.2
3.9
4.2
4.3
6.7
3.4
2.3
4.1
2.0
S.D.
3.4
4.2
3.1
2.9
2.0
1.1
1.2
2.2
4.1
1.3
0.8
3.5
4.0
4.2
6.6
3.6
1.6
3.0
1.5
Total
Exposed
Population
8,754
21,903
2,077
6,960
2,408
9,212
20,022
23,893
16,387
16,065
974
11,176
2,799
17,412
96,191
16,295
1,212
18,015
3,178
Alternative Estimate of Fish Tissue Response Times
5 Year Lag (2025)
Per Block
Group
Mean
3.8
3.3
3.2
3.8
2.4
1.2
1.2
2.4
5.0
1.4
1.0
3.5
3.9
3.8
5.5
3.0
2.2
3.8
2.0
S.D.
2.9
3.2
3.1
2.5
1.7
0.9
1.1
1.9
3.5
1.1
0.7
3.1
3.6
3.4
5.2
2.8
1.5
2.7
1.3
Total
Exposed
Population
8,135
17,578
1,994
6,041
2,112
7,961
18,143
22,126
14,579
14,978
847
10,010
2,605
15,065
78,817
14,044
1,172
16,766
3,202
50 Year Lag (2070)
Per Block
Group
Mean
4.7
5.8
3.6
5.5
3.4
1.8
1.6
2.9
6.9
1.8
1.5
4.7
4.8
5.5
9.1
4.4
2.4
4.7
2.0
S.D.
4.6
6.2
3.5
3.9
2.6
1.5
1.5
2.7
5.2
1.7
1.0
4.5
4.9
5.9
9.3
5.1
1.9
3.7
1.8
Total
Exposed
Population
9,992
30,553
2,244
8,800
3,000
11,714
23,779
27,429
20,003
18,239
1,227
13,508
3,187
22,105
130,939
20,798
1,293
20,515
3,129
-------
Table 10-9 summarizes the mercury concentration data for freshwater fish in each of the
distance and waterbody type categories.13 Some of the key results from this table are the
following:
A relatively small percentage of block groups have lake samples within 0-10
miles (26 percent) and > 10-20 miles (49 percent);
• A larger percentage of block groups have river samples within 0-10 miles (52
percent) and > 10-20 miles (69 percent);
• Almost 94 percent of block groups have at least one lake sample within 100
miles;
• Over 99 percent of block groups have at least one river sample within 100 miles;
and
• For all samples within 100 miles, the mean (median) concentration in lake
samples is 0.21 ppm (0.18 ppm) and in river samples is 0.23 ppm (0.2 ppm).
Table 10-9. Average Estimated Mercury Concentrations (ppm) in Freshwater Fish by
Distance Interval from Block Group Centroids; Base Case 2001
Distance from Centroid
Lake Sampling Sites
0-10 miles
> 10-20 miles
>20-50 miles
>50-100 miles
0-100 miles
River Sampling Sites
0-10 miles
> 10-20 miles
>20-50 miles
>50-100 miles
0-1 00 miles
Na
44,327
80,524
147,696
161,837
163,402
86,269
114,505
157,663
163,311
164,084
Min
0.0036
0.0036
0.0063
0.0063
0.0036
0.0004
0.0004
0.0004
0.0335
0.0004
Mean
0.209
0.204
0.201
0.229
0.213
0.217
0.221
0.224
0.232
0.225
Max
2.640
2.640
2.225
1.054
2.640
3.300
3.300
3.190
1.022
3.300
Median
0.150
0.164
0.177
0.193
0.180
0.165
0.184
0.203
0.225
0.202
"Number of block groups (out of 164,950) with at least one sample in the distance interval.
13 Note that each of the sampling site mercury concentration estimates reported in Table 10-4 is included multiple
times in the estimates reported in Table 10-9, because they are all located within 100 miles of multiple block groups.
10-35
-------
10. 3.3 Angler Destination Approach
Rather than focusing on the point of origin (i.e., residential location) of freshwater
anglers, as is done in the population centroid approach, the angler destination approach focuses
on recreational fishing behavior and determines the areas that are more likely to be fished, using
information on inland watersheds (defined by HUCs).14 The advantage of the NSFHWR, as
described above, is that it is based on a relatively large number of observations and, therefore,
can provide reliable estimates of angler activities and demographics on a state level. However, it
does not provide information at a finer level of spatial resolution than the state. In contrast, the
NSRE provides a much smaller sample on angler activities and demographics, but it also
provides more detailed information on the destination (i.e., HUC) of fishing trips.
The flow diagram in Figure 10-5 illustrates the main components of this approach and
how they are interrelated including the spatial scale of the data used for various model
components. To implement the angler destination approach, the following steps were applied.
First, for each state, data from the NSFHWR were used to provide the total annual
number of lake- and river-fishing days in the state (Dls and Dra) by both residents and
nonresidents in 2001 (see Table 10-4).
Second, data from the NSRE were used to estimate the portion of these state- level fishing
days that occur in each HUC within the state. The details of this analysis of the NSRE data are
provided in Appendix E-2. To summarize, the following process was applied:
1 . Counted the number of lake and river trips in NSRE 1 994 to each HUC in the United
States (Tlh, T,,,), and used these counts as indicators of the "level of use" for each
HUC.
2. Used a negative binomial regression model to estimate the HUC-level determinants
of the level-of-use counts (determinants include the number of lake or river miles in
the HUC, population within 50 miles of the HUC, size of the HUC, etc). In a
simplified form, the regression model assumes that the probability of observing the
value th for Tlh (or T.J can be expressed as:
)'* for 5 = j
.
lh •
14 A site-choice random utility model (RUM) approach was also considered for modeling fishing behavior, but the
data and analytical requirements of this alternative approach were determined to be beyond the scope of this
analysis.
10-36
-------
TO
C
o
U)
1
ere
s
>>
a
BS
»•*•
5'
D
•a
•o
o
Description
Level of Use Indicators:
Number of reported last river trips to HUC
Number of reported last lake trips to HUC
Vector of HUC characteristics
Estimated Negative Binomial Count Model
MODEL PREDICTED Level of Use Indicators:
Prediced level of use indicator for rivers in HUC
Prediced level of use indicator for lakes in HUC
MODEL PREDICTED RELATIVE Level of Use Indicators:
Predicted portion of state-level river trips to the HUC
Predicted portion of state-level lake trips to the HUC
Annual river fishing days (by residents and nonresidents combined)
Annual lake fishing days (by residents and nonresidents combined)
Predicted annual river-fishing days in HUC
Predicted annual lake-fishing days in HUC
Average annual number of fishing days per angler
Angler-year equivalents
Portion of anglers that are female of childbearing age
Portion of anglers that are male, married, and age 18-44
Fertility rate
Angler-year equivalents for households with pregnant women
Average Hg fish tissue concentration in rivers
Average Hg fish tissue concentration in rivers
Weighted average Hg concentration In recreatlonally-caught fish
Spatial Scale Data Source/ Computation
HUC NSRE1994
HUC
HUC
HUC
HUC/State
USGS. Census, GIS
Eg. (3.8)
Eg. (3.8)
State
State
HUC
HUC
US
HUC
State
State
State
HUC
HUC
HUC
HUC
Eq.
(3.10)
Eq. (3.9)
NSFHWR
NSFHWR
Eq. (3.12)
Eq.(3.11)
NSFHWR
Eq. (3.13)
NSFHWR
NSFHWR
Vital Stats
Eq.(3.14)
NLFWA/GIS
NLFWA/GIS
Eg. (3.15)
-------
where
Xh represents a vector of HUC-level characteristics and pf represents a corresponding
vector of coefficients. These coefficient vectors were estimated using maximum
likelihood methods and are represented as p,and pr.
3. Used the results of the estimated econometric model to predict a level-of-use
indicator for each HUC (frh ,TA), based on observed HUC-level characteristics.
4. For each state, approximate the percentage of state-level lake- and river-fishing
days that occur in each HUC as:
p,h = (Eq-10-9)
Prh =
f (Eq. 10.10)
rh
Third, fishing days in each HUC (h) were estimated for each waterbody type (A,h and A.J
by combining the state-level fishing day totals from the NSFHWR (Step 1) with the HUC-level
portions estimates (Step 2):
Alh=plhs*Dls (Eq. 10.11)
(Eq. 10.12)
Fourth, "angler-year equivalents" were estimated for each HUC (NJ, using the
nationwide estimate from the NSFWHR of the average number of fishing days per year per
angler (d =16.42 days/yr). In other words, each 16.42 estimated angler days in HUC h in 2001
were treated as the equivalent of one angler-year to the HUC in 2001.
NAh = (Alh + AJ/d (Eq. 10.13)
Fifth, the number of these angler-year equivalents that are specifically for pregnant
women were approximated as
NPAh = NAh * (nfs + nms) * fs (Eq. 1 0. 1 4)
where
nf. = percentage of anglers fishing in state s that are female and of childbearing age
(NSFHWR) and
nms = percentage of anglers fishing in state s that are male and married and between
the ages of 18 and 44 (NSFHWR).
fs = state-level fertility rate (see Eq.[10.2])
The base year for the NPAh estimates is 2001, which is the most recent year for
NSFHWR data. To estimate NPAh for subsequent years, it was assumed that this exposed
10-38
-------
population associated with each HUC would grow at the same rate as the projected annual
population growth for females 15-44 in the entire 37-state study area.15 Population growth rates
based on Woods and Poole (2001) population projections for females 15-44 in the study area
were used to estimate NPAh up to 2025. For years beyond 2025, it was assumed that NPAh in
the study area increase linearly at the same rate as the annual population growth rate for females
15-44 from 2024 to 2025.
Sixth, the georeferenced mercury concentration estimates (summarized in Tables 10-2
and 10-3) were used to estimate average mercury concentrations in each HUC for each
waterbody type (CHglh, CHg^). This approach assumes that the sample averages of CHg (from
available NLFA and NLFTS sampling data) for each waterbody type in each HUC provide an
unbiased estimate of the true average mercury concentrations in the corresponding waterbody
type and HUC. The weighted (by number of fishing days in each waterbody type) average
mercury concentration in freshwater fish in each HUC was estimated as follows:
(A* CHg,. ) + (A. * CHg. )
CHg, = ^^ *»> y* ^^. (Eq. 10.15)
•"•Ih + Arh
As in the case of the population centroid approach, an alternative method is to apportion
NPh to different water bodies (j = river or lake). Each subpopulation (NPjh) can then be matched
with the average mercury concentration for the corresponding waterbody j (CHgjh).
Summary of Results. This section summarizes some of the key results from applying
the angler destination approach outlined above. Figures 10-6 and 10-7 display the estimated
distributions of lake- and river-fishing days across HUCs (Alh and A^) in 2001. The HUCs with
the highest estimated density of lake fishing (i.e., 200 or more annual lake-fishing days per
square mile) are widely distributed across the study area. As expected, many are located in areas
with extensive lake shoreline miles such as the Great Lakes region and in Minnesota, and several
are also located in Tennessee, Kentucky, and Florida. The HUCs with the highest density of
river fishing (i.e., 200 or more annual river fishing days per square mile) are most heavily
concentrated in Pennsylvania, New York, Connecticut, and Massachusetts, as well as in North
Carolina, Tennessee and Florida.
Table 10-10 reports the estimated annual number of prenatally exposed children across
states and HUCs in 2001. It is important to note that, in contrast to the estimates reported in
Tables 10-7 and 10-8 for the population centroid approach, the states in this table refer to where
fishing took place and where the mercury exposure originated from rather than state of
residence. Using the angler destination approach, the size of the exposed population of interest
15 Woods and Poole county level population projections were not used to predict future exposed populations in the
angler destination approach because the exposed population are based on fishing destination rather than residential
location.
10-39
-------
f
Annual Lake Fishing
Days Per Sq. Mi.
20-50
50 - 100
100-200
200 or more
-------
I
Annual River Fishing
Days Per Sq. Mi.
-------
Table 10-10. State-Level Summary of Exposed Population Estimates: Angler Destination
Approach
Estimated Annual Number of Prenatally Exposed Children in 2001
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
Average per HUC
251
253
397
4
59
421
406
208
636
603
133
330
212
340
333
135
344
448
299
230
Total
13,077
14,669
4,762
8
592
22,748
21,530
12,069
33,696
23,517
11,929
15,840
12,491
6,808
7,318
2,977
20,998
36,700
19,720
12,417
State
NC
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Study Area
Average per HUC
335
54
77
165
809
412
679
350
433
167
329
65
340
309
357
102
470
139
313
Total
19,451
2,684
5,477
2,805
10,520
21,401
30,556
23,791
24,704
835
11,837
3,798
20,723
64,537
18,224
1,726
24,429
5,149
586,516
(i.e., prenatally exposed children) was estimated to average more than 15,000 per state in the
study area, with a total of almost 590,000 in the entire area. The state with the highest estimated
number of prenatally exposed children is Texas, with almost 65,000, followed by Minnesota and
Ohio, both with over 30,000. This total exposed population estimate is roughly 35 percent
greater than the estimate based on the population centroid approach. For reasons discussed in
more detail in Section 10.7, the assumptions underlying the angler destination approach are
likely to overestimate the exposed population of interest, whereas those underlying the
population centroid approach are likely to underestimate the exposed population.
10.4 Estimation of Mercury Exposures, IQ Decrements, and Lost Future Earnings
This section describes the methods and results for estimating mercury exposures through
consumption of noncommercial freshwater fish. It describes how the estimates derived and
summarized in the previous section—in particular, distributions of (1) the annual number of
prenatally exposed children and (2) the average mercury concentrations in freshwater fish
consumed by their mothers—were combined to calculate mercury ingestion levels from
recreationally caught freshwater fish in this population. In addition it describes methods for
translating these exposure estimates into corresponding estimates of (1) expected reductions in
IQ levels achieved by the prenatally exposed children and (2) expected reductions in the value of
future earnings resulting from the IQ decrements.
10-42
-------
10.4.1 Modeling Approach for Estimating Individual Exposures
Both the population centroid approach and the angler destination approach described
above define distinct subpopulations of women of childbearing age in angler households.
Women in each subpopulation i are estimated to be exposed to the same average levels of
mercury in recreationally caught freshwater fish. In the population centroid approach these
subpopulations are grouped by Census block group (and still further by income and urban-rural
area categories), whereas in the angler destination approach they are grouped by HUC.
To estimate average daily mercury ingestion rates for each subpopulation i, a similar
method was applied using results from both the population centroid and the angler destination
approach. In both cases, the ingestion rates were calculated based on the following equation:
li = CHgFC; * C = (CHgFUj * CCF) * C (10.16)
where
Hgl = average daily mercury ingestion rate (fj.g/day);
CHgFU = average mercury concentration in uncooked freshwater fish (jig/g
= ppm);
CCF = cooking conversion factor: ratio of mercury concentration in
cooked fish to mercury concentration in uncooked fish (= 1.5);
CHgFC = average mercury concentration in cooked freshwater fish (ppm);
and
C = average daily self-caught freshwater cooked fish consumption rate
(g/day) = 8 g/day.
To determine an appropriate daily fish consumption rate (C) for the analysis, EPA
conducted an extensive review of existing literature characterizing self-caught freshwater fish
consumption including: (a) EPA's 1997 Exposure Factors Handbook (EPA, 1997b), (b) EPA
Office of Water's documentation presenting methodologies for deriving Ambient Water Quality
Criteria (USEPA, 2000c), (c) studies recommended by peer reviewers of the mercury benefits
model, and (d) studies identified in the open literature, in order to insure that the best available
data were used in modeling ingestion for the recreational fisher population. In conducting this
literature review, the modeling team applied several key criteria in assessing the applicability of
consumption data presented in studies:
1 . Self-caught freshwater consumption data for regions relevant to the study area:
The ingestion rate data had to apply to the consumption of freshwater fish caught
by fishers at locations representative of the 37 state study area. Many of the
studies that were reviewed included saltwater fish species, or some component of
commercially caught (i.e., store-purchased) fish, either of which resulted in
exclusion of the study from consideration. In addition, some of the studies,
provided estimates for populations located in areas of the United States outside of
the 37 state study area, where fishing behavior could differ systematically (in
terms of types offish harvested and seasonally) from the eastern half of the
country.
10-43
-------
2. Consumption rates (including meal event frequency data) that supported
generation of an annual-averaged daily consumption rate for fish: The
concentration-response function for IQ decrements is based on maternal hair
mercury concentrations, which, in turn, is based on the estimated annual-averaged
daily exposure level for methylmercury in ug/kg bodyweight/day. This exposure
level is itself based on an annual-averaged daily consumption rate for fish (for the
study population, or modeled individuals within that population) combined with
the methylmercury concentration in ingested fish. Because the analysis required
annual-averaged daily fish consumption estimates, the underlying consumption
data had to include information on meal event frequency over longer periods.
This is especially true for the high-end percentile consumption estimates required
to generate the lognormal distribution of recreational fisher consumption rates.
Many of the fish consumption studies that were reviewed, provided data for fish
consumption over a relatively short survey period (days) and did not include
information on the long-term frequency of self-caught fish meals. Only those
studies providing long-term frequency data were ultimately selected for this
analysis.
3. Coverage for the entire recreational fisher population including consumers and
non-consumers (catch and release): Modeling of recreational fisher exposure can
be conducted either focusing on consumers only (i.e., excluding catch-and-release
fishers), or by modeling "all fishers" including those who consume their catch and
those who engage in catch-and-release activity. Because we are not able to
identify in the data the portion of recreational anglers that are consumers only, we
have selected a consumption rate that applies to "all fishers" "(i-e., one that
incorporates a 0 g/d consumption rate for non-consumers). 16
Based on application of the above criteria to available studies characterizing recreational
freshwater fish consumption, it was decided that the ingestion rates for recreational freshwater
fishers specified as "recommended" in the EPA's Exposure Factors Handbook (mean of 8 g/day
and 95th percent of 25 g/day), represented the most appropriate values to use in this analysis.
These recommended values were derived based on ingestion rates from four studies conducted in
Maine, Michigan and Lake Ontario (Ebert et al., 1992; Connelly et al., 1996; West et al., 1989;
West et al., 1993) that matched the suitability criteria presented above (i.e., annual-averaged
daily intake rates for self-caught freshwater fish by all recreational fishers including consumers
and non-consumers). The mean values presented in these four studies ranged from 5 to 17 g/day,
while the 95th percent values ranged from 13 to 39 g/day (Note: the 39 g/day value actually
represents a 96th percent value). The EPA "recommended values" were developed by
16 Note: exposure modeling for recreational fishers could have focused on consumers and excluded those engaging
in catch-and-release (i.e., non-consumers). However this would require representative information on "percent
consumers" within the recreational fisher population to allow the overall recreational fisher population to be parsed
appropriately. While this information is available from a number offish consumption surveys, the values can differ
significantly across those surveys, suggesting that there is uncertainty associated with our ability to differentiate
(especially at the national level) recreational fishers based on consumption versus non-consumption status.
Consequently, EPA decided to model recreational fisher exposure focusing on all fishers and did not attempt to
differentiate consumers from non-consumers.
10-44
-------
considering the range and spread of means and 95th percent values presented in the four studies.
The average daily fish consumption rate of 8 g/day correspond to approximately (8 x 365 7117
=) 25 fish meals per year using the average fish meal size of 117 g/meal from Pao et al. (1982).
EPA recognizes that use of mean and 95th percent consumption rates based on these four studies
may not be representative of fishing behavior across the entire 37 state study area and that there
may be regional trends in consumption that differ from the values used in this analysis.
However, EPA believes that these four studies do represent the best available data for
developing recreational fisher ingestion rates for the 37 state study area that meet all of the study
suitability criteria presented above. It should be noted that there are a large number of local-
scale studies, including Creel surveys, that characterize fishing practices for specific waterbodies
or groups of waterbodies in particular geographic areas. However, these studies often do not
meet one or more of the suitability criteria presented above and are not readily generalizable to
larger regional areas. Consequently, these local-scale studies were not used in deriving ingestion
rates for this analysis.
Because the consumption rate estimate C is for cooked fish and the mercury
concentrations are estimated for uncooked filet, a conversion factor (CCF) was applied to
estimate mercury concentrations in cooked fish. Cooking fish tends to reduce the overall weight
offish by approximately one-third (Great Lakes Sport Fish Advisory Task Force, 1993).
Because volatilization of mercury is unlikely to occur during cooking, the overall amount of
mercury will stay unchanged during cooking, and the concentration of mercury will increase by
a factor of roughly 1.5 (Morgan, Berry, and Graves, 1997).
10.4,2 Modeling Approach for Estimating IQ Effects and Lost Earnings
Estimating the IQ decrements in children that result from mothers' ingestion of mercury
required two steps. First, based on the estimated average daily maternal ingestion rate, the
expected mercury concentration in the hair of exposed pregnant women was estimated as
follows:
CHgHj = (0.08)-1 * (Hgl/W) (Eq. 10.17)
where
CHgH = average mercury concentration in maternal hair (ppm)
W = average body weight for female adults ages 15-44 (= 64 kg)
This conversion rate between average daily ingestion rate and maternal hair
concentration is based on the one compartment toxicokinetic model used for deriving EPA's
reference dose for MeHg by Swartout and Rice (2000). Uncertainty and variability in various
input parameters was analyzed using Monte Carlo simulation to establish a relationship between
the mercury ingestion dose and hair mercury concentration. The simulation results indicated that
the median conversion factor was 0.08 jag/Kg-day of mercury ingestion for 1 ppm of mercury
concentration in hair. The 2002 EPA Workshop on Methylmercury Neurotoxicity recommended
that this one compartment model might be better suited than PBPK model in modeling dose-
response (EPA, 2002c). The average body weight estimate (W) was based on EPA's Exposure
Factor Handbook (EPA, 1997a).
10-45
-------
Second, to estimate the expected IQ decrement in offspring resulting from in utero
exposure to mercury through mothers' fish consumption, the following dose-response
relationship was applied:
; = 0.131 * CHgHj (Eq. 10.18)
where
dlQ = IQ decrement in exposed mother/child (IQ pts)
This dose response relationship is based on the statistical analysis in Section 9 that
integrates the results of the available epidemiological studies of mercury and neurobehavior
effects (see Section 9 for additional details).
The monetary value of losses resulting from IQ decrements were then assessed in terms
of foregone future earnings for the affected individuals. These losses were estimated using to the
following equation:
Vj = VIQj * dlQj (Eq. 10.19)
where
V = present value of net earnings losses per IQ point loss per prenatally
exposed child (1999 dollars)
VIQ = value per change in IQ point (= $8,807)
This valuation approach for assessing losses associated with IQ decrements is based on
an approach used by EPA to assess benefits for reductions in lead exposures (EPA, 2000a). For
that analysis, EPA used results from a study by Salkever (1995) to estimate the effects of IQ loss
on expected future earnings and years of education.
Salkever (1995) analyzes data from the National Longitudinal Study of Youth (NLSY)
and uses a three equation regression model to estimate the relationships between IQ levels,
educational attainment, and expected future earnings. The results of this study indicate that the
average effect (for men and women combined) of a one point decrease in IQ is:
1 . A 2.379 percent decrease in future earnings; and
2. A 0.1007 decrease in years of schooling.
To estimate the expected monetary value these effects, EPA first estimated the average
present value of future earnings at the time of birth for a person born in the U.S. Using earnings
data from the 1992 Current Population Survey (CPS) and discounting at a 3 percent annual rate,
this present value was estimated to be $366,021 in 1992 dollars.
EPA then estimated the average direct and indirect costs associated with one additional
year of schooling. Based on Department of Education data, the average annual expenditure per
student was estimated to be $5,500, and the average annual opportunity cost (lost income from
being in school) was estimated to be $10,925. Assuming that these costs were incurred at age 19
(based on an average of 12.9 years of education among those over age 25 in the U.S.) the
10-46
-------
combined present value of these two costs at time of birth (discounted at 3 percent) were
estimated to be $9,367 per additional year of schooling in 1992 dollars.
Combining these estimates with the results from the Salkever (1995) study summarized
above implies that the average present value of net earnings losses associated with a one point
decrease in IQ is $7,765 in 1992 dollars. This value is calculated as the average present value of
lost earnings per IQ point loss ($8,708 = $366,021 * 0.2379) minus the partially offsetting
change in average education costs per IQ point loss ($943 = $9,367 * 0.1007). Corrected for
inflation using the GDP deflator, the average present value of net earnings losses per IQ point
loss is $8,807 in 1999 dollars17.
It is important to note that this value per IQ point lost is considered a "cost of illness"
measure rather than a measure of willingness-to-pay (WTP) to prevent a loss of an IQ point. The
cost-of-illness approach simply measures ex post costs and does not attempt to measure the loss
in utility due to pain and suffering or the costs of any averting behaviors that individuals have
taken to avoid the illness altogether (EPA, 2000b). However, the cost-of-illness estimate may be
considered a lower bound estimate of WTP (Harrington and Portney, 1987; Berger et al., 1987).
The main reason that the cost of illness understates total WTP is the failure to account for many
effects of disease beyond those associated solely with net earnings.
10.5 Model Results: Estimated Benefits of Utility Mercury Emission Controls
Based on the modeling approach described above, mercury ingestion levels, IQ
decrements, and lost future earnings were estimated for each modeled subpopulation and then
aggregated across the study area. These estimates were calculated for the following conditions
and emissions control scenarios:
• 2001 Base Case
• 2001 Utility Emissions Zero-Out
• 2020 Base Case with CAIR
• 2020 Utility Emissions Zero-Out
• 2020 CAMR Control Option 1
• 2020 CAMR Control Option 2
For the 2001 Base Case, it was assumed the mercury emissions, and therefore mercury
concentration levels in fish, would remain constant at their currently observed levels
(summarized in Table 10-2).
For each of the other five emissions control scenarios, mercury concentration levels were
re-estimated for the entire study area. To estimate the effect of emissions controls, EPA first
17 The value per IQ estimate using a 7 percent discount rate is $1,580 per IQ point.
10-47
-------
conducted air quality modeling runs and estimated mercury deposition levels across the study
area for baseline (i.e., 2001 Base Case) emissions conditions. Based on the format of the air
quality modeling results, the study area was segmented into roughly 10,000 36x36 km grids, and
separate mercury deposition estimates were generated for each grid. EPA then repeated this
process for each of the five emissions control scenarios. Comparing each set of model results to
baseline results, EPA estimated percent reductions in mercury deposition relative to baseline for
each grid and for each emissions control scenario. Based on results from EPA's Mercury MAPs
(MMaps) approach discussed in Section 3, reductions in mercury fish tissue concentrations in
each grid were assumed to be directly proportional to the estimated mercury deposition
reductions for the grid. As Section 3 discusses, this proportionality assumption is limited to
areas where air deposition is the primary source of mercury loading to a waterbody, and is
applicable at steady-state (i.e., when the waterbody comes to equilibrium after a given change in
mercury loading).
Therefore, to re-estimate fish tissue mercury concentrations at each mercury sampling
point for each emissions control scenario, the following steps were applied:
1. Each lake and river sampling point in the study area was mapped to its corresponding
air quality grid using GIS.
2. The baseline (Base Case 2001) mercury fish tissue concentration estimates for each
sampling point (as summarized in Table 10-2) were reduced by the same percentage
as the estimated percent reduction in mercury deposition for its corresponding grid.
The resulting estimates of changes in mercury fish tissue concentrations are summarized in
Tables 10-11 and 10-12. Table 10-11 reports results relative to the 2001 baseline levels for two
emissions control scenarios: 2001 Utility Emissions Zero-Out and 2020 Base Case with CAIR.
Under the 2001 Utility Emissions Zero-Out scenario, the fish tissue mercury
concentrations in lakes and rivers were estimated to decline by an average of 11 and 15 percent
respectively across the study area, with the largest percentage reductions occurring in
Pennsylvania, Ohio, and West Virginia. The percent reductions across sampling points varied
from less than one percent to over 67 percent relative to 2001 baseline levels.
Under the 2020 Base Case with CAIR, mercury concentrations were estimated to decline
on average by similar amount—11 percent for lakes and 14 percent for rivers. Again the largest
percentage reductions were estimated to occur in Pennsylvania, Ohio, and West Virginia;
however, the range of estimated reductions across sampling points was much larger for this
scenario, going from over 10 percent increases in mercury levels at some points to over 60
percent reductions at others.
10-48
-------
Table 10-11. Effects of Emission Control Scenarios—Percent Reduction in Estimated Fish
Tissue Mercury Concentrations from 2001 Base Case'
Mean (Min-Max) Percent Reduction Across Sampling Sites
Utility Mercury Emissions Zero-Out in 2001
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
ND
NE
NH
NJ
NY
OH
OK
Lake Sites
10.6(0.6-67.7)
19(8-41.3)
7.3(4.7-18.3)
7.7(4.7-10.1)
—
19(16.7-22.5)
4.1(0.6-15.9)
15.3(5.4-33.6)
7.8(4.1-25.3)
23.4(9.9-40.3)
22.7(15.9-32.4)
5.3(3.6-8.6)
17.1(2.7-21.5)
5.6(3.1-11.3)
9.3(5.9-21.4)
33.6(22.8-50.5)
3.1(2.1-5)
14 (2.6 - 42.8)
2.3(0.9-9.5)
8.9(5.9-15.5)
6.9 (4.8 - 9.4)
18.5(10.1-30.8)
2.9(1.3-5)
5(1.7-25.3)
5.9(4.3-6.8)
16.9(10-28)
13.9(6.1-30.5)
27.2(15.4-45.6)
7.3(1.8-18.9)
River Sites
15.3(0.5-67.7)
16.8(7.6-53.4)
7.2(4.4-18.3)
8.3(4.7-10.1)
28.5(28.5-28.5)
19.6(12.3-29.5)
4.3(0.5-45.5)
15.2(4.8-42.3)
7.2(4.1-25.3)
16.4(6.2-30)
19.7(7.5-51.9)
7.6(2.9-13.2)
23.9(2.7-38.1)
4.4 (3 - 7)
3.6(3.6-3.6)
32.4(13.9-58.5)
3.4(2.1-6)
14.5(2.8-27.8)
2.9 (0.9 - 8.2)
8.9(6.2-11.2)
10.1(5.2-15)
16.4(8.5-51.1)
6.1(1.2-16.1)
5.4(1.4-25.3)
6(4.9-9)
13.1(10-28)
12.5(5.5-30.4)
29.8(15.4-55.8)
8.3(1.8-18.9)
Base Case with CAIR in 2020
Lake Sites
10.7 (-24.5 -61.8)
19.1 (-1.7 -41.2)
6 (-0.5 -10.2)
15.1(5.4-46.1)
—
15.6(11.7-19.7)
8.2 (-3. 1-19.3)
15.5(7-32.7)
4.5 (-2.7 -6.1)
15.5 (-0.9 -31.4)
12.1 (-0.9-33.6)
4.2(2.8-4.9)
22.1(15.6-41.4)
6.3(2.7-12.4)
11(4.6-18.1)
32(21-48.4)
5.6 (-2.9 - 37.2)
9.8 (0.5 - 30.7)
4.5 (-24.5 -4 1.9)
5.9 (-5.9 -2 1.7)
8.5(6.4-9.6)
18.3(12-30.9)
4.3(3-5.1)
2.4 (-15. 1-6.1)
8.4(3.8-15.6)
9.4(0.4-16)
14.3 (-7.6 -26.7)
25.6(16.1-45.3)
5.4 (-3.7 -13.8)
River Sites
14.4 (_27.l -61. 8)
15.5 (-3.3 -43 .4)
6.3 (-3.6 -13. 2)
13.1(5.1-46.1)
39.2(39.2-39.2)
17.6(2.5-25.7)
7 (-3. 1-40.5)
16(6.8-41.5)
2.9 (-12.5 -6.1)
8.1 (-15.2 -27.1)
16 (-9.3 -44.4)
3.8 (-0.9 -9.8)
23.1(2.9-41.4)
5.4(3.2-15.8)
55.9(55.9-55.9)
30.6(5.1-56.9)
4.6 (-13.3 -37.2)
16.8(3.7-40.1)
2.4 (-24.5 -4 1.9)
2.9 (-0.9 -5. 3)
17(4.2-50.3)
17.4(4.9-49.8)
6.3(2.6-10.4)
3.7 (-15. 1-8.7)
9.4(2.2-30.1)
2.6 (-7.6 -16)
11. 2 (-7.6 -35.9)
27.1 (9.7-53.7)
6.6 (-3.7 -35.8)
(continued)
10-49
-------
Table 10-11. Effects of Emission Control Scenarios—Percent Reduction in Estimated Fish
Tissue Mercury Concentrations from 2001 Base Case (continued)
Mean
(Min-Max) Percent Reduction Across Sampling Sites
Utility Mercury Emissions Zero-Out in 2001
State
PA
SC
TN
TX
VA
VT
WI
WV
Lake Sites
31.7(10.3-67.7)
15.3(13.4-20.2)
20.3(9.9-37.4)
9(1.1-64.4)
25.6(16.6-41.1)
6.8(5.7-8.3)
5.6(2.3-40.3)
33.2(32.8-34.3)
River Sites
34.8(10.3-67.7)
14.8(4.3-28.9)
21.7(11.5-37.4)
6.8 (0.5 - 30.7)
23.8(9.2-35.7)
6.9 (2 - 8.3)
8(2.8-40.3)
37.4(20.3-60.6)
Base Case with CAIR in 2020
Lake Sites
27.5(4.6-61.8)
14.8(12.5-20.7)
2 1.3 (-7.4 -38.9)
10.3(3.3-55.6)
23 (-4.8 -3 1.3)
10.8(7.5-22.6)
5.9 (-4.8 -24.1)
33.5(32.6-34.5)
River Sites
33 (-11.9 -61.8)
13. 5 (-2 1.5 -44.3)
22.4(12.7-38.9)
6.8 (-14.5 -26.2)
26(17.2-39.2)
11.9(4.7-30.1)
5.8 (-27. 1-35.6)
36.8(21.4-60.2)
' For summary purposes data are reported at a state level in this table. The analysis accounted for within state
variations in estimated percent reductions in mercury concentrations.
Using the resulting mercury concentration estimates from the 2020 Base Case with CAIR
scenario as a new reference point, Table 10-12 reports estimated reductions in mercury levels for
the other three emissions control scenarios. Under the 2020 Utility Emissions Zero-Out
scenario, the fish tissue mercury concentrations in lakes and rivers were estimated to decline by
an average of 5 and 6 percent respectively across the study area, with over 10 percent reductions
in both lake and river levels estimated to occur in Illinois, Michigan, and Missouri.
The 2020 CAMR Control Option 1 was estimated to reduce fish concentrations estimates
relative to the 2020 baseline by an average of less than 1 percent, whereas the 2020 CAMR
Control Option 2 was estimated to reduce these concentrations by an average of between 1 and 2
percent. In both cases, states with relatively high estimated reductions (3 percent or more
reductions in both lakes and rivers) included Pennsylvania and New Jersey.
Although Mercury Maps assumes that reductions in mercury deposition levels result in
roughly proportionate reductions in steady state fish tissue concentrations (everything else being
equal), there is uncertainty regarding how long it takes for these effects to occur (i.e., for fish
tissue concentrations to reach a new steady state), which is discussed in detail in Section 3 of this
report. Based on the response times from the case studies discussed in Section 318, to reflect the
possibility that reductions in mercury emissions in 2001 and in 2020 would have lagged effects
on fish tissue concentrations, we present a range of benefits based on the 10 and 20 year lags as
18 Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems
to reach a new steady state after a reduction in mercury deposition rates can be as short as 5 years or as long as 50
years or more. The medium response scenarios also varied widely but were generally on the order of one to three
decades.
10-50
-------
Table 10-12. Effects of Emission Control Scenarios—Percent Reduction in Estimated Fish Tissue Mercury Concentrations from
2020 Base Case with CAIRa
Mean (Min-Max) Percent Reduction Across Sampling Sites
Utility Mercury Emissions Zero-Out in 2020
Study
Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
Lake Sites
5.41(0.32-45.8)
4.82(2.58-13.16)
5.55(2.85-22.33)
3.03(2.4-3.86)
—
8.83(7.18-9.88)
1.5(0.32-4.16)
5.43(1.95-16.36)
9.02(4.25-25.26)
13.2(7.99-18.79)
14.08(7.03-25.12)
6.51(4.24-11.06)
5.72(1.84-12.06)
3.24(1.57-6.35)
2.92(2.1-5.59)
8.53(6.03-11)
2.3(0.95-10.41)
10.21(2.41-29.78)
2.49(0.96-9.29)
10.34(4.96-23.21)
River Sites
6.38(0.27-33.19)
5.31(2.39-24.74)
5.43(2.6-23.08)
2.95(2.4-3.86)
16.32(16.32-16.32)
10.51(7.05-18.64)
1.24(0.27-5.67)
4.77(1.79-33.19)
8.84(3.89-25.26)
11.56(4.51-24.5)
9.28(2.85-26.27)
9.78(3.07-16.52)
7.21(1.84-14.98)
2.51(1.57-6.17)
2.16(2.16-2.16)
8.02(3.04-12.7)
2.46 (0.95 - 10.41)
10.88(2.94-22.25)
3.01(1.06-12.15)
11.51(6.59-16.52)
CAMR Control Option 1
Lake Sites
0.7 (-^.68 - 36.5)
0.37 (-4.34 -1.91)
2.46(0.68-17.95)
0.67(0.56-1.08)
—
2.38(0.55-3.85)
0.1 (-1.39 -2.34)
0.39 (-0.44 -1.25)
1.71(0.45-7.56)
1.38(0.52-2.43)
1.95 (-0.13 -6.72)
1.22(0.63-2.41)
0.2(0.06-0.35)
0.69(0.24-4.02)
0.69(0.49-1.38)
2.63(1.52-4.36)
0.98(0.26-6.05)
0.61 (-3.63 -6.72)
0.2(0.05-0.83)
1.73(0.86-4.59)
River Sites
0.99 (-8.96 -19.49)
0.8 (-4.34 -5.93)
2.24(0.6-18.38)
0.76(0.54-1.08)
1.57(1.57-1.57)
3.56 (-7.46 -10.12)
-0.01 (-8.96 -2.34)
0.41 (-0.44 -1.68)
1.31(0.43-7.56)
1.19 (-0.42 -4.35)
0.63 (-3. 78 -6.66)
0.89(0.34-3.02)
0.36 (-0.41 -1.14)
0.37(0.16-0.7)
-1.33
(-1.33 — 1.33)
2.31 (-0.74 -4.54)
1.04(0.26-6.05)
0.6 (-0.97 -1.98)
0.22 (0.05 - 0.66)
1.22(0.59-2.15)
CAMR Control Option 2
Lake Sites
1.27 (-3.55 -36.92)
1.19(0.24-2.38)
2.97(0.92-18.28)
0.77(0.65-1.19)
—
2.84(0.96-4.47)
0.23 (-1.26 -2.57)
2.58 (-0.09 - 12.63)
3.09(1.31-11.87)
2.05(1.07-3.96)
2.37(0.14-7.31)
2.89(2.13-4.42)
0.53(0.29-0.65)
0.92(0.35-4.19)
0.8(0.58-1.5)
2.88(1.76-4.59)
1.18(0.33-8.59)
1.15 (-3.35 -7.31)
0.43(0.18-2.14)
2.95(1.8-6.09)
River Sites
1.65 (-8.83 -28. 13)
1.37 (-0.33 -6.11)
2.85(0.82-18.72)
0.87(0.65-1.19)
2.37(2.37-2.37)
4.01 (-6.75 -10.12)
0.12 (-8.83 -2.57)
1.95 (-0.09 -28. 13)
2.41(1.01-11.87)
1.78 (-0.1 1-4.76)
0.97 (-3 .62 -7.23)
3.95(1.73-6.14)
0.66 (-0.2 1-1. 4)
0.52(0.24-0.89)
-1.24
(-1.24 — 1.24)
2.79 (-0.39 -5.77)
1.29(0.33-8.59)
1.04 (-0.57 -2.61)
0.57(0.18-6.13)
3.39(1.49-5.9)
(continued)
-------
Table 10-12. Effects of Emission Control Scenarios—Percent Reduction in Estimated Fish Tissue Mercury Concentrations from
2020 Base Case with CAIR (continued)
Mean (Min-Max) Percent Reduction Across Sampling Sites
Utility Mercury Emissions Zero-Out in 2020
State
MS
NC
ND
NE
NH
NJ
NY
-------
central estimates. We also provide results for the 5 and 50 year lags to demonstrate how
benefits would differ under shorter and longer lag periods.
Results based on the modeling framework described above were estimated using
exposure estimates from both the population centroid approach and the angler destination
approaches. These results are summarized and discussed separately below.
The monetized benefits of the two CAMR control options—2020 CAMR Control Option
1 and 2020 CAMR Control Option 2—were estimated as the reduction in total IQ related losses
in the affected population relative to 2020 baseline conditions (2020 Base Case with CAIR). In
other words, benefits were calculated as the difference between (1) the estimated aggregate
present value of earnings losses in the 2020 baseline scenario and (2) the estimated aggregate
present value of earnings losses in the 2020 control scenario.
10.5.1 Results for the Population Centroid Approach
Applying the population centroid approach and equations 10.16 and 10.17 to estimate
mercury ingestion levels requires matching block group subpopulations (N) with corresponding
average mercury levels. As discussed in Section 10.1.2 one method for matching is to first
subdivide the exposed population in each block group (NPA) according to demographic group i,
distance interval j, and waterbody type k (lake or river) and then estimate the size of each
subpopulation, NPAp, using equation 10.7. This method assumes that each individual in the
subpopulation receives all of his/her noncommercial freshwater fish from the same distance
interval j and waterbody type k; therefore, each individual can be matched with the
corresponding average mercury concentration from that distance interval and waterbody type
(CHgjk). The results reported in this section were generated using this method to match
subpopulations and mercury concentrations.
One of the limitations of the available mercury concentration estimates, is that they do
not provide lake and river estimates for each possible location in the study area. As reported in
Table 10-9, for many block groups in the study area, mercury sampling locations did not exist
for each of the specified distance categories. To address these data limitations, a simple spatial
extrapolation method was used to provide estimates of exposures for all the distance intervals
within and beyond 100 miles of each block group centroid. For the four distance intervals within
100 miles of each centroid, if they did not contain a lake or river sampling location for mercury,
they were assigned the average lake or river concentration estimate (respectively) from the other
distance categories within 100 miles of the centroid. For exposures beyond 100 miles from each
centroid, the mean lake and river concentrations from the entire study area (0.23 ppm for lakes
and 0.25 ppm for rivers, as reported in Table 10-2) were used.
Table 10-13 reports model results for mercury ingestion rates under baseline conditions
in 2001. It disaggregates results by separating the estimated exposed subpopulations according
to their assumed distance to the freshwater fish they consume. It reports the distribution of
estimated Hgl levels for subpopulations in the four distance intervals below 100 miles. The
distributions are roughly similar for subpopulations in each distance interval, with an average
(and 75th percentile) Hgl of between 2.5 and 3 jig/day.
10-53
-------
Table 10-13. Estimated Distribution of Mercury Ingestion by Distance Traveled to Fish:
Population Centroid Approach—2001 Base Case
Average Daily Maternal Ingestion of Mercury (jag/day) by Distance Interval
0-10 miles
1st percentile
5th percentile
25th percentile
50th percentile
75th percentile
95th percentile
99th percentile
Mean
Std Dev
0.48
0.92
1.48
2.03
2.74
4.56
6.51
2.50
1.65
>10-20 miles
0.42
0.80
1.45
2.01
2.74
4.61
6.73
2.47
1.79
>20-50 miles >50-100 miles
0.53
0.91
1.49
2.13
2.75
4.66
6.86
2.46
1.37
0.80
1.12
1.70
2.23
2.93
4.67
6.46
2.71
1.23
>100 miles
2.72
2.72
2.76
2.76
2.79
2.86
2.90
2.77
0.23
Average
1.25
1.47
1.81
2.26
2.82
4.16
5.06
2.58
1.01
Table 10-14 summarizes model results for the 2001 Base Case based on exposure
estimates from the population centroid approach in 2001.19 The estimated annual number of
prenatally exposed children in freshwater angler households is just over 434,000. The states
with the largest estimated exposed populations are Texas and Illinois. The average daily
mercury ingestion rate for pregnant women in freshwater angler households (Hgl) was estimated
to be 2.44 jig/day for the entire study area. The states with the highest average rates were
primarily in New England, which is also where the highest estimates of average mercury
concentrations in freshwater fish were located. For example, the average rates for Maine and
New Hampshire were both estimated to be above 4 jig/day under baseline conditions. Average
IQ decrements in prenatally exposed children were estimated to be over 0.06 points for the study
area, with the highest average (per capita) reductions estimated again in New England states.
Under baseline conditions in 2001, the total IQ losses associated with self-caught freshwater fish
consumption were estimated to be 27,100 and the present value in 2001 of foregone net earnings
associated with these IQ decrements was estimated to sum to almost $240 million (in 1999
dollars).
Table 10-15 summarizes model results for the 2020 Base Case with CAIR16, based on the
exposure estimates from the population centroid approach for 2020, as reported in Table 10-8.
To provide a direct comparison with the 2001 Base Case results, the results reported in
Table 10-15 assume that all mercury emissions reductions associated with CAIR have been
implemented by 2020 and that there is no lag between emissions reductions and fish tissue
mercury concentrations. Due to population growth between 2001 and 2020, the estimated annual
number of prenatally exposed children in freshwater angler households is almost 482,000. The
average daily mercury ingestion rate for pregnant women in freshwater angler households (Hgl)
was estimated to be roughly 12 percent below the 2001 Base Case level, at 2.14 ^.g/day for the
19 For comparison purposes between the 2001 Base Case and the 2020 Base Case with CAIR, the benefits presented
in Tables 10-14 and 10-15 do not reflect potential lags in fish tissue response times for a change in mercury
deposition.
10-54
-------
Table 10-14. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2001 Base Case3
Annual Number of
Prenatally Exposed
Children
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
434,059
9,948
9,832
4,134
236
836
21,218
20,255
8,448
27,670
13,640
8,246
10,322
10,709
5,663
5,881
2,378
17,028
22,806
16,804
7,909
13,483
Average Daily Maternal
Ingestion of Mercury
(jig/day/person)
Mean
2.44
1.99
2.56
3.37
1.80
2.26
3.74
2.59
1.54
1.69
2.04
3.29
2.18
2.51
3.85
1.60
4.86
2.40
1.74
2.28
2.63
2.73
S.D.
1.01
0.72
0.57
0.90
0.20
0.28
0.88
0.68
0.19
0.28
0.33
1.00
0.47
0.69
0.35
0.28
0.73
0.34
0.30
0.97
0.52
0.66
IQ Decrements in Prenatally Exposed
Children
Mean
0.062
0.051
0.066
0.086
0.046
0.058
0.096
0.066
0.039
0.043
0.052
0.084
0.056
0.064
0.098
0.041
0.124
0.061
0.045
0.058
0.067
0.070
S.D.
0.026
0.018
0.015
0.023
0.005
0.007
0.022
0.017
0.005
0.007
0.008
0.026
0.012
0.018
0.009
0.007
0.019
0.009
0.008
0.025
0.013
0.017
Total
27,094
507
645
357
11
48
2,029
1,340
334
1,193
713
694
574
689
557
240
296
1,044
1,017
982
532
943
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2001; 1999$)
Mean
$550
$449
$578
$760
$406
$509
$842
$583
$348
$380
$460
$741
$490
$567
$867
$360
$1,095
$540
$393
$515
$592
$616
S.D.
$227
$161
$128
$202
$45
$63
$198
$153
$43
$63
$75
$225
$106
$155
$80
$62
$165
$76
$68
$218
$118
$150
Total
$238,626,988
$4,464,277
$5,680,355
$3,141,787
$95,779
$425,082
$17,872,632
$11,800,171
$2,937,846
$10,507,395
$6,280,533
$6,112,146
$5,059,688
$6,066,865
$4,909,319
$2,117,268
$2,604,794
$9,198,072
$8,955,172
$8,648,998
$4,681,160
$8,305,285
(continued)
-------
Table 10-14. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2001 Base Case (continued)
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Annual Number of
Prenatally Exposed
Children
2,101
5,138
1,966
6,838
16,877
21,457
13,117
15,146
755
9,010
2,508
13,317
55,802
11,793
1,288
15,848
3,651
Average Daily Maternal
Ingestion of Mercury
(lag/day/person)
Mean
2.22
1.73
4.25
3.08
3.06
2.54
2.49
3.34
3.66
3.10
2.01
2.28
2.01
1.95
3.58
2.18
2.66
S.D.
0.55
0.33
0.41
0.48
0.76
0.55
0.43
1.98
0.30
0.71
0.91
0.40
0.56
0.54
0.65
0.38
0.40
IQ Decrements in Prenatally Exposed
Children
Mean
0.057
0.044
0.109
0.079
0.078
0.065
0.064
0.086
0.094
0.079
0.051
0.058
0.052
0.050
0.092
0.056
0.068
S.D.
0.014
0.008
0.011
0.012
0.020
0.014
0.011
0.051
0.008
0.018
0.023
0.010
0.014
0.014
0.017
0.010
0.010
Total
119
228
214
538
1,320
1,395
837
1,295
71
715
129
776
2,875
588
118
882
248
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2001; 1999$)
Mean
$500
$391
$957
$693
$689
$573
$562
$753
$825
$699
$453
$513
$454
$439
$806
$490
$599
S.D.
$124
$73
$93
$107
$172
$125
$97
$447
$68
$161
$205
$90
$127
$121
$146
$85
$90
Total
$1,050,953
$2,008,393
$1,881,273
$4,739,979
$11,628,981
$12,288,120
$7,369,896
$11,405,684
$623,444
$6,301,059
$1,1-35,391
$6,834,282
$25,324,716
$5,174,750
$1,038,103
$7,770,204
$2,187,136
Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table. For comparison purposes with the 2020 Base Case with CAIR, benefits presented in this table do not reflect potential lags in fish
tissue response to a change in mercury deposition.
-------
Table 10-15. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2020 Base Case with CAIR"
o
-0
Annual Number of
Prenatally Exposed
Children
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
481,987
9,871
10,877
4,437
211
894
27,339
24,796
8,325
29,953
14,665
9,222
10,303
10,857
5,867
6,533
2,137
17,539
25,696
17,805
7,974
16,264
Average Daily Maternal
Ingestion of Mercury
(ja.g/day/person)
Mean
2.14
1.69
2.34
2.93
1.47
1.98
3.46
2.17
1.44
1.50
1.74
3.16
1.72
2.30
3.44
1.32
4.47
2.04
1.67
2.15
2.38
2.22
S.D.
0.80
0.72
0.52
0.75
0.06
0.23
0.88
0.55
0.15
0.18
0.26
0.97
0.32
0.62
0.25
0.21
0.54
0.26
0.24
0.96
0.47
0.59
IQ Decrements in Prenatally Exposed
Children
Mean
0.0548
0.0432
0.0599
0.0750
0.0375
0.0507
0.0885
0.0556
0.0368
0.0384
0.0445
0.0809
0.0441
0.0589
0.0879
0.0337
0.1144
0.0522
0.0427
0.0551
0.0609
0.0569
S.D.
0.0204
0.0185
0.0133
0.0192
0.0016
0.0060
0.0226
0.0140
0.0038
0.0045
0.0067
0.0249
0.0082
0.0158
0.0064
0.0053
0.0137
0.0067
0.0060
0.0245
0.0120
0.0150
Total
26,413
427
651
333
8
45
2,420
1,378
306
1,149
652
746
454
640
516
220
245
915
1,096
980
486
925
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2001; 1999$)
Mean
$1,412
$1,129
$2,687
$1,119
$161
$796
$2,339
$2,534
$1,024
$1,028
$1,198
$2,896
$1,267
$1,606
$899
$527
$1,885
$954
$2,366
$1,902
$1,992
$1,546
S.D.
$1,598
$1,205
$1,771
$809
$133
$617
$2,854
$2,760
$956
$1,014
$1,138
$2,729
$915
$1,218
$581
$424
$1,049
$749
$2,064
$1,838
$1,501
$1,474
Total
$232,623,719
$3,759,332
$5,736,031
$2,931,055
$69,675
$399,568
$21,317,014
$12,132,565
$2,696,242
$10,121,039
$5,746,041
$6,573,870
$4,001,455
$5,634,355
$4,544,090
$1,937,548
$2,154,061
$8,061,103
$9,656,733
$8,634,048
$4,278,349
$8,149,444
(continued)
-------
Ul
oo
Table 10-15. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and Foregone Earnings: Population
Centroid Approach—2020 with CAIR (continued)
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TO
TX
VA
VT
WI
WV
Annual Number of
Prenatally Exposed
Children
2,005
5,829
2,023
7,554
17,537
21,660
14,132
14,621
807
9,634
2,587
14,372
73,600
13,318
1,156
16,324
3,261
Average Daily Maternal
Ingestion of Mercury
(jag/day/person)
Mean
2.02
1.61
3.83
2.82
2.71
1.94
2.32
2.64
3.25
2.66
1.85
1.89
1.83
1.60
3.15
1.99
1.84
S.D.
0.50
0.28
0.26
0.28
0.46
0.32
0.37
1.43
0.09
0.60
0.99
0.26
0.43
0.47
0.52
0.33
0.26
IQ Decrements in Prenatally Exposed
Children
Mean
0.0516
0.0413
0.0980
0.0721
0.0694
0.0496
0.0595
0.0675
0.0830
0.0681
0.0475
0.0483
0.0469
0.0410
0.0806
0.0509
0.0472
S.D.
0.0128
0.0072
0.0067
0.0072
0.0117
0.0083
0.0095
0.0366
0.0022
0.0155
0.0252
0.0066
0.0111
0.0120
0.0133
0.0085
0.0068
Total
103
241
198
545
1,216
1,074
841
986
67
657
123
694
3,450
546
93
831
154
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2001; 1999$)
Mean
$1,452
$1,332
$1,998
$737
$710
$1,011
$2,552
$836
$719
$2,023
$1,616
$1,523
$2,117
$1,012
$1,549
$1,667
$853
S.D.
$1,261
$875
$1,444
$556
$653
$786
$1,842
$754
$455
$1,812
$1,918
$1,302
$2,118
$917
$936
$1,186
$515
Total
$910,706
$2,118,974
$1,745,904
$4,797,708
$10,713,538
$9,458,550
$7,402,873
$8,687,193
$590,322
$5,782,498
$1,081,144
$6,113,673
$30,388,891
$4,805,964
$820,772
$7,316,669
$1,354,724
Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table. For comparison purposes with the Base Cases in 2001, benefits presented in this table do not incorporate potential lags in fish tissue
response to a change in mercury deposition.
-------
entire study area.20 The states with the highest estimated average rates continued to be primarily
in New England. Average IQ decrements in prenatally exposed children were estimated to be
less than 0.06 points for the study area. Under the 2020 conditions with CAIR, the total IQ
losses associated with prenatal exposures to self-caught freshwater fish consumption in 2020
were estimated to be 26,400 IQ points and the present value in 2020 of foregone net earnings
associated with these IQ decrements was estimated to sum to almost $233 million (in 1999
dollars).
Tables 10-16, 10-17, and 10-18 report estimated reductions in exposures, IQ decrements,
and net earnings losses associated with CAIR and the two utility emissions zero-out scenarios.
Based on the response times from the case studies discussed in Section 321, to reflect the
possibility that reductions in mercury emissions in 2020 would have lagged effects on fish tissue
concentrations, we present a range of benefits based on the 10 and 20 year lags as central
estimates. We also provide results for the 5 and 50 year lags to demonstrate how benefits would
differ under shorter and longer lag periods. Differences in annual benefit estimates across the
selected lag periods are attributable to two factors. The first factor is differences in the size of
the exposed population, due to the inclusion of projected population growth factors across time.
For example, from 2001 to 2021, the exposed population was estimated to grow by 12 percent to
over 486,000. The second factor is differences in the present value of IQ-related losses. To
make the estimates of lost future earnings comparable across the four lag periods, they were all
expressed as present values in 2001 or 2020; therefore, losses to children born in future years
(2006, 2011, 2021, and 2051) were discounted assuming a 3 percent discount rate22.
Table 10-16 reports results for the 2020 with CAIR scenario, by comparing them to
conditions with (1) 2001 Base Case mercury levels in fish and (2) exposed population levels
estimated to occur in the year of the selected lag periods. The per capita IQ decrements avoided
were estimated to decrease by 0.007 points (12 percent). Total IQ decrements avoided are
estimated to decrease by approximately 3,900 to 4,200 points in 2020 under a 10 to 20 year lag
period. The mercury emission reductions associated with CAIR are estimated to reduce the
present value of total net earnings losses due to prenatal exposures in 2020 by approximately
$20.5 to $25.3 million under the 10 to 20 year lag periods. Under the alternative lag periods
considered in the analysis, the total IQ decrements avoided and monetary value of benefits are
3,500 IQ points at a value of $28.1 million under a 5 year lag and 5,200 IQ points at a value of
$10.5 million under a 50 year lag.
20 The average daily mercury ingestion rate given here of 2.14 ug/day from freshwater fish is below the EPA's
Reference Dose (RfD) for mercury of 5.8 ug/day. This estimate does not account for total exposure from
consumption from other rifhs sources. See Section 11 of this report for a detailed discussion of the implication of
the RfD on this analysis.
21 Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems
to reach a new steady state after a reduction in mercury deposition rates can be as short as 5 years or as long as 50
years or more. The medium response scenarios also varied widely but were generally on the order of one to three
decades.
22 Due to the large number of tables and calculations provided in this section, the implications on results of using a 7
percent discount rate are presented in the summary tables for each approach analyzed.
10-59
-------
Table 10-17 reports results for the 2001 Zero-Out relative to the 2001 Base Case. Under
this scenario, the average daily mercury ingestion rate for pregnant women in freshwater angler
households was estimated to decrease by roughly 13 percent for the entire study area. The
average per capita IQ decrements in prenatally exposed children were also estimated to decline
by roughly 13 percent under the zero out scenario, decreasing by an average of 0.008 points over
the entire study area. As shown in Table 10-17, the states that were estimated to benefit from the
largest per capita reductions in IQ losses are Pennsylvania and West Virginia (0.02 points in
each state). Total IQ decrements avoided are estimated to decrease by approximately 3,700 to
3,900 points in 2020 under a 10 to 20 year lag period. The mercury emission reductions
associated with the 2001 zero out of utility emissions are estimated to reduce the present value of
total net earnings losses due to prenatal exposures in 2020 by approximately $19.0 to $24.0
million under the 10 to 20 year lag periods. Under the alternative lag periods considered in the
analysis, the total IQ decrements avoided and monetary value of benefits are 3,600 IQ points at a
value of $27.5 million under a 5 year lag and 5,000 IQ points at a value of $10.0 million under a
50 year lag.
Table 10-18 reports comparable results for the 2020 Utility Emissions Zero-Out relative
to the 2020 Base Case with CAIR. The average daily mercury ingestion rate for pregnant
women in freshwater angler households and average per capita IQ decrements in prenatally
exposed children were estimated to decrease by roughly 5 percent for the entire study area.
Total IQ decrements avoided are estimated to decrease by approximately 1,500 to 1,600 points in
2020 under a 10 to 20 year lag period. The mercury emission reductions associated with the zero
out of emission remaining after 2020 with CAIR are estimated to reduce the present value of
total net earnings losses due to prenatal exposures in 2020 by approximately $8.0 to $10.0
million under the 10 to 20 year lag periods. Under the alternative lag periods considered in the
analysis, the total IQ decrements avoided and monetary value of benefits are 1,500 IQ points at a
value of $11.0 million under a 5 year lag and 2,000 IQ points at a value of $4.1 million under a
50 year lag.
10-60
-------
Table 10-16. 2020 Base Case
Demographics)—Population
with CAIR: Modelled Avoided Losses Relative to 2001 Base Case (Applied to 2020
Centroid Approach"'b
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Per Capita'
0.0073
0.0078
0.0047
0.0114
0.0085
0.0069
0.0071
0.0104
0.0028
0.0049
0.0080
0.0041
0.0112
0.0049
0.0105
0.0068
0.0095
0.0093
0.0020
0.0038
0.0063
0.0125
10-Yr
Lag
3,866
80
57
57
2
7
226
299
24
158
127
41
121
54
68
49
20
173
55
73
53
238
20-Yr
Lag
4,209
83
62
64
2
7
260
344
25
170
136
45
127
55
75
54
20
182
60
79
56
274
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
3,695
78
54
54
2
6
209
276
24
151
122
39
118
54
65
47
20
168
53
70
51
220
50-Yr.
Lag
5,238
94
79
84
3
9
361
481
27
207
164
58
145
59
95
68
20
209
74
96
64
380
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$25,335,445
$522,582
$372,468
$373,838
$12,639
$43,569
$1,481,167
$1,959,059
$157,468
$1,032,355
$830,840
$269,820
$793,923
$355,039
$447,789
$322,312
$132,502
$1,130,786
$363,410
$480,351
$346,536
$1,558,876
20-Yr Lag
$20,524,068
$406,589
$304,097
$310,480
$10,569
$34,696
$1,266,474
$1,679,796
$120,449
$828,493
$663,683
$221,159
$619,390
$269,529
$365,837
$262,666
$98,398
$886,044
$292,938
$385,610
$271,555
$1,333,737
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$28,068,173
$591,997
$410,802
$408,213
$13,744
$48,735
$1,589,056
$2,098,096
$179,995
$1,149,790
$927,759
$296,914
$898,068
$407,422
$493,684
$355,859
$153,759
$1,276,121
$403,744
$534,904
$391,058
$1,671,787
50-Yr. Lag
$10,522,314
$189,434
$158,587
$167,847
$5,794
$17,108
$724,896
$966,526
$53,675
$415,887
$329,616
$116,312
$290,574
$117,651
$191,062
$136,439
$40,296
$420,204
$148,529
$193,700
$128,809
$764,276
(continued)
-------
o
I
to
Table 10-16. 2020 Base Case with CAIR: Modelled Avoided Losses Relative to 2001 Base Case Applied to 2020
Demographics—Population Centroid Approach (continued)
Total Avoided IQ Decrements
Central Estimate of Fish
Tissue Response Times
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Per Capita0
0.0049
0.0028
0.0105
0.0067
0.0086
0.0152
0.0042
0.0168
0.0107
0.0110
0.0037
0.0097
0.0045
0.0080
0.0102
0.0047
00200
10-Yr
Lag
10
18
23
56
162
344
64
258
10
115
10
154
385
118
12
81
64
20-Yr
Lag
10
20
25
62
171
361
69
267
10
123
10
169
439
129
12
85
63
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
10
17
22
53
157
336
61
253
9
111
10
146
358
112
12
79
64
50-Yr.
Lag
11
25
32
80
199
413
84
296
13
148
12
214
599
163
13
97
62
Total Avoided Net Earnings Losses
(Present Yalue in 2020; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$64,360
$118,046
$152,746
$369,285
$1,062,160
$2,256,841
$419,008
$1,689,870
$62,288
$752,632
$65,022
$1,007,148
$2,522,650
$774,228
$79,133
$533,092
S4iqifi09
20-Yr Lag
$49,733
$96,254
$123,654
$303,096
$835,940
$1,762,498
$336,105
$1,304,303
$50,737
$600,633
$50,419
$823,356
$2,138,657
$631,324
$59,908
$415,898
S309.364
Alternative Estimate of Fish Tissue
Response Times
5-Yr Lag
$73,175
$130,292
$169,288
$406,047
$1,195,819
$2,551,488
$466,796
$1,922,502
$68,790
$840,877
$73,792
$1,109,960
$2,720,686
$854,523
$90,938
$603,022
$488 673
50-Yr. Lag
$22,767
$50,057
$63,299
$159,866
$400,747
$828,953
$168,535
$595,299
$26,328
$297,638
$23,289
$430,603
$1,204,386
$328,354
$25,949
$195,109
$123914
0 Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table.
b Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction in
mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades.
c Per capita IQ decrements and mercury ingestion rate vary only very slightly across different lag periods. Therefore, for brevity sake, we report the results for the
10 year lag period case.
-------
Table 10-17. 2001 Utility Mercury Emissions Zero Out: Modelled Avoided Losses Relative to 2001 Base Case—Population
Centroid Approach"*b
Ov
u>
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Per Capita'
0.0081
0.0077
0.0056
0.0076
0.0088
0.0091
0.0044
0.0097
0.0038
0.0081
0.0092
0.0066
0.0109
0.0043
0.0078
0.0077
0.0057
0.0114
0.0029
0.0063
0.0058
0.0125
10-Yr
Lag
3,674
76
57
32
2
8
105
216
31
232
129
57
111
45
44
47
13
196
70
106
46
183
20-Yr
Lag
3,891
77
61
34
2
8
122
247
32
247
137
62
113
46
46
51
12
202
77
112
47
206
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
3,623
76
56
31
2
8
100
206
32
228
127
55
112
45
44
47
13
196
68
106
46
176
50-Yr.
Lag
4,960
87
80
47
3
10
184
376
34
312
167
84
129
49
61
67
12
236
98
141
54
310
Total Avoided Net Earnings Losses
(Present Value in 2001; 3% discount rate;
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$24,080,244
$498,871
$374,531
$207,370
$12,861
$51,035
$690,798
$1,417,902
$205,618
$1,520,561
$848,492
$372,660
$724,909
$296,048
$290,190
$308,041
$82,477
$1,287,068
$457,894
$697,528
$302,039
$1,201,286
20-Yr Lag
$18,973,876
$373,165
$296,529
$166,027
$9,006
$40,007
$595,815
$1,205,174
$154,912
$1,204,058
$666,643
$302,207
$548,625
$225,716
$225,540
$248,152
$59,203
$983,452
$373,857
$546,921
$227,761
$1,002,879
1999$)
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$27,522,949
$580,312
$426,616
$238,981
$15,185
$58,919
$756,618
$1,566,735
$240,401
$1,728,824
$966,985
$420,109
$847,860
$344,048
$337,552
$353,594
$99,791
$1,492,774
$518,260
$803,148
$348,550
$1,333,984
50-Yr. Lag
$9,963,920
$174,977
$159,757
$94,933
$5,024
$20,280
$370,192
$754,952
$68,796
$626,223
$335,634
$169,416
$259,550
$98,620
$122,917
$134,345
$24,305
$474,548
$196,272
$282,823
$108,785
$623,774
(continued)
-------
Table 10-17. 2001 Utility Mercury Emissions Zero Out: Modelled Avoided Losses Relative to 2001 Base Case—Population
Centroid Approach (continued)
o
6\
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Per Capita'
0.0031
0.0042
0.0073
0.0111
0.0108
0.0159
0.0046
0.0203
0.0086
0.0117
0.0037
0.0095
0.0043
0.0113
0.0069
0.0069
0.0205
10-Yr
Lag
6
23
15
78
184
338
61
300
7
109
9
130
279
140
8
109
71
20-Yr
Lag
6
25
15
85
190
342
66
294
7
113
9
138
324
151
8
113
66
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
6
22
15
77
184
339
60
306
7
108
9
129
259
137
9
109
72
50-Yr.
Lag
6
32
19
114
226
395
82
328
9
139
11
182
473
200
8
130
64
Total Avoided Net Earnings Losses
(Present Value in 2001; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$40,482
$149,297
$95,231
$512,870
$1,207,790
$2,214,893
$398,767
$1,965,224
$44,208
$712,105
$59,949
$852,552
$1,828,979
$918,053
$55,273
$712,244
$464r152
20-Yr Lag
$29,829
$121,372
$72,725
$415,666
$926,866
$1,666,293
$319,584
$1,435,975
$34,111
$550,761
$46,251
$672,804
$1,577,641
$736,190
$38,890
$550,450
$322.820
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$47,850
$168,120
$111,107
$586,162
$1,399,430
$2,577,917
$455,579
$2,326,920
$50,576
$817,218
$69,514
$978,646
$1,968,963
$1,043,174
$66,685
$827,480
$548 360
50-Yr. Lag
$12,925
$64,976
$38,443
$228,680
$454,284
$793,664
$164,099
$658,018
$18,333
$279,334
$21,198
$366,004
$950,221
$401,475
$16,971
$260,986
$128 187
" Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table.
b Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction in
mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades.
c Per capita IQ decrements and mercury ingestion rate vary only very slightly across different lag periods. Therefore, for brevity sake, we report the results for the
10 year lag period case.
-------
Table 10-18. 2020 with CAIR Emissions Zero Out: Modelled Avoided Losses Relative to 2020 with CAIR Base
Case—Population Centroid Approach9'b
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Per Capita'
0.0029
0.0018
0.0032
0.0023
0.0025
0.0039
0.0012
0.0024
0.0026
0.0039
0.0035
0.0064
0.0023
0.0016
0.0027
0.0023
0.0033
0.0054
0.0015
0.0050
0.0019
0.0030
10-Yr
Lag
1,520
18
38
11
1
4
39
68
22
125
55
65
25
17
17
17
7
101
43
95
16
58
20-Yr
Lag
1,651
19
41
13
1
4
45
78
23
135
58
72
26
18
19
19
7
107
47
104
17
66
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
1,455
18
36
11
1
4
36
63
22
120
53
61
24
17
16
16
7
98
41
91
15
53
50-Yr.
Lag
2,043
22
52
17
1
5
62
108
24
166
68
96
30
19
24
23
7
124
58
130
19
92
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate;
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$9,962,464
$120,628
$247,726
$74,045
$3,735
$24,599
$254,611
$445,245
$144,142
$820,256
$357,823
$422,799
$163,530
$113,423
$113,281
$110,453
$46,197
$664,343
$282,405
$621,907
$103,267
$378,000
20-Yr Lag
$8,050,132
$93,754
$202,009
$61,506
$3,123
$19,586
$217,598
$379,786
$109,892
$660,566
$282,024
$353,482
$127,946
$86,041
$92,592
$90,289
$34,351
$522,489
$228,469
$505,203
$80,568
$322,827
1999$)
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$11,052,917
$136,729
$273,412
$80,846
$4,062
$27,519
$273,240
$478,394
$165,047
$911,783
$402,531
$459,854
$184,696
$130,207
$124,858
$121,735
$53,574
$748,223
$313,104
$687,897
$116,811
$405,830
50-Yr. Lag
$4,104,007
$43,564
$105,073
$33,261
$1,712
$9,653
$124,432
$216,388
$48,533
$334,213
$135,681
$193,683
$60,455
$37,480
$48,405
$47,211
$14,122
$250,057
$116,785
$260,597
$37,801
$184,367
(continued)
-------
Table 10-18. 2020 with CAIR Emissions Zero Out: Modelled Avoided Losses Relative to 2020 with CAIR Base
Case—Population Centroid Approach (continued)
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Per Capita0
0.0015
0.0032
0.0031
0.0041
0.0032
0.0037
0.0027
0.0053
0.0026
0.0036
0.0020
0.0021
0.0014
0.0051
0.0021
0.0037
0.0030
10-Yr
Lag
3
20
7
34
60
84
41
81
2
37
5
34
119
75
2
64
9
20-Yr
Lag
3
22
7
37
64
88
44
85
3
40
6
37
135
83
2
67
9
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
3
19
7
32
58
82
39
80
2
36
5
32
111
71
2
62
9
50-Yr.
Lag
3
28
9
48
75
101
55
96
3
48
6
47
184
106
3
76
9
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$20,371
$133,935
$44,812
$222,831
$391,969
$549,932
$266,732
$533,357
$15,007
$243,906
$34,762
$222,943
$782,514
$491,642
$16,030
$417,403
$61r899
20-Yr Lag
$15,544
$108,936
$36,300
$182,732
$309,697
$430,087
$215,406
$414,507
$12,219
$194,560
$26,989
$181,794
$660,448
$403,406
$12,151
$325,491
$45.764
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$23,314
$148,042
$49,647
$245,137
$440,351
$621,252
$296,026
$604,567
$16,577
$272,572
$39,424
$246,064
$846,244
$540,675
$18,410
$472,274
$7lr987
50-Yr. Lag
$6,882
$56,345
$18,608
$96,203
$149,881
$203,004
$109,671
$192,572
$6,335
$96,312
$12,507
$94,553
$368,728
$212,643
$5,282
$152,519
$18.490
* Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table.
b Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction in
mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades.
c Per capita IQ decrements and mercury ingestion rate vary only very slightly across different lag periods. Therefore, for brevity sake, we report the results for the
10 year lag period case.
-------
Table 10-19 report similar calculations for the 2020 CAMR Option 1. Relative to the
2020 Base Case with CAIR, this option is estimated to reduce per capita IQ decrements across
the study area by 0.0006 points. Total IQ decrements avoided under Option 1 are estimated to
decrease by approximately 318 to 346 points in 2020 under a 10 to 20 year lag period. The
mercury emission reductions associated with CAMR Option 1 are estimated to reduce the
present value of total net earnings losses due to prenatal exposures in 2020 by approximately
$1.7 to $2.0 million under the 10 to 20 year lag periods. Under the alternative lag periods
considered in the analysis, the total IQ decrements avoided and monetary value of benefits are
300 IQ points at a value of $2.3 million under a 5 year lag and 350 IQ points at a value of $0.8
million under a 50 year lag.
Table 10-20 report results of the 2020 CAMR Option 2. Relative to the 2020 Base Case
with CAIR, this option is estimated to reduce per capita IQ decrements across the study area by
0.0009 points. Total IQ decrements avoided under Option 2 are estimated to decrease by
approximately 475 to 520 points in 2020 under a 10 to 20 year lag period. The mercury emission
reductions associated with CAMR Option 2 are estimated to reduce the present value of total net
earnings losses due to prenatal exposures in 2020 by approximately $2.5 to $3.1 million under
the 10 to 20 year lag periods. Under the alternative lag periods considered in the analysis, the
total IQ decrements avoided and monetary value of benefits are 450 IQ points at a value of $3.4
million under a 5 year lag and 650 IQ points at a value of $1.3 million under a 50 year lag.
Table 10-21 summarizes the annual benefit estimates for CAMR Control Options 1 and
2, and it compares them to aggregate estimates associated with CAIR emissions reductions and
the two Utility Emissions Zero-Out scenarios (in 2001 and 2020). To assess the sensitivity of
these results to the assumed rate at which future gains/losses are discounted, the benefit
estimates are also reported assuming 3 and 7 percent discount rates. In addition, we present a 1
percent discount rate for the 50 year lag period to reflect the discount rate for inter-generational
effects as is recommended in the EPA Guidelines for Economic Analysis.
In comparison to the other mercury emissions reductions, the estimated benefits of
Option 1 are roughly 21 percent of the estimated benefits that would be achieved by eliminating
utility mercury emissions (i.e., 2020 Utility Emissions Zero-Out). The estimated benefits of
Option 2 are approximately 31 percent as large as those for the 2020 Utility Emissions Zero-Out
scenario.
10.5.2 Results for the Angler Destination Approach
This section summarizes results from applying the angler destination approach. As
reported in Table 10-3, many of the HUCs in the study area do not contain lake and/or river
mercury sampling locations. As was done in the population centroid approach, a simple spatial
extrapolation method was used to address these data limitations. For HUCs that do not contain a
lake (river) sample, they were assigned the lake (river) mean from the entire study area (e.g.,
0.23 ppm for lakes and 0.25 ppm for rivers in the 2001 Base Case).
10-67
-------
Table 10-19. Estimated Benefits of 2020 CAMR Control Option 1: Relative to 2020 with CAIR— Population Centroid
Approach0' b
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
Study Area
State
AL
AR
CT
DC
DE
FL
o GA
oo IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Per Capita'
0.0006
0.0002
0.0013
0.0006
0.0005
0.0014
0.0002
0.0003
0.0004
0.0005
0.0004
0.0012
0.0003
0.0003
0.0006
0.0008
0.0013
0.0004
0.0002
0.0009
0.0004
0.0005
10-Yr
Lag
318
3
15
3
0
1
6
8
3
18
7
12
3
4
4
6
3
8
4
18
4
9
20-Yr
Lag
346
3
16
3
0
1
7
9
3
19
7
14
3
4
5
6
3
8
5
19
4
11
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
304
3
14
3
0
1
6
8
3
17
6
12
3
4
4
5
3
8
4
17
3
9
50-Yr.
Lag
430
3
20
4
0
2
9
13
4
24
8
19
3
4
6
8
3
10
6
25
4
15
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$2,086,359
$16,779
$98,522
$19,128
$692
$8,852
$39,199
$53,256
$21,968
$115,700
$43,038
$81,222
$18,766
$23,896
$27,173
$37,122
$18,016
$52,581
$29,069
$116,271
$23,119
$60,245
20-Yr Lag
$1,687,988
$13,010
$79,453
$15,904
$579
$7,046
$33,373
$45,606
$16,694
$93,378
$33,644
$68,174
$14,509
$18,081
$22,211
$30,494
$13,390
$41,328
$23,439
$94,896
$17,953
$51,447
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$2,313,079
$19,042
$109,429
$20,872
$752
$9,904
$42,166
$57,081
$25,196
$128,451
$48,632
$88,131
$21,330
$27,468
$29,949
$40,798
$20,898
$59,240
$32,289
$128,264
$26,217
$64,684
50-Yr. Lag
$862,951
$6,009
$40,326
$8,617
$317
$3,470
$18,946
$26,179
$7,307
$47,477
$15,862
$37,650
$6,652
$7,820
$11,612
$16,113
$5,497
$19,750
$11,893
$49,451
$8,323
$29,376
(continued)
-------
Table 10-19. Estimated Benefits of 2020 CAMR Control Option 1: Relative to 2020 with CAIR—Population Centroid
Approach (continued)
o\
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Per Capita1
0.0004
0.0007
0.0008
0.0016
0.0010
0.0003
0.0004
0.0021
0.0006
0.0003
0.0004
0.0004
0.0003
0.0031
0.0005
0.0003
0.0008
10-Yr
Lag
1
5
2
14
19
6
5
33
1
3
1
7
29
46
1
6
3
20-Yr
Lag
1
5
2
15
21
6
6
34
1
3
1
7
33
51
1
6
3
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
1
4
2
13
19
6
5
32
0
3
1
6
27
44
1
6
3
50-Yr.
Lag
1
6
2
19
25
7
7
39
1
4
1
9
43
66
1
7
3
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate;
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$5,528
$29,969
$11,734
$89,843
$127,759
$38,067
$35,259
$213,167
$3,351
$20,552
$7,649
$43,676
$190,955
$304,403
$4,145
$38,029
$17.661
20-Yr Lag
$4,219
$24,408
$9,488
$73,692
$101,285
$29,742
$28,314
$166,443
$2,728
$16,486
$5,945
$35,203
$159,347
$250,091
$3,146
$29,668
$13.174
1999$)
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$6,326
$33,099
$13,013
$98,824
$143,262
$43,026
$39,256
$241,023
$3,701
$22,896
$8,669
$48,527
$207,925
$334,512
$4,757
$43,019
$20,448
50-Yr. Lag
$1,869
$12,662
$4,844
$38,814
$49,417
$14,004
$14,233
$78,245
$1,414
$8,267
$2,763
$17,845
$86,980
$132,187
$1,373
$13,917
$5,468
0 Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table.
b Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction in
mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades.
c Per capita IQ decrements and mercury ingestion rate vary only very slightly across different lag periods. Therefore, for brevity sake, we report the results for the
10 year lag period case.
-------
o
o
Table 10-20.
Approach8' b
Estimated Benefits of 2020
CAMR Control Option 2: Relative to 2020 with C AIR— Population Centroid
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Per Capita0
0.0009
0.0005
0.0017
0.0007
0.0006
0.0016
0.0003
0.0009
0.0007
0.0008
0.0006
0.0027
0.0004
0.0005
0.0008
0.0009
0.0016
0.0007
0.0003
0.0016
0.0006
0.0007
10-Yr
Lag
475
5
20
4
0
2
11
27
6
27
10
27
5
5
5
7
3
13
9
30
5
13
20-Yr
Lag
518
5
22
4
0
2
12
31
6
30
10
30
5
5
6
7
3
13
10
33
5
15
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
453
5
19
3
0
1
10
25
6
26
9
26
5
5
5
6
3
12
9
29
5
12
50-Yr.
Lag
648
6
28
5
0
2
17
42
7
37
12
40
5
6
7
9
3
15
13
42
6
20
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate;
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$3,112,816
$33,043
$133,134
$23,456
$927
$9,979
$68,824
$176,240
$40,034
$179,218
$62,282
$178,850
$30,227
$34,511
$33,380
$43,528
$22,357
$82,012
$62,124
$196,787
$34,120
$83,460
20-Yr Lag
$2,527,403
$25,636
$108,432
$19,500
$775
$7,943
$58,661
$150,021
$30,470
$144,755
$48,840
$148,340
$23,483
$26,132
$27,285
$35,723
$16,613
$64,560
$50,114
$161,225
$26,506
$71,474
1999$)
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$3,444,104
$37,489
$147,041
$25,598
$1,008
$11,165
$73,983
$189,602
$45,881
$198,881
$70,258
$195,450
$34,270
$39,654
$36,790
$47,864
$25,936
$92,320
$68,989
$216,604
$38,683
$89,452
50-Yr. Lag
$1,302,261
$11,857
$56,252
$10,564
$425
$3,913
$33,374
$85,143
$13,394
$73,728
$23,206
$79,973
$10,899
$11,327
$14,266
$18,838
$6,816
$30,968
$25,453
$84,711
$12,301
$41,030
(continued)
-------
Table 10-20. Estimated Benefits of 2020 With CAIR Control Option 2:
Approach (continued)
Relative to 2020 with CAIR—Population Centroid
Total Avoided
IQ Decrements
Central Estimate of Alternative Estimate
Fish Tissue Response of Fish Tissue
Times Response Times
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TO
TX
VA
VT
WI
WV
Per Capita0
0.0005
0.0015
0.0010
0.0018
0.0012
0.0001
0.0014
0.0023
0.0007
0.0008
0.0006
0.0006
0.0005
0.0033
0.0007
0.0010
0.0010
10-Yr
Lag
1
10
2
15
23
1
21
35
1
8
2
10
44
49
1
17
3
20-Yr
Lag
1
10
2
17
24
1
23
37
1
9
2
11
49
54
1
18
3
5-Yr 50-Yr.
Lag Lag
1
9
2
14
22
1
20
34
1
8
2
10
41
46
1
17
3
1
13
3
21
29
2
28
42
1
10
2
14
66
69
1
20
3
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$6,901
$62,393
$14,166
$99,911
$149,088
$8,284
$136,825
$228,368
$4,147
$52,469
$10,937
$67,360
$285,177
$321,251
$5,233
$110,883
$20.928
20-Yr Lag
$5,270
$50,688
$11,457
$81,948
$118,099
$6,667
$110,719
$178,561
$3,377
$41,957
$8,498
$54,560
$239,171
$263,882
$3,973
$86,510
$15.579
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$7,896
$69,011
$15,710
$109,901
$167,255
$9,212
$151,679
$258,016
$4,581
$58,555
$12,399
$74,632
$309,587
$353,067
$6,005
$125,426
$24.255
50-Yr. Lag
$2,337
$26,150
$5,852
$43,160
$57,508
$3,368
$56,625
$84,236
$1,751
$20,888
$3,946
$27,962
$131,872
$139,419
$1,734
$40,588
$6.427
° Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table.
b Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction in
mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades.
c Per capita IQ decrements and mercury ingestion rate vary only very slightly across different lag periods. Therefore, for brevity sake, we report the results for the
10 year lag period case.
-------
Table 10-21. Summary of Annual Benefit Estimates: Population Centroid Approach9
to
Central Estimate of Fish
Tissue Response Times
Annual Number of Prenatally Exposed Children
2001 Base Case
2020 Base Case (with CAIR)
Total Value of Benefits (1999$s)
2001 Utility Emissions Zero-Out (Relative to 2001 Base Case)
1% Discount Rate; Present Value in 2001
3% Discount Rate; Present Value in 2001
7% Discount Rate; Present Value in 2001
2020 Base Case with CAIR (Relative to 2001 Base Case applied to 2020
demographics)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 Utility Emissions Zero-Out (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
BENEFITS OF CONTROL OPTIONS
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
10-Yr Lag
452,575
528,721
N/A
$24,080,244
$16,451,115
N/A
$25,335,445
$17,308,643
N/A
$9,962,464
$6,806,145
N/A
$2,086,359
$1,425,357
N/A
$3,112,816
$2r 126.6 10
20-Yr Lag
486,487
577,910
N/A
$18,973,876
$8,855,743
N/A
$20,524,068
$9,579,269
N/A
$8,050,132
$3,757,266
N/A
$1,687,988
$787,840
N/A
$2,527,403
$lr!79r624
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
442,938
504,127
N/A
$27,522,949
$22,748,995
N/A
$28,068,173
$23,199,647
N/A
$11,052,917
$9,135,749
N/A
$2,313,079
$1,911,867
N/A
$3,444,104
$2,846.712
50-Yr Lag
632,017
725,474
$26,559,677
$9,963,920
$1,482,868
$28,048,124
$10,522,314
$1,565,971
$10,939,580
$4,104,007
$610,774
$2,300,269
$862,951
$128,428
$3,471,289
$1,302,261
$193.807
Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction
in mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on
the order of one to three decades.
-------
Figure 10-8 displays the estimated spatial distribution of 2001 baseline average mercury
ingestion rates across HUCs in the study area. Maine, Pennsylvania, Massachusetts, and New
Hampshire are the main states with high levels for average mercury ingestion rates. Given that
the fish consumption rate is constant across the study area, the higher mercury ingestion rate in
these states is primarily due to the relatively high average fish tissue mercury concentrations in
these states (as evident from Table 10-2). Figure 10-9 displays the 2001 baseline distribution of
total IQ point decrements across HUCs in the study area. States like New York, Texas, Florida,
Pennsylvania, and Ohio have higher total IQ decrements because of higher mercury
concentration and higher number of anglers together. The distribution of estimated percentage
reductions in per capita IQ decrements associated with the 2001 Utility Emissions Zero-Out
scenario are displayed in Figure 10-10. Higher percentage reductions from the baseline risk
levels are mostly in Pennsylvania, Ohio, Virginia and West Virginia.
Table 10-22 summarizes model results for the 2001 Base Case, based on exposure
estimates from the angler destination approach in 2001. The estimated annual number of
prenatally exposed children in freshwater angler households is close to 587,000. The states with
the largest estimated exposed populations are Texas, Minnesota and Illinois. The average daily
mercury ingestion rate for pregnant women in freshwater angler households (Hgl) was estimated
to be 2.7 u.g/day for the entire study area. The states with the highest average rates were
primarily in the Northeast (except for Florida). For example, the average rates for Maine,
Pennsylvania and New Hampshire were all estimated to be above 4 jag/day under baseline
conditions.23 Average IQ decrements in prenatally exposed children were estimated to be 0.07
points for the study area.
Under baseline conditions in 2001, the total IQ losses associated with self-caught
freshwater fish consumption were estimated to be 40,200 IQ points and the present value in 2001
of foregone earnings associated with these IQ decrements was estimated to sum to $354 million
(in 1999 dollars).
23 The average daily mercury ingestion rate given here of 2.7 ug/day from freshwater fish is below the EPA's
Reference Dose (RfD) for mercury of 5.8 ug/day. This estimate does not account for total exposure from
consumption from other rifhs sources. See Section 11 of this report for a detailed discussion of the implication of
the RfD on this analysis.
10-73
-------
Average Dally
Mercury Ingestion (ug/day)
-------
8 »
— • c
g 3
S' "?
3 -i
o tt
K Si
f5 —
T 2
vi
Si
O O"
o
n -;
s: W
09 VI
n a-
o
e
n
I
7
•8
Total IQ Points
C3oto<1
C31to<1°
50tD<100
100 or more
I
-------
1 1'
n ^
9 o
G S.
S. "*
E =*"'
<* 5.
M o
3 s
s* 2,
B, •*>
5' M
§ a.
v> —
?
!
8s
S S3
O Q.
O B
Percentage Reduction
c
y:
-------
-0
-J
Table 10-22. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and Foregone Earnings:
Destination Approach — 2001 Base Case"
Annual Number
of Prenatally
Exposed Children
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
586,516
13,077
14,669
4,762
8
592
22,748
21,530
12,069
33,696
23,517
11,929
15,840
12,491
6,808
7,318
2,977
20,998
36,700
19,720
12,417
19,451
Average Daily Maternal
Ingestion of Mercury
(Hg/day/person)
Mean
2.68
1.70
2.32
2.95
1.08
2.06
3.56
2.91
1.95
1.84
2.38
2.90
2.39
2.47
3.45
1.32
5.54
2.72
2.03
2.64
2.84
2.79
S.D.
1.55
1.31
1.01
1.34
0.06
0.60
1.41
1.58
0.72
0.96
0.78
1.03
0.60
0.93
1.08
1.00
1.39
1.16
0.71
0.53
0.57
1.02
IQ Decrements in Prenatally Exposed
Children
Mean
0.069
0.043
0.059
0.075
0.028
0.053
0.091
0.074
0.050
0.047
0.061
0.074
0.061
0.063
0.088
0.034
0.142
0.070
0.052
0.068
0.073
0.071
S.D.
0.040
0.034
0.026
0.034
0.002
0.015
0.036
0.040
0.018
0.025
0.020
0.026
0.015
0.024
0.028
0.026
0.035
0.030
0.018
0.013
0.014
0.026
Total
40,207
569
872
359
0
31
2,074
1,603
601
1,586
1,430
885
969
790
602
247
422
1,462
1,903
1,332
901
1,389
: Angler
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2001; 1999$)
Mean
$604
$383
$524
$665
$242
$463
$803
$656
$438
$415
$536
$653
$539
$557
$778
$297
$1,249
$613
$457
$595
$639
$629
S.D.
$350
$296
$229
$302
$13
$136
$318
$355
$162
$217
$177
$233
$134
$210
$243
$225
$312
$262
$161
$119
$128
$229
Total
$354,113,271
$5,009,131
$7,680,796
$3,165,411
$1,978
$274,454
$18,269,235
$14,115,893
$5,291,262
$13,971,988
$12,597,628
$7,793,163
$8,536,621
$6,961,559
$5,299,186
$2,175,909
$3,717,653
$12,875,688
$16,756,418
$11,729,835
$7,937,478
$12,237,634
(continued)
-------
Table 10-22. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and Foregone Earnings: Angler
Destination Approach—2001 Base Case (continued)
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Annual Number
of Prenatally
Exposed Children
2,684
5,477
2,805
10,520
21,401
30,556
23,791
24,704
835
11,837
3,798
20,723
64,537
18,224
1,726
24,429
5,149
Average Daily Maternal
Ingestion of Mercury
(Hg/day/person)
Mean
2.78
2.26
4.45
3.35
3.67
2.88
2.42
4.75
3.28
3.06
2.49
2.44
2.47
2.11
3.68
2.48
3.03
S.D.
0.37
0.79
.33
.47
.46
.05
.18
5.05
0.60
1.01
0.66
0.74
0.65
1.06
1.69
0.62
1.00
IQ Decrements in Prenatally Exposed
Children
Mean
0.071
0.058
0.114
0.086
0.094
0.074
0.062
0.121
0.084
0.078
0.064
0.063
0.063
0.054
0.094
0.063
0.077
S.D.
0.009
0.020
0.034
0.037
0.037
0.027
0.030
0.129
0.015
0.026
0.017
0.019
0.017
0.027
0.043
0.016
0.026
Total
191
317
319
903
2,011
2,255
1,470
3,000
70
925
242
1,295
4,084
982
163
1,551
399
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2001; 1999$)
Mean
$626
$510
$1,002.
$756
$828
$650
$544
$1,069
$740
$689
$561
$550
$557
$474
$830
$559
$682
S.D.
$83
$178
$301
$330
$328
$236
$265
$1,138
$135
$228
$149
$167
$147
$239
$381
$139
$225
Total
$1,680,883
$2,795,383
$2,809,423
$7,952,715
$17,712,372
$19,859,741
$12,950,097
$26,420,063
$618,220
$8,150,271
$2,131,921
$11,407,980
$35,973,218
$8,647,351
$1,431,637
$13,660,658
$3,512,420
* Benefits analyses using the angler destination approach were conducted at a HUC level, but the results are aggregated and reported at a state level in this table.
-------
Table 10-23 summarizes model results for the 2020 Base Case with CAIR based on the
exposure estimates from the angler destination approach. Due to population growth between
2001 and 2020, the estimated annual number of prenatally exposed children in freshwater angler
households is approximately 675,000. The average daily mercury ingestion rate for pregnant
women in freshwater angler households (Hgl) was estimated to be roughly 13 percent below
levels from the 2001 Base Case, at 2.33 ng/day for the entire study area. The states with the
highest estimated average rates continued to be primarily in Northeast. Average IQ decrements
in prenatally exposed children were estimated to be less than 0.06 points for the study area.
Under the 2020 conditions with CAIR, the total IQ losses associated with self-caught freshwater
fish consumption were estimated to be 40,200 and the present value in 2020 of foregone net
earnings associated with these IQ decrements was estimated at $354 million (in 1999 dollars).
These aggregate estimates are very similar to the 2001 Base Case results - the increase in
exposed population from 2001 to 2020 and the reduction in mercury levels due to CAIR have
roughly offsetting effects on the estimates of total IQ decrements.
Tables 10-24, 10-25, and 10-26 report estimated reductions in exposures, IQ decrements,
and net earnings losses associated with CAIR and the two utility emissions zero-out scenarios.
Based on the response times from the case studies discussed in Section 324, to reflect the
possibility that reductions in mercury emissions in 2020 would have lagged effects on fish tissue
concentrations, we present a range of benefits based on the 10 and 20 year lags as central
estimates. We also provide results for the 5 and 50 year lags to demonstrate how benefits would
differ under shorter and longer lag periods.
Table 10-24 reports results for the 2020 with CAIR scenario, by comparing them to
conditions with (1) 2001 Base Case mercury levels in fish and (2) exposed population levels in
2020 to 2070. The per capita IQ decrements were estimated to decrease by 0.0089 points (13
percent). Total IQ decrements were estimated to decrease by 6,700 to 7,400 points in 2020
under a 10 to 20 year lag period. The mercury emission reductions associated with CAIR are
estimated to reduce the present value of total net earnings losses due to prenatal exposures in
2020 by approximately $44.7 to $36 million under the 10 to 20 year lag periods. Under the
alternative lag periods considered in the analysis, the total IQ decrements avoided and monetary
value of benefits are 6,300 IQ points at a value of $48.1 million under a 5 year lag and 9,400 IQ
points at a value of $19 million under a 50 year lag.
Table 10-25 reports results for the 2001 Zero-Out relative to the 2001 Base Case. The
average per capita IQ decrements in prenatally exposed children were estimated to decline by
roughly 13 percent under the zero out scenario, decreasing by an average of 0.009 points over
the entire study area. As shown in Table 10-25 (and Figure 10-10), the states that were
estimated to benefit from the largest per capita reductions in IQ losses are Pennsylvania, West
Virginia, and Ohio (0.18 or more points in each state). Total IQ decrements avoided are
estimated to decrease by approximately 5,600 to 6,200 points in 2020 under a 10 to 20 year lag
24 Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems
to reach a new steady state after a reduction in mercury deposition rates can be as short as 5 years or as long as 50
years or more. The medium response scenarios also varied widely but were generally on the order of one to three
decades.
10-79
-------
period. The mercury emission reductions associated with the 2001 zero out of utility emissions
are estimated to reduce the present value of total net earnings losses due to prenatal exposures in
2020 by approximately $30.0 to $37.0 million under the 10 to 20 year lag periods. Under the
alternative lag periods considered in the analysis, the total IQ decrements avoided and monetary
value of benefits are 5,500 IQ points at a value of $41.5 million under a 5 year lag and 8,200 IQ
points at a value of $16.5 million under a 50 year lag.
Table 10-26 reports results for the 2020 Utility Emissions Zero-Out relative to the 2020
Base Case with CAIR. The average daily mercury ingestion rate for pregnant women in
freshwater angler households and average per capita IQ decrements in prenatally exposed
children were estimated to decrease by roughly 6 percent for the entire study area. Total IQ
decrements avoided are estimated to decrease by approximately 2,400 to 2,700 points in 2020
under a 10 to 20 year lag period. The mercury emission reductions associated with CAIR are
estimated to reduce the present value of total net earnings losses due to prenatal exposures in
2020 by approximately $13.1 to $16.0 million under the 10 to 20 year lag periods. Under the
alternative lag periods considered in the analysis, the total IQ decrements avoided and monetary
value of benefits are 2,300 IQ points at a value of $17.7 million under a 5 year lag and 3,500 IQ
points at a value of $6.9 million under a 50 year lag.
10-80
-------
o
oo
Table 10-23. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and Foregone Earnings:
Destination Approach— 2020 with CAIR"
Annual Number
of Prenatally
Exposed Children
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
674,357
15,035
16,866
5,475
9
681
26,155
24,755
13,876
38,743
27,039
13,716
18,213
14,362
7,828
8,414
3,423
24,143
42,196
22,673
14,277
22,364
Average Daily Maternal
Ingestion of Mercury
Gig/day/person)
Mean
2.33
1.44
2.17
2.52
0.68
1.80
3.25
2.52
1.77
1.59
2.03
2.62
2.03
2.28
3.00
1.06
5.23
2.46
1.92
2.39
2.52
2.30
S.D.
1.28
1.19
0.92
1.28
0.07
0.54
1.40
1.46
0.62
0.81
0.70
1.02
0.55
0.84
0.94
0.79
1.36
1.11
0.70
0.47
0.52
0.92
IQ Decrements in Prenatally Exposed
Children
Mean
0.060
0.037
0.055
0.064
0.017
0.046
0.083
0.064
0.045
0.041
0.052
0.067
0.052
0.058
0.077
0.027
0.134
0.063
0.049
0.061
0.065
0.059
S.D.
0.033
0.030
0.024
0.033
0.002
0.014
0.036
0.037
0.016
0.021
0.018
0.026
0.014
0.021
0.024
0.020
0.035
0.028
0.018
0.012
0.013
0.023
Total
40,211
554
935
353
0
31
2,178
1,595
629
1,573
1,407
920
947
838
602
227
458
1,520
2,069
1,384
921
1,316
Angler
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2020; 1999S)
Mean
$525
$325
$488
$567
$153
$405
$733
$567
$399
$357
$458
$591
$458
$514
$677
$238
$1,178
$555
$432
$538
$568
$518
S.D.
$289
$268
$208
$288
$15
$121
$316
$329
$139
$182
$157
$229
$125
$189
$211
$177
$306
$250
$158
$107
$116
$207
Total
$354,146,357
$4,880,859
$8,238,433
$3,105,516
$1,440
$276,132
$19,180,880
$14,048,148
$5,537,508
$13,849,691
$12,391,617
$8,102,384
$8,339,087
$7,376,996
$5,299,712
$2,002,919
$4,033,529
$13,388,662
$18,222,891
$12,187,216
$8,113,273
$11,587,446
(continued)
-------
oo
NJ
Table 10-23. Summary of Estimated Mercury Exposures, with Associated IQ Decrements and Foregone Earnings: Angler
Destination Approach—2020 with CAIR (continued)
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Annual Number
ofPrenatally
Exposed Children
3,086
6,297
3,225
12,096
24,606
35,133
27,354
28,404
961
13,610
4,367
23,827
74,202
20,954
1,984
28,088
5,921
Average Daily Maternal
Ingestion of Mercury
(Hg/day/person)
Mean
2.56
2.07
4.07
3.06
3.17
2.20
2.23
3.60
2.93
2.66
2.23
2.00
2.21
1.75
3.25
2.32
2.31
S.D.
0.33
0.70
1.24
1.39
1.26
0.70
1.07
3.85
0.56
0.93
0.56
0.64
0.57
0.93
1.42
0.59
0.64
IQ Decrements in Prenatally Exposed
Children
Mean
0.066
0.053
0.104
0.078
0.081
0.056
0.057
0.092
0.075
0.068
0.057
0.051
0.057
0.045
0.083
0.059
0.059
S.D.
0.008
0.018
0.032
0.036
0.032
0.018
0.027
0.098
0.014
0.024
0.014
0.016
0.015
0.024
0.036
0.015
0.016
Total
202
334
336
947
1,998
1,975
1,562
2,619
72
926
249
1,217
4,199
938
165
1,667
349
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2020; 1999$)
Mean
$578
$467
$918
$689
$715
$495
$503
$812
$660
$599
$502
$450
$498
$394
$731
$523
$520
S.D.
$75
$159
$278
$314
$283
$158
$240
$867
$127
$209
$125
$143
$129
$209
$321
$134
$144
Total
$1,782,188
$2,940,513
$2,959,412
$8,337,532
$17,595,195
$17,390,082
$13,758,988
$23,070,030
$634,231
$8,154,050
$2,191,439
$10,719,756
$36,978,242
$8,261,322
$1,451,289
$14,680,994
$3,076,753
Benefits analyses using the angler destination approach were conducted at a HUC level, but the results are aggregated and reported at a state level in this table. For
comparison purposes with the Base Cases in 2001, benefits presented in this table do not incorporate potential lags in fish tissue response to a change in mercury
deposition.
-------
Table 10-24. 2020 Base Case with CAIR: Modelled Avoided Losses Relative to 2001 Base Case Applied to 2020
Demographics—Angler Destination Approach"'b
9
oo
U)
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Per
Capita'
0.0089
0.0066
0.0040
0.0111
0.0101
0.0066
0.0079
0.0100
0.0045
0.0065
0.0088
0.0071
0.0092
0.0050
0.0115
0.0067
0.0080
0.0067
0.0028
0.0065
0.0081
0.0126
10-Yr Lag
6,679
111
75
67
0.11
5
230
275
69
279
264
108
186
79
100
63
30
178
131
164
128
313
20-Yr Lag
7,363
122
82
74
0.12
5
253
303
76
308
291
119
205
87
110
69
33
197
145
181
141
345
Alternative Estimate
of Fish Tissue
Response Times
S-Yr Lag
6,337
105
71
64
0.10
5
218
261
65
265
250
103
176
75
95
60
29
169
125
155
121
297
50-Yr.
Lag
9,415
156
105
95
0.15
7
324
388
97
393
372
152
262
111
141
89
43
251
185
231
180
441
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$43,769,459
$725,459
$489,454
$440,961
$689
$32,559
$1,506,686
$1,801,801
$451,069
$1,829,053
$1,728,205
$708,499
$1,218,938
$517,930
$654,970
$411,975
$198,943
$1,168,843
$861,393
$1,073,033
$836,533
$2,050,491
20-Yr Lag
$35,903,739
$595,088
$401,495
$361,717
$565
$26,708
$1,235,923
$1,478,003
$370,009
$1,500,357
$1,417,632
$581,176
$999,885
$424,854
$537,266
$337,940
$163,191
$958,792
$706,594
$880,201
$686,202
$1,682,001
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$48,142,771
$797,945
$538,359
$485,021
$758
$35,812
$1,657,230
$1,981,831
$496,139
$2,011,806
$1,900,882
$779,290
$1,340,731
$569,681
$720,412
$453,138
$218,821
$1,285,630
$947,461
$1,180,247
$920,117
$2,255,369
50-Yr. Lag
$18,913,979
$313,491
$211,506
$190,551
$298
$14,070
$651,080
$778,607
$194,919
$790,384
$746,804
$306,162
$526,736
$223,812
$283,030
$178,026
$85,969
$505,089
$372,232
$463,687
$361,489
$886,073
(continued)
-------
Table 10-24. 2020 Base Case with CAIR: Modelled Avoided Losses Relative to 2001 Base Case Applied to 2020
Demographics—Angler Destination Approach (continued)
o
oo
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TO
TX
VA
VT
WI
WV
Per
Capita1
0.0055
0.0049
0.0095
0.0076
0.0128
0.0176
0.0047
0.0292
0.0091
0.0102
0.0068
0.0114
0.0067
0.0091
0.0111
0.0041
0.0184
10-Yr Lag
19
34
34
102
349
686
142
921
10
153
33
302
552
212
25
129
121
20-Yr Lag
21
38
38
112
385
756
157
1,015
11
169
36
333
609
234
27
142
134
Alternative Estimate
of Fish Tissue
Response Times
S-Yr Lag
18
33
32
96
331
651
135
874
9
145
31
287
524
201
23
123
115
50-Yr.
Lag
27
49
48
143
492
967
201
1,298
14
216
46
426
778
299
35
182
171
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$124,232
$225,882
$223,606
$665,803
$2,287,436
$4,495,750
$933,671
$6,034,162
$63,239
$1,004,906
$214,525
$1,979,290
$3,619,204
$1,388,296
$160,835
$846,939
$794r199
20-Yr Lag
$101,907
$185,289
$183,422
$546,153
$1,876,365
$3,687,828
$765,883
$4,949,775
$51,874
$824,316
$175,974
$1,623,596
$2,968,804
$1,138,808
$131,932
$694,737
$651.475
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$136,645
$248,451
$245,948
$732,328
$2,515,989
$4,944,952
$1,026,961
$6,637,078
$69,557
$1,105,313
$235,960
$2,177,054
$3,980,824
$1,527,011
$176,906
$931,562
$873.553
50-Yr. Lag
$53,684
$97,610
$96,626
$287,712
$988,464
$1,942,736
$403,465
$2,607,526
$27,327
$434,247
$92,702
$855,305
$1,563,957
$599,921
$69,501
$365,985
$343.195
" Benefits analyses using the angler destination approach were conducted at a HUC level, but for summary purposes the results are aggregated and reported at a state
level in this table.
b Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction in
mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades.
c Estimated per capita IQ decrements and mercury ingestion rate do not vary across different lag periods with the angler destination approach.
-------
Table 10-25. 2001 Utility Mercury Emissions Zero Out: Modelled Avoided Losses Relative to 2001 Base Case—Angler
Destination Approach"'b
o
oo
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Per
Capita1
0.0090
0.0066
0.0051
0.0060
0.0081
0.0066
0.0055
0.0093
0.0051
0.0076
0.0095
0.0074
0.0087
0.0050
0.0087
0.0069
0.0051
0.0081
0.0018
0.0067
0.0075
0.0122
10-Yr Lag
5,636
91
79
30
0
4
134
213
65
273
237
94
147
66
63
53
16
181
69
141
99
252
20-Yr Lag
6,148
100
86
33
0
5
146
233
71
298
258
102
160
73
69
58
18
197
75
153
108
275
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
5,460
89
77
29
0
4
130
207
63
264
229
91
142
64
61
52
16
175
67
136
96
244
50-Yr.
Lag
8,202
133
115
44
0
6
195
310
95
397
345
136
214
97
92
78
24
263
100
205
143
367
Total Avoided Net Earnings Losses
(Present Value in 2001; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$36,934,050
$599,322
$517,992
$198,764
$460
$27,293
$876,952
$1,397,305
$426,459
$1,788,859
$1,552,030
$614,414
$961,667
$435,766
$414,894
$350,194
$106,020
$1,184,445
$450,121
$921,773
$646,120
$1,652,481
20-Yr Lag
$29,978,191
$486,451
$420,437
$161,331
$373
$22,153
$711,794
$1,134,148
$346,143
$1,451,960
$1,259,734
$498,700
$780,554
$353,698
$336,756
$284,241
$86,053
$961,376
$365,349
$748,173
$524,435
$1,341,266
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$41,479,456
$673,079
$581,740
$223,226
$516
$30,652
$984,877
$1,569,269
$478,942
$2,009,010
$1,743,036
$690,029
$1,080,017
$489,395
$465,954
$393,292
$119,068
$1,330,212
$505,517
$1,035,213
$725,636
$1,855,848
50-Yr. Lag
$16,477,629
$267,379
$231,095
$88,676
$205
$12,176
$391,240
$623,389
$190,259
$798,075
$692,417
$274,113
$429,035
$194,411
$185,099
$156,234
$47,299
$528,424
$200,815
$411,236
$288,258
$737,232
(continued)
-------
Table 10-25. 2001 Utility Mercury Emissions Zero Out: Modelled Avoided Losses Relative to 2001 Base Case—Angler
Destination Approach (continued)
o
oo
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Per
Capita'
0.0051
0.0047
0.0068
0.0125
0.0125
0.0180
0.0052
0.0323
0.0091
0.0106
0.0065
0.0109
0.0064
0.0103
0.0070
0.0042
0.0187
10-Yr Lag
15
28
20
140
284
586
133
851
8
134
26
241
438
200
13
109
103
20-Yr Lag
16
30
22
153
310
639
145
928
9
146
29
263
478
218
14
119
112
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
14
27
20
135
275
568
128
824
8
129
25
234
425
193
13
106
99
50-Yr.
Lag
21
40
30
203
414
853
193
1,238
12
194
38
351
638
291
19
159
149
Total Avoided Net Earnings Losses
(Present Value in 2001; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$95,044
$181,100
$133,480
$916,294
$1,863,495
$3,841,667
$868,975
$5,575,486
$52,962
$875,061
$172,035
$1,580,236
$2,873,128
$1,308,459
$84,701
$716,280
$672r317
20-Yr Lag
$77,145
$146,993
$108,341
$743,727
$1,512,540
$3,118,159
$705,320
$4,525,445
$42,987
$710,259
$139,635
$1,282,627
$2,332,026
$1,062,035
$68,749
$581,382
$545r698
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$106,741
$203,388
$149,907
$1,029,061
$2,092,832
$4,314,454
$975,919
$6,261,651
$59,480
$982,753
$193,206
$1,774,712
$3,226,719
$1,469,489
$95,125
$804,432
$755r058
50-Yr. Lag
$42,403
$80,795
$59,550
$408,792
$831,373
$1,713,908
$387,682
$2,487,428
$23,628
$390,397
$76,751
$705,001
$1,281,807
$583,751
$37,788
$319,559
$299.945
• Benefits analyses using the angler destination approach were conducted at a HUC level, but for summary purposes the results are aggregated and reported at a state
level in this table.
b Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction in
mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades.
c Estimated per capita IQ decrements and mercury ingestion rate do not vary across different lag periods with the angler destination approach.
-------
Table 10-26. 2020 with CAIR Emissions Zero Out: Modelled Avoided Losses Relative to 2020 with CAIR Base Case—Angler
Destination Approach"b
oo
-0
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Per
Capita1
0.0033
0.0018
0.0035
0.0019
0.0024
0.0025
0.0019
0.0028
0.0029
0.0031
0.0037
0.0038
0.0027
0.0024
0.0029
0.0017
0.0031
0.0047
0.0014
0.0035
0.0030
0.0032
10-Yr Lag
2,452
30
65
12
0
2
54
77
45
135
111
59
55
38
25
16
12
126
65
87
47
79
20-Yr Lag
2,703
33
72
13
0
2
60
85
50
149
123
65
61
42
28
17
13
139
71
96
52
87
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
2,327
28
62
11
0
2
52
73
43
128
106
56
52
36
24
15
11
120
61
83
45
75
50-Yr.
Lag
3,457
42
92
17
0
3
77
109
63
190
157
83
77
54
35
22
16
178
91
123
67
111
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$16,071,187
$196,083
$426,322
$77,362
$166
$12,610
$356,504
$506,464
$294,703
$884,090
$730,635
$383,870
$359,669
$249,120
$163,677
$103,428
$76,013
$827,885
$423,007
$570,043
$309,813
$515,616
20-Yr Lag
$13,183,067
$160,845
$349,709
$63,459
$136
$10,344
$292,437
$415,448
$241,743
$725,212
$599,334
$314,886
$295,034
$204,351
$134,263
$84,841
$62,353
$679,107
$346,989
$467,602
$254,137
$422,956
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$17,676,972
$215,675
$468,919
$85,091
$182
$13,870
$392,125
$557,068
$324,149
$972,426
$803,637
$422,225
$395,607
$274,011
$180,032
$113,762
$83,608
$910,605
$465,273
$627,000
$340,769
$567,135
50-Yr. Lag
$6,944,799
$84,733
$184,226
$33,430
$72
$5,449
$154,055
$218,857
$127,349
$382,040
$315,727
$165,881
$155,423
$107,651
$70,729
$44,694
$32,847
$357,752
$182,793
$246,331
$133,879
$222,812
(continued)
-------
oo
oo
Table 10-26. 2020 with CAIR Emissions Zero Out: Modelled Avoided Losses Relative to 2020 with CAIR Base Case—Angler
Destination Approach (continued)
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Per
Capita'
0.0025
0.0026
0.0039
0.0057
0.0038
0.0043
0.0031
0.0081
0.0034
0.0036
0.0027
0.0026
0.0026
0.0037
0.0023
0.0035
0.0038
10- Yr Lag
9
18
14
77
105
166
94
254
4
55
13
70
211
86
5
108
25
20-Yr Lag
10
20
15
85
115
183
104
280
4
61
14
77
232
95
5
119
27
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
8
17
13
73
99
158
90
241
3
52
12
66
200
82
5
102
24
50-Yr.
Lag
12
25
20
109
147
234
133
358
5
78
18
99
297
121
7
152
35
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999S)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$57,208
$117,725
$91,363
$504,947
$685,370
$1,089,536
$618,670
$1,664,222
$23,529
$360,945
$85,529
$458,050
$1,381,894
$563,901
$32,641
$706,115
$162r463
20-Yr Lag
$46,927
$96,569
$74,944
$414,204
$562,204
$893,738
$507,490
$1,365,148
$19,301
$296,080
$70,159
$375,735
$1,133,556
$462,563
$26,775
$579,220
$133.267
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$62,924
$129,488
$100,492
$555,399
$753,850
$1,198,399
$680,485
$1,830,505
$25,880
$397,009
$94,075
$503,817
$1,519,968
$620,244
$35,902
$776,667
$178.696
50-Yr. Lag
$24,721
$50,872
$39,481
$218,201
$296,167
$470,818
$267,344
$719,156
$10,168
$155,974
$36,959
$197,936
$597,154
$243,677
$14,105
$305,131
$70.205
* Benefits analyses using the angler destination approach were conducted at a HUC level, but the results are aggregated and reported at a state level in this table.
b Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction in
mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades.
' Estimated per capita IQ decrements and mercury ingestion rate do not vary across different lag periods with the angler destination approach.
-------
Table 10-27 report similar calculations for the 2020 CAMR Option 1. Relative to the
2020 Base Case with CAIR, this option is estimated to reduce per capita IQ decrements across
the study area by 0.0006 points. Total IQ decrements avoided under Option 1 are estimated to
decrease by approximately 460 to 500 points in 2020 under a 10 to 20 year lag period. The
mercury emission reductions associated with CAMR Option 1 are estimated to reduce the
present value of total net earnings losses due to prenatal exposures in 2020 by approximately
$2.5 to $3.0 million under the 10 to 20 year lag periods. Under the alternative lag periods
considered in the analysis, the total IQ decrements avoided and monetary value of benefits are
430 IQ points at a value of $3.3 million under a 5 year lag and 650 IQ points at a value of $1.3
million under a 50 year lag.
Table 10-28 report results of the 2020 CAMR Option 2. Relative to the 2020 Base Case
with CAIR, this option is estimated to reduce per capita IQ decrements across the study area by
0.0009 points. Total IQ decrements avoided under Option 2 are estimated to decrease by
approximately 700 to 775 points in 2020 under a 10 to 20 year lag period. The mercury emission
reductions associated with CAMR Option 2 are estimated to reduce the present value of total net
earnings losses due to prenatal exposures in 2020 by approximately $3.8 to $4.6 million under
the 10 to 20 year lag periods. Under the alternative lag periods considered in the analysis, the
total IQ decrements avoided and monetary value of benefits are 660 IQ points at a value of $5.0
million under a 5 year lag and 990 IQ points at a value of $2.0 million under a 50 year lag.
Table 10-29 summarizes the annual benefit estimates for CAMR Control Options 1 and
2, and it compares them to aggregate estimates associated with CAIR emissions reductions and
the two Utility Emissions Zero-Out scenarios. To assess the sensitivity of these results to the
assumed rate at which future gains/losses are discounted, the benefit estimates are also reported
assuming 3 and 7 percent discount rates. In addition, we present a 1 percent discount rate for the
50 year lag period to reflect the discount rate for inter-generational effects as is recommended in
the EPA Guidelines for Economic Analysis. Using a 3 percent discount rate, the aggregate
present value in 2020 of avoided net earnings losses due to reductions in mercury exposures for
selected years were estimated to range between $2.5 to $3.0 million for Option 1 and between
$3.8 to $4.6 million for Option 2.
In comparison to the other mercury emissions reductions, the estimated benefits of
Option 1 are roughly 19 percent of the estimated benefits that would be achieved by eliminating
utility mercury emissions (i.e., 2020 Utility Emissions Zero-Out). The estimated benefits of
Option 2 are almost 29 percent as large as those for the 2020 Utility Emissions Zero-Out
scenario.
10.5.3 Comparison of Results from Two Approaches
Table 10-30 summarizes and compares results from the population centroid approach and
the angler destination approach, for each of the selected years (i.e., lagged effects). Assuming a
3 percent annual discount rate, the estimated annual benefits associated with the CAMR Option
1 range from $1.7 to $2.0 million with the population centroid approach and $2.5 to $3.0 million
with the angler destination approach, for an average range across both approaches under the 10
to 20 year lag period of $1.7 to 3.0 million. For CAMR Option 2, the annual benefits range from
10-89
-------
$2.5 to $2.1 million with the population centroid approach and $3.8 to 4.6 million with the
angler
10-90
-------
Table 10-27.
Approach0 b
Estimated
Benefits of 2020 With CAIR Control Option 1: Relative to 2020 with CAIR— Angler Destination
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Per
Capita'
0.0006
0.0002
0.0014
0.0004
0.0003
0.0004
0.0002
0.0003
0.0004
0.0004
0.0004
0.0006
0.0003
0.0005
0.0005
0.0004
0.0012
0.0004
0.0001
0.0006
0.0005
0.0005
10-Yr Lag
457
4
27
3
0
0
5
8
7
16
12
9
6
9
5
4
4
11
6
15
7
12
20- Yr Lag
504
4
30
3
0
0
6
9
7
17
13
10
7
9
5
4
5
12
7
16
8
13
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
434
4
26
2
0
0
5
7
6
15
12
9
6
8
4
4
4
10
6
14
7
12
50-Yr.
Lag
644
6
38
4
0
0
8
11
9
22
17
13
9
12
7
5
6
15
9
21
10
17
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$2,995,451
$25,793
$176,298
$17,025
$20
$1,997
$35,495
$51,482
$43,839
$102,905
$79,906
$60,012
$40,766
$56,228
$30,738
$24,727
$29,092
$68,912
$40,853
$96,596
$48,514
$79,502
20-Yr Lag
$2,457,145
$21,158
$144,616
$13,966
$17
$1,638
$29,116
$42,230
$35,961
$84,412
$65,546
$49,227
$33,440
$46,123
$25,214
$20,283
$23,864
$56,528
$33,511
$79,237
$39,796
$65,215
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$3,294,747
$28,370
$193,913
$18,726
$22
$2,197
$39,041
$56,626
$48,219
$113,187
$87,890
$66,008
$44,840
$61,846
$33,810
$27,197
$31,998
$75,797
$44,935
$106,248
$53,361
$87,446
50-Yr. Lag
$1,294,416
$11,146
$76,183
$7,357
$9
$863
$15,338
$22,247
$18,944
$44,468
$34,530
$25,933
$17,616
$24,298
$13,283
$10,685
$12,571
$29,779
$17,654
$41,742
$20,964
$34,355
(continued)
-------
o
K)
Table 10-27. Estimated Benefits of 2020 With CAIR Control Option 1:
Approach (continued)
Relative to 2020 with CAIR—Angler Destination
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue
Response
Times
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Per
Capita1
0.0006
0.0004
0.0012
0.0023
0.0008
0.0003
0.0004
0.0030
0.0006
0.0003
0.0004
0.0005
0.0004
0.0016
0.0006
0.0003
00011
10-Yr Lag
2
3
4
31
22
14
11
93
1
5
2
14
32
37
1
9
7
20-Yr Lag
2
3
5
34
24
15
13
103
1
6
2
15
35
41
1
9
8
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999S)
Alternative Estimate
of Fish
Tissue
Response Times
5-Yr
Lag
2
3
4
29
21
13
11
89
1
5
2
13
30
35
1
8
7
50-Yr.
Lag
3
4
6
43
31
19
16
132
1
7
3
20
45
52
2
12
10
Central Estimate
of Fish Tissue
Response Times
10-Yr Lag
$12,868
$20,484
$28,501
$200,521
$143,881
$88,663
$75,345
$612,298
$3,987
$32,924
$13,706
$91,916
$209,545
$240,792
$7,971
$55,817
$45.531
20-Yr Lag
$10,556
$16,803
$23,379
$164,486
$118,024
$72,730
$61,805
$502,263
$3,271
$27,008
$11,243
$75,398
$171,888
$197,519
$6,538
$45,787
$37 349
Alternative
Estimate of Fish
Tissue Response Times
5-Yr Lag
$14,154
$22,530
$31,349
$220,556
$158,257
$97,522
$82,874
$673,477
$4,386
$36,214
$15,075
$101,100
$230,482
$264,851
$8,767
$61,395
$50 080
50-Yr. Lag
$5,561
$8,852
$12,316
$86,650
$62,175
$38,314
$32,559
$264,591
$1,723
$14,228
$5,923
$39,720
$90,550
$104,053
$3,444
$24,120
$19 675
• Benefits analyses using the angler destination approach were conducted at a HUC level, but the results are aggregated and reported at a state level in this table.
b Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction in
mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades.
c Estimated per capita IQ decrements and mercury ingestion rate do not vary across different lag periods with the angler destination approach.
-------
Table 10-28.
Approach01 b
Estimated
Benefits of 2020 With CAIR Control Option 2: Relative to 2020 with C AIR— Angler Destination
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue
Response
Times
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Per
Capita'
0.0009
0.0004
0.0018
0.0005
0.0004
0.0006
0.0004
0.0010
0.0008
0.0006
0.0006
0.0012
0.0005
0.0007
0.0007
0.0005
0.0014
0.0007
0.0003
0.0010
0.0008
0.0007
10-Yr Lag
700
7
35
3
0
0
11
28
12
25
19
18
11
12
6
5
5
19
12
25
12
18
20-Yr Lag
772
8
38
3
0
1
12
31
13
28
21
20
12
13
7
5
6
21
14
27
14
19
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999$)
Alternative Estimate
of Fish
Tissue
Response Times
5-Yr
Lag
664
7
33
3
0
0
11
27
11
24
18
17
10
11
6
5
5
18
12
23
12
17
50-Yr.
Lag
987
10
49
4
0
1
16
40
17
35
27
26
15
16
9
7
8
27
18
35
17
25
Central Estimate
of Fish Tissue
Response Times
10-Yr Lag
$4,586,570
$48,579
$226,689
$20,176
$28
$3,145
$73,160
$186,772
$77,054
$163,649
$123,657
$120,199
$71,142
$76,098
$42,065
$32,228
$35,489
$124,334
$81,863
$161,994
$80,345
$115,883
20-Yr Lag
$3,762,327
$39,849
$185,951
$16,551
$23
$2,580
$60,013
$153,207
$63,207
$134,240
$101,435
$98,598
$58,357
$62,423
$34,506
$26,436
$29,112
$101,990
$67,151
$132,882
$65,907
$95,058
Alternative
Estimate of Fish
Tissue Response Times
5-Yr Lag
$5,044,846
$53,433
$249,339
$22,192
$31
$3,460
$80,470
$205,434
$84,753
$180,000
$136,012
$132,209
$78,250
$83,702
$46,268
$35,448
$39,035
$136,757
$90,042
$178,179
$88,373
$127,462
50-Yr. Lag
$1,981,982
$20,992
$97,959
$8,719
$12
$1,359
$31,614
$80,709
$33,297
$70,717
$53,435
$51,941
$30,742
$32,884
$18,177
$13,926
$15,336
$53,728
$35,375
$70,002
$34,719
$50,076
(continued)
-------
o
vb
-p..
Table 10-28. Estimated Benefits of 2020 With CAIR Control Option 2:
Approach (continued)
Relative to 2020 with CAIR—Angler Destination
Total Avoided IQ Decrements
Central Estimate of
Fish Tissue Response
Times
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Per
Capita1
0.0008
0.0008
0.0014
0.0025
0.0011
0.0007
0.0013
0.0033
0.0009
0.0008
0.0007
0.0008
0.0006
0.0018
0.0007
0.0008
0.0013
10-Yr Lag
3
6
5
33
30
27
39
105
1
12
3
21
53
41
2
26
9
20-Yr Lag
3
6
5
37
33
30
43
116
1
13
4
23
59
46
2
29
9
Alternative Estimate
of Fish Tissue
Response Times
5-Yr
Lag
2
5
5
31
28
26
37
100
1
12
3
20
51
39
1
25
8
50-Yr.
Lag
4
8
7
47
42
38
55
148
1
17
5
30
75
58
2
37
12
Total Avoided Net Earnings Losses
(Present Value in 2020; 3% discount rate; 1999$)
Central Estimate of Fish Tissue
Response Times
10-Yr Lag
$17,003
$36,187
$32,423
$217,220
$194,435
$178,335
$256,552
$687,982
$5,986
$79,822
$21,853
$137,697
$349,061
$271,563
$9,959
$170,194
$55r746
20-Yr Lag
$13,947
$29,684
$26,597
$178,184
$159,493
$146,287
$210,448
$564,346
$4,911
$65,478
$17,926
$112,952
$286,332
$222,761
$8,169
$139,609
$45r728
Alternative Estimate of Fish
Tissue Response Times
5-Yr Lag
$18,702
$39,802
$35,663
$238,924
$213,862
$196,154
$282,186
$756,723
$6,585
$87,798
$24,036
$151,456
$383,938
$298,697
$10,954
$187,199
$61.316
50-Yr. Lag
$7,347
$15,637
$14,011
$93,867
$84,021
$77,064
$110,863
$297,296
$2,587
$34,493
$9,443
$59,503
$150,839
$117,350
$4,304
$73,546
$24.089
' Benefits analyses using the angler destination approach were conducted at a HUC level, but the results are aggregated and reported at a state level in this table.
b Case studies of individual ecosystems (as presented in Section 3) show that the time necessary for aquatic systems to reach a new steady state after a reduction in
mercury deposition rates can be as short as 5 years or as long as 50 years or more. The medium response scenarios also varied widely but were generally on the
order of one to three decades.
c Estimated per capita IQ decrements and mercury ingestion rate do not vary across different lag periods with the angler destination approach.
-------
Table 10-29. Summary of Annual Benefit Estimates: Angler Destination Approach
Central Estimate of Fish Tissue Alternative Estimate of Fish Tissue
Response Times Response Times
Annual Number of Prenatally Exposed Children
2001 Base Case
2020 Base Case (with CAIR)
Total Value of Benefits (1999$s)
2001 Utility Emissions Zero-Out (Relative to 2001 Base Case)
1% Discount Rate; Present Value in 2001
3% Discount Rate; Present Value in 2001
7% Discount Rate; Present Value in 2001
2020 Base Case with CAIR (Relative to 2001 Base Case applied to 2020
demographics)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 Utility Emissions Zero-Out (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
BENEFITS OF CONTROL OPTIONS
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate: Present Value in 2020
10-Yr Lag
624,862
748,424
N/A
$36,934,050
$25,232,565
N/A
$43,769,459
$29,902,373
N/A
$16,071,187
$10,979,497
N/A
$2,995,451
$2,046,429
N/A
$4,586,570
$3.133,448
20-Yr Lag
681,608
825,065
N/A
$29,978,191
$13,991,825
N/A
$35,903,739
$16,757,477
N/A
$13,183,067
$6,152,979
N/A
$2,457,145
$1,146,832
N/A
$3,762,327
$1.756.004
5-Yr Lag
605,347
710,103
N/A
$41,479,456
$34,284,695
N/A
$48,142,771
$39,792,234
N/A
$17,676,972
$14,610,837
N/A
$3,294,747
$2,723,262
N/A
$5,044,846
$4.169.799
50-Yr Lag
909,371
1,054,990
$43,922,522
$16,477,629
$2,452,263
$50,416,821
$18,913,979
$2,814,850
$18,511,953
$6,944,799
$1,033,551
$3,450,377
$1,294,416
$192,640
$5,283,142
$1,981,982
$294.966
-------
Table 10-30. Summary and Comparison of Annual Benefit Estimates:
Approach
Population Centroid Approach vs. Angler Destination
o
vb
Central Estimate of Fish Tissue Alternative Estimate of Fish Tissue
Response Times Response Times
Population Centroid Approach
Annual Number of Prenatally Exposed Children
2001 Base Case
2020 Base Case (with CAIR)
BENEFITS OF CONTROL OPTIONS
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
Angler Destination Approach
Annual Number of Prenatally Exposed Children
2001 Base Case
2020 Base Case (with CAIR)
BENEFITS OF CONTROL OPTIONS
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate: Present Value in 2020
10-Yr Lag
452,575
528,721
N/A
$2,086,359
$1,425,357
N/A
$3,112,816
$2,126,610
624,862
748,424
N/A
$2,995,451
$2,046,429
N/A
$4,586,570
$3.133.448
20-Yr Lag
486,487
577,910
N/A
$1,687,988
$787,840
N/A
$2,527,403
$1,179,624
681,608
825,065
N/A
$2,457,145
$1,146,832
N/A
$3,762,327
$1.756.004
5-Yr Lag
442,938
504,127
N/A
$2,313,079
$1,911,867
N/A
$3,444,104
$2,846,712
605,347
710,103
N/A
$3,294,747
$2,723,262
N/A
$5,044,846
$4.169.799
50-Yr Lag
632,017
725,474
$2,300,269
$862,951
$128,428
$3,471,289
$1,302,261
$193,807
909,371
1,054,990
$3,450,377
$1,294,416
$192,640
$5,283,142
$1,981,982
$294.966
-------
destination approach, for an average range across both approaches under the 10 to 20 year lag
period of $1.7 to 3.8 million. The main source of difference between the two approaches are the
estimated sizes of the exposed populations. The estimates from the angler destination approach
are generally between 35 percent and 40 percent higher than from the other approach. As
described in Section 10.1.2 the two approaches use different modeling assumptions to identify
numbers of pregnant women in angler households. As discussed in more detail in Section 10.7,
the assumptions used in the population centroid approach are likely to underestimate these
numbers, whereas the assumptions of the angler destination approach are likely to overestimate
them. Therefore, the results of the two modeling approaches can be interpreted as providing
lower and upper bound average estimates of exposed populations and benefits.
10.5.4 Sensitivity Analysis of Alternative Dose-Response Functions
EPA conducted a sensitivity analysis of alternative dose-response functions provided in
Section 9 of this report. Specifically, we estimated the benefits with a lower dose-response
function with a beta coeffiicent of-0.108, and a higher dose-response function with a beta
coefficient of-0.233. The findings of the sensitivity analysis are present in Table 10-31 and
show that with the lower dose-response function, the benefits would decrease by about 18
percent, while the higher dose-resopnse function would increase benefits by about 56 percent.
Table 10-31. Summary and Comparison of Annual Benefit Estimates Under Alternative
IQ Dose-Response Assumptions: Population and Angler Destination Approach
5-Yr Lag
20-Yr. Lag
Population Centroid Approach
2020 Base Case (with CAIR)
BENEFITS OF CONTROL OPTION 1
504,127
577,910
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR; 3% Discount Rate; Present Value
in 2020)
IQ Dose-Response Coefficient = -0.108 $ 1,906,966 $ 1,391,623
IQ Dose-Response Coefficient = -0.131 $ 2,313,079 $ 1,687,988
IQ Dose-Response Coefficient =-0.233 $ 4,114,102 $ 3,002,299
Angler Destination Approach
Annual Number of Prenatally Exposed Children
2020 Base Case (with CAIR)
BENEFITS OF CONTROL OPTION 1
710,103
825,065
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR; 3% Discount Rate; Present Value
in 2020)
IQ Dose-Response Coefficient = -0.108 $ 2,716,280 $ 2,025,738
IQ Dose-Response Coefficient = -0.131 $ 3,294,747 $ 2,457,145
IQ Dose-Response Coefficient = -0.233 $ 5,860,123 S 4,370,341
10-97
-------
10.5.5 Distribution ofPer-Capita IQ Changes for the Exposed Population (in support of
distributional equity analysis)
In addition to considering the net benefits associated with reductions in mercury fish
tissue concentrations, an additional factor to consider as part of a cost-benefit (as stipulated in
Executive Order 12866) is the distributional equity of benefits in relation to the distribution of
societal costs. To support an assessment of distributional equity in the context of this analysis,
EPA has modeled per-capita IQ changes for the population of modeled recreational fishers.
Consideration of the distribution of IQ changes across the modeled population using these results
supports a determination regarding the degree of equity in benefits associated with this
regulation.
Thus far, the analysis of mercury ingestion levels has assumed a constant rate of
freshwater fish consumption (C) across the exposed population. However, the population
centroid approach can also be adapted to allow for variation in C. As described above in Section
10.1.4, the constant consumption rate (C) is based on the recommended average rate for
freshwater recreational anglers in EPA's Exposure Factor Handbook (EFH) (EPA, 1997b).
However, the EFHaho provides guidelines regarding the variability in average daily fish
consumption rates across the recreational freshwater fishing population. It recommends mean
and 95th percentile daily self-caught fish freshwater fish consumption rate for freshwater
recreational anglers of 8 g/day and 25 g/day respectively. The distribution of consumption rate
is skewed to the right because the 95th percentile is more than 3 times the mean. Therefore, the
variability in the consumption rate can be represented by a distribution such as lognormal
distribution. The EFH contains examples of lognormal distributions fitted to fish consumption
data. Based on EFH recommendations, it is also possible to incorporate the variation in
consumption rates by allowing it to vary randomly (within the defined distribution) across the
exposed population.
To create a full distribution of consumption rates that is consistent with the EFH
recommendations, the consumption rates was assumed to be log-normally distributed across the
population with a mean of 8 g/day and a 95th percentile of 25 g/day; that is, the specified
distribution for the consumption rate was log-normal (8, 10.45). Lognormal distribution was
deemed to be appropriate because of its skewed shape and because it would not allow for fish
consumption rates in the negative range. Drawing randomly from this distribution, a
consumption rate was assigned to each of the roughly 165,000 block groups in the study area.
The fish consumption in each block group was assumed to be constant for the modeling
purposes. This approach was used to avoid the computational burden of assigning a random
consumption rate to a fractional person in a block group (the estimated number of exposed
persons in each block group is not an integer) and because the estimated number of exposed
persons per block group is relatively small. Although some degree of randomness is lost by
assuming a constant consumption rate in a block group, the inter-individual variability in
consumption rate is adequately represented because the average number of exposed individuals
in a block group is only 2.6 with standard deviation of 2.5. In effect, a sample of 165,000 people
is used to represent the consumption rate variability in approximately 435,000 people, which is
adequate for the modeling purposes. After randomly assigning consumption rates to the block
groups, the population centroid approach was reapplied using Eq. (10.16) with a variable
10-98
-------
consumption rate to calculate average daily mercury ingestion rates for the estimated exposed
population (NPA) in each block group.
Based on this adapted version of the population centre id approach, Figures 10-11 to 10-
18 show how the avoided IQ decrements (i.e., benefits) are distributed in the exposed population
under alternative emissions reductions scenarios. In all of these figures, the consumption rate is
allowed to vary across the exposed population as described above.
Figures 10-11 and 10-12 report results for the 2001 Utility Emissions Zero-Out scenario,
relative to the 2001 Base Case. Most of the prenatally exposed children avoid less than 0.025 IQ
point decrements; however, a significant number have avoided IQ decrements between 0.025
and 0.1 IQ points. About 1,300 prenatally exposed children have avoided IQ loss greater than
0.1 IQ points. Figure 10-12 shows the cumulative distribution of avoided IQ decrements across
the exposed population. Reduction in IQ decrements (i.e., benefits) due to the Zero-Out scenario
are less than 0.02 IQ points from the baseline levels for more than 90 percent of the prenatally
exposed children.
Figures 10-13 and 10-14 report results for the 2020 Utility Emissions Zero-Out scenario,
relative to the 2020 Base Case with CAIR. Again, a large majority of the prenatally exposed
children avoid less than 0.025 IQ point decrements. In this case, less than 100 prenatally
exposed children have avoided IQ loss greater than 0.1 IQ points. Figure 10-14 shows the
cumulative distribution of avoided IQ decrements across the exposed population. Reductions in
IQ decrements (i.e., benefits) due to the 2020 Zero-Out scenario are less than 0.01 IQ points
from the 2020 baseline levels for more than 90 percent of the prenatally exposed children.
Figures 10-15 through 10-18 report results for the CAMR Control Options relative to the
2020 Base Case with CAIR. In both cases 11 or less prenatally exposed children are estimated
to avoid IQ losses of more than 0.1 IQ points. Figures 10-16 and 10-18 show the cumulative
distributions of avoided IQ decrements for the two CAMR control options. In both cases,
reductions in IQ decrements (i.e., benefits) are less than 0.003 IQ points from the 2020 baseline
levels for more than 90 percent of the prenatally exposed children.
10-99
-------
| 400,000
| 350,000
c
•o 300,000
3
CO
o
3 r- O 0
- CM T- (D »-
in in
I-. CM
CO ^~
0 0
2 S
in ^~
« <=
o
IQ Points
CM ^
in
^
d
3
in
o
- 0 1-
in
CM
in
o
3
in
o
- 0 »
in
in
c!
3
in
in
o
-or
in
CM
CD
o
Q
CD
o
4 O C
in
-------
JS
ra
TO
T3
c
•a
oiov-if)CMiocoio^-iointf>
0.
1.000 0.001 0.002 0.003
0.004 0.005
IQ Points
0.006 0.007 0.008 0.009
Figure 10-14. Cumulative Distribution of Modelled Avoided IQ Decrements (Benefits) due
to Mercury Emissions Reductions: CAMR Control Option 1 Relative to 2020 Base Case
with CAIR; Population Centroid Approach; Variable Consumption Rate
10-101
-------
at
"m
3
|O
'•&
_C
•o
0>
1
X
Ul
"5
1
I
z
600,000-i;
i
'
500,000- '
400,000-
300,000-
200,000-
100,000-
1
(
(
(
<
»
NJ
O
—
?
CO
U) CM CM
mr^cMcM^rcMooooo^-ooooooooooooooo
ninmininiAinininininminin
Nf^cMf--r>jr»-CNt--cNri-CMi^cNh-
DOT-T-cMCMcoco^-^-inincoco
^docicidocDciooooo
oooooooooooooo
Din^-mcMincoin^ininmcoin
° ci ^ o ^ o « ci * o « c= ®
O 0 O O O 0 O
000
in u>
CM f^
1*- o
o „,
0 >
0 <
IQ Points
Figure 10-15. Distribution of Modelled Avoided IQ Decrements (Benefits) due to Mercury
Emissions Reductions: CAMR Control Option 2 Relative to 2020 Base Case with CAIR;
Population Centroid Approach; Variable Consumption Rate
01
CL
1.0
0.9
0.8-
0.7-
0.6-
0.5-
0.4-
0.3-
0.2-
0.1-
0.0
0.000
0.002
0.004
0.006
IQ Points
0.008
0.010
0.012
Figure 10-16. Cumulative Distribution of Modelled Avoided IQ Decrements (Benefits) due
to Mercury Emissions Reductions: CAMR Control Option 2 Relative to 2020 Base Case
with CAIR; Population Centroid Approach; Variable Consumption Rate
10-102
-------
euo.uuo-
J2 500,000-
ra
13
1 400.000-
g 300,000-
UJ 200,000-
*5
£ 100,000-
E
f\
u
s.
•^
•^
in T-
in in
CM l»-
0 0
D 0
S S
o m
0
ci
S ¥
in
CM
o
o
^_
o
o
•«- en
in
^~
O
o
in
ci
CM
'in
CM
CM
0
0
CM
0
T- O O
in
CM
0
0
in
CM
ci
o o
in
CM
CO
o
0
CO
ci
o
'in
f*H,
CO
0
o
in
CO
ci
IQ
0 * O
in
CM
0
0
ci
Points
O CM
in
^f
0
0
in
o
o o
in
CM
in
o
o
in
ci
o o
in
in
0
o
in
in
ci
o o
in
CM
-------
It is important to note that these results based on a variable consumption rate only
represent one realization of the random distribution offish consumption rates across the exposed
populations. The results would be somewhat different if the randomization process were
repeated; however, the summary results would most likely be very similar. In addition, the
results presented at the tail of this distribution are based on a mathematical calculation of the
results of the benefit modeling and are not based on any empirical evidence the Agency has that
people consume large amounts of recreationally caught freshwater fish (i.e., 2-3 meals per day)
such that they would incur this level of IQ loss.
10.6 Analysis of Potentially High-Risk Subpopulations
In addition to considering the full distribution of IQ benefits across the modeled study
population of prenatally exposed children it is also possible to consider distributional equity by
examining segments of the exposed population expected to experience disproportionately high
impacts (and hence benefits) due to (a) proximity to high methylmercury fish concentrations
and/or (b) relatively high fish consumption rates. In this analysis, to further examine this issue
of distributional equity, EPA examined several special sub-populations believed to be at elevated
risk of methylmercury fish exposure. The analyses conducted for these potential high-risk
subpopulations are described in this section. Consideration of IQ change and benefits estimates
generated for these special sub-populations can support a determination regarding the potential
for distributional equity playing a key factor in benefits for this regulation.
In this section of the report the high-risk populations of interest are groups of individuals
who are exposed to relatively high levels of mercury through consumption of self-caught
freshwater fish. In particular, they are groups with high rates offish consumption. The term
"subsistence fishing"25 may apply to the subpopulations analyzed in this section, depending on
how this term is interpreted. These subpopulations are not necessarily from households for
whom fishing is a primary activity for survival; however, their high rates of noncommercial
freshwater fish consumption suggest that it is an integral part of their diet.
Below high-risk subpopulations are addressed in three ways, each of which is an
adaptation of the population centroid approach. In the first case, fish consumption rates were
allowed to vary systematically across the exposed population. The analysis focused on mercury
ingestion by the those in the top fifth percentile of consumption rate distribution. In the second
case, the analysis focused on the portion of the exposed population with incomes below the
poverty threshold. Because we are lacking data on consumption rates by income categories, it
was assumed that a portion of these populations have noncommercial freshwater fish
consumption rates comparable to the population examined in the first case (top fifth percentile of
the total consumption rate distribution). In the third case, the analysis focused on two ethnic
subpopulations in the U.S. with high expected rates of freshwater fish consumption—the Hmong
and the Chippewa in Minnesota, Wisconsin, and Michigan. Using existing studies of these
populations' fishing and fish consumption behaviors, the population centroid approach was
adapted and applied to estimate mercury exposures for prenatally exposed children in these
ethnic groups.
25 "Subsistence fishing" is defined as the taking offish for the sustenance of families, communities, and cultures.
10-104
-------
10.6.1 Mercury Ingestion Estimates for Individuals in the Upper Range of the Fish
Consumption Distribution
In this section of the report, mercury ingestion levels for potentially high risk
subpopulations are analyzed by incorporating information about the variability in average daily
fish consumption rates across the recreational freshwater fishing population. According to
EPA's Exposure Factors Handbook (EFH) (EPA, 1997b), the recommended mean and 95th
percentile daily self-caught fish freshwater fish consumption rate (C) for freshwater recreational
anglers are 8 g/day and 25 g/day respectively. In this analysis of potentially high risk
individuals, the consumption rate is allowed to vary systematically across the population, using
the same method described in Section 10.4.
To create a full distribution of consumption rates that is consistent with the EFH
recommendations, the consumption rate was assumed to be log-normally distributed across the
population with a mean of 8 g/day and a 95th percentile of 25 g/day. Drawing randomly from
this distribution, a different consumption rate was assigned to each of the roughly 165,000 block
groups in the study area. The population centroid approach was then reapplied and combined
with Eq. (10.16) using a variable consumption rate to calculate average daily mercury ingestion
rates for the estimated exposed population (NPA) in each block group.
To specifically address potentially high risk groups, Tables 10-32 and 10-33 report
results only for exposed populations that were randomly assigned consumption rates higher than
95* percentile consumption rate of the recreational freshwater angler (i.e., above 25 g/day).
Table 10-32 reports detailed result for the 2001 Base Case. The annual number of prenatally
exposed children in this high risk group was estimated to be 22,300 and the mean Hgl across
block groups was 12.95 fig/day. Average IQ decrements in this group were estimated to be 0.33
points and the total present value of foregone earnings was estimated to be $65 million. As with
the estimates reported for the recreational freshwater angler in Section 10.1.4, these estimates are
based on observed mercury levels in freshwater fish (as summarized in Table 10-2) and therefore
include all sources of mercury freshwater fish.
Table 10-33 summarizes similar model results for the 2020 Base Case with CAIR. The
annual number of prenatally exposed children in this high risk group in 2020 was estimated to be
24,700 and the mean Hgl across block groups was 11.33 fxg/day, which is 12 percent lower than
in the 2001 Base Case. Average IQ decrements in this group were estimated to be 0.29 points
and the total present value of foregone earnings was estimated to be $63 million.
Table 10-34 reports estimates of beneficial changes to this subsistence population under
the five emissions control scenarios, including estimates for five lag periods and three assumed
discount rates. The 2001 Utility Emissions Zero-Out and the 2020 Base Case with CAIR are
both estimated to result in per capita IQ decrements that are on average between 0.042 and 0.047
less than under the 2001 Base Case. Compared to the 2020 Base Case with CAIR, the 2020
Utility Emissions Zero-Out is estimated to reduce per capita IQ decrements by an average of
10-105
-------
o
o
Table 10-32. Summary of Estimated Mercury Exposures for Consumption-Based Subsistence Population, with Associated IQ
Decrements and Foregone Earnings: Population Centroid Approach — 2001 Base Case8
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Annual Number
of Prenatally
Exposed Children
22,302
507
466
218
13
37
1,114
1,097
391
1,407
813
369
589
679
303
262
122
884
1,093
820
392
756
Average Daily Maternal
Ingestion of Mercury
(Hg/day/person)
Mean
12.95
10.56
15.86
17.94
8.39
10.71
19.49
14.03
8.02
8.70
11.55
17.87
10.87
13.81
19.83
7.71
25.80
12.39
8.87
12.51
12.37
14.76
S.D.
10.19
7.03
19.71
11.59
3.68
4.86
10.00
10.06
3.89
4.95
6.91
11.98
6.25
10.09
11.33
4.00
10.63
7.23
4.80
8.75
6.13
13.52
IQ Decrements in Prenatally Exposed
Children
Mean
0.331
0.270
0.406
0.459
0.215
0.274
0.499
0.359
0.205
0.223
0.295
0.457
0.278
0.353
0.507
0.197
0.660
0.317
0.227
0.320
0.316
0.378
S.D.
0.261
0.180
0.504
0.297
0.094
0.124
0.256
0.257
0.100
0.127
0.177
0.307
0.160
0.258
0.290
0.102
0.272
0.185
0.123
0.224
0.157
0.346
Total
7,390
137
189
100
3
10
555
394
80
313
240
169
164
240
154
52
80
280
248
262
124
285
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2001; 1999$)
Mean
$2,918
$2,379
$3,573
$4,042
$1,890
$2,413
$4,391
$3,161
$1,808
$1,961
$2,602
$4,027
$2,449
$3,111
$4,468
$1,736
$5,813
$2,791
$1,998
$2,819
$2,787
$3,327
S.D.
$2,297
$1,584
$4,441
$2,611
$829
$1,095
$2,253
$2,268
$877
$1,116
$1,556
$2,700
$1,408
$2,274
$2,553
$900
$2,395
$1,629
$1,083
$1,971
$1,382
$3,047
Total
$65,086,047
$1,206,122
$1,666,437
$882,544
$24,954
$90,402
$4,891,366
$3,469,580
$707,055
$2,759,572
$2,116,091
$1,487,071
$1,441,147
$2,111,880
$1,352,521
$455,573
$707,079
$2,467,205
$2,184,445
$2,311,088
$1,090,960
$2,514,047
(continued)
-------
o
o
Table 10-32. Summary of Estimated Mercury Exposures for Consumption-Based
Decrements and Foregone Earnings: Population Centroid Approach—2001 Base
Subsistence Population, with Associated IQ
Case (continued)
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Annual Number
of Prenatally
Exposed Children
127
246
63
365
874
1,150
759
775
30
465
124
686
2,586
668
50
812
192
Average Daily Maternal
Ingestion of Mercury
(jig/day/person)
Mean
11.00
8.68
19.21
15.70
15.92
14.63
12.98
18.10
18.00
16.90
9.60
12.95
10.53
10.02
17.22
10.50
13.21
S.D.
6.17
3.87
6.75
13.68
9.85
13.53
7.69
13.21
11.51
11.82
5.29
8.44
5.69
6.66
8.05
5.20
8.78
IQ Decrements in Prenatally Exposed
Children
Mean
0.281
0.222
0.492
0.402
0.407
0.374
0.332
0.463
0.460
0.432
0.246
0.331
0.269
0.256
0.441
0.269
0.338
S.D.
0.158
0.099
0.173
0.350
0.252
0.346
0.197
0.338
0.295
0.303
0.135
0.216
0.146
0.170
0.206
0.133
0.225
Total
36
55
31
146
356
430
252
359
14
201
31
227
696
171
22
218
65
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2001; 1999$)
Mean
$2,478
$1,956
$4,330
$3,538
$3,586
$3,297
$2,924
$4,078
$4,055
$3,808
$2,163
$2,917
$2,373
$2,257
$3,880
$2,365
$2,976
S.D.
$1,390
$872
$1,520
$3,082
$2,219
$3,049
$1,732
$2,976
$2,594
$2,664
$1,192
$1,901
$1,282
$1,501
$1,814
$1,171
$1,977
Total
$314,521
$480,417
$272,212
$1,289,682
$3,135,000
$3,790,446
$2,218,085
$3,158,572
$121,504
$1,771,388
$269,236
$2,002,465
$6,134,208
$1,507,902
$192,461
$1,920,550
$570,259
Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table.
-------
o
oo
Table 10-33. Summary of Estimated Mercury Exposures for Consumption-Based Subsistence Population, with Associated IQ
Decrements and Foregone Earnings: Population Centroid Approach — 2020 with CAIR"
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Annual Number
of Prenatally
Exposed Children
24,724
509
514
234
12
40
1,447
1,331
382
1,516
875
404
583
690
313
294
113
900
1,224
852
392
930
Average Daily Maternal
Ingestion of Mercury
(|ig/day/person)
Mean
11.33
9.05
14.22
15.59
6.83
9.44
18.03
11.72
7.44
7.74
9.66
17.44
8.51
12.61
17.74
6.33
23.73
10.52
8.39
11.89
11.15
12.02
S.D.
8.34
6.51
17.84
10.13
2.48
3.83
9.63
7.91
3.78
4.16
5.52
12.14
4.42
7.73
10.02
2.91
10.51
5.69
4.14
8.86
5.13
10.12
IQ Decrements in Prenatally Exposed
Children
Mean
0.290
0.232
0.364
0.399
0.175
0.242
0.461
0.300
0.190
0.198
0.247
0.446
0.218
0.323
0.454
0.162
0.607
0.269
0.215
0.304
0.285
0.308
S.D.
0.213
0.166
0.456
0.259
0.063
0.098
0.246
0.202
0.097
0.107
0.141
0.311
0.113
0.198
0.256
0.075
0.269
0.146
0.106
0.227
0.131
0.259
Total
7,165
118
187
93
2
10
668
399
73
300
216
180
127
223
142
48
69
242
263
259
112
286
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2020; 1999$)
Mean
$7,485
$6,141
$14,079
$6,079
$824
$4,025
$13,271
$14,174
$4,741
$5,374
$6,852
$15,136
$6,283
$9,201
$4,807
$2,268
$11,602
$4,774
$11,582
$10,289
$9,472
$9,228
S.D.
$10,476
$6,676
$18,853
$6,149
$636
$2,572
$18,187
$17,435
$5,054
$5,914
$7,988
$15,844
$5,304
$8,479
$3,952
$1,840
$10,809
$4,888
$10,530
$12,852
$7,507
$10,983
Total
$63,100,915
$1,037,813
$1,647,238
$820,718
$18,136
$84,529
$5,878,893
$3,515,077
$639,995
$2,644,042
$1,904,965
$1,589,252
$1,118,388
$1,959,919
$1,249,821
$419,500
$603,321
$2,134,093
$2,316,415
$2,284,126
$985,087
$2,519,356
(continued)
-------
o
o
Table 10-33. Summary of Estimated Mercury Exposures for Consumption-Based Subsistence Population, with Associated IQ
Decrements and Foregone Earnings: Population Centroid Approach—2020 with CAIR (continued)
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Annual Number
of Prenatally
Exposed Children
123
282
64
402
908
1,176
820
746
33
496
120
763
3,418
765
45
835
174
Average Daily Maternal
Ingestion of Mercury
Qig/day/person)
Mean
9.85
8.07
17.30
14.30
14.09
11.19
12.11
14.42
15.93
14.36
8.75
10.48
9.62
8.20
15.17
9.53
9.13
S.D.
5.41
3.61
5.13
11.13
8.81
11.31
6.61
9.76
8.03
9.95
4.38
7.36
5.48
5.08
5.87
4.26
5.05
IQ Decrements in Prenatally Exposed
Children
Mean
0.252
0.206
0.443
0.366
0.360
0.286
0.310
0.369
0.407
0.367
0.224
0.268
0.246
0.210
0.388
0.244
0.233
S.D.
0.138
0.092
0.131
0.285
0.225
0.289
0.169
0.250
0.206
0.254
0.112
0.188
0.140
0.130
0.150
0.109
0.129
Total
31
58
28
147
327
337
254
275
13
182
27
205
841
161
17
204
41
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2020; 1999$)
Mean
$8,054
$5,896
$7,138
$3,713
$3,723
$6,163
$13,318
$4,615
$3,679
$10,696
$6,765
$9,441
$10,962
$5,238
$5,454
$8,155
$4,714
S.D.
$5,855
$4,237
$3,622
$3,428
$4,724
$10,259
$10,817
$4,977
$3,200
$12,041
$3,700
$11,605
$15,938
$4,930
$2,936
$7,315
$2,898
Total
$273,830
$512,923
$249,817
$1,295,668
$2,881,885
$2,964,188
$2,237,462
$2,422,759
$117,744
$1,604,395
$236,769
$1,803,311
$7,410,196
$1,414,187
$152,718
$1,794,118
$358,264
Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table. For comparison purposes with the Base Cases in 2001, benefits presented in this table do not incorporate potential lags in fish tissue
response to a change in mercury deposition.
-------
Table 10-34. Summary of Annual Benefit Estimates for Consumption-Based Subsistence Population: Population Centroid
Approach
Central Estimate of Fish Tissue
Response Times
Annual Number of Prenatally Exposed Children
2001 Base Case
2020 Base Case (with CAIR)
Per Capita Avoided IQ Decrements
2001 Utility Emissions Zero-Out (Relative to 2001 Base Case)
2020 Base Case with CAIR (Relative to 2001 Base Case applied to 2020 demographics)
2020 Utility Emissions Zero-Out (Relative to 2020 Base Case with CAIR)
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
Total Value of Benefits (1999$s)
2001 Utility Emissions Zero-Out (Relative to 2001 Base Case)
1% Discount Rate; Present Value in 2001
3% Discount Rate; Present Value in 2001
7% Discount Rate; Present Value in 2001
2020 Base Case with CAIR (Relative to 2001 Base Case applied to 2020 demographics)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 Utility Emissions Zero-Out (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
BENEFITS OF CONTROL OPTIONS
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate- Present Value in 7(\">C\
10-YrLag
23,232
27,126
0.0439
0.0462
0.0152
0.0032
0.0048
N/A
$6,678,524
$4,562,627
N/A
$7,036,390
$4,807,114
N/A
$2,705,281
$1,848,191
N/A
$573,373
$391,716
N/A
$850,046
$580. 7H
20-Yr Lag
24,955
29,661
0.0432
0.0469
0.0151
0.0032
0.0048
N/A
$5,255,162
$2,452,760
N/A
$5,701,309
$2,660,991
N/A
$2,185,277
$1,019,942
N/A
$463,559
$216,358
N/A
$689,683
$"?21.898
Alternative Estimate of Fish Tissue
Response Times
5-Yr Lag
22,747
25,859
0.0442
0.0451
0.0153
0.0032
0.0048
N/A
$7,636,211
$6,311,683
N/A
$7,749,431
$4,807,114
N/A
$3,001,948
$2,481,249
N/A
$635,939
$525,633
N/A
$940,902
$777699
50-Yr Lag
32,450
37,266
0.0423
0.0449
0.0149
0.0032
0.0047
$7,358,513
$2,760,562
$410,837
$7,794,954
$2,924,924
$435,204
$2,967,472
$1,113,254
$165,679
$630,695
$236,607
$35,213
$945,748
$354,799
S'i2 8m
-------
0.015 points. CAMR Options 1 and 2, by comparison, are estimated to reduce them by an
average of 0.003 and 0.005 points respectively.
Under CAMR Option 1, the aggregate benefits for this subsistence population assuming
10 and 20 year lag periods are estimated to be $0.6 million and $0.5 million, respectively, for
those prenatally exposed in 2020. These estimates are 21 percent as large as those for the 2020
Utility Emissions Zero-Out. For CAMR Option 2, these values are estimated to both be $0.9
million and 31 percent as large as corresponding estimates for the 2020 Utility Emissions Zero-
Out.
10.6.2 Mercury Ingestion Estimates for Individuals in Low Income, High Fish Consumption
Households
Another potentially high-risk subpopulation in the U.S. is comprised of individuals in
low income households who rely on self-caught freshwater fish as part of their diet (EPA,
2002b). Although studies have documented these types of behaviors among low income groups
in specific locations in the U.S. (West, 1992; Belton, Roundy, and Weinstein, 1986), a broad
definition of this behavior at the national level is not available.
To assess potentially high exposures among low income subpopulations, the population
centroid approach was adapted to focus specifically on low-income populations. Mercury
ingestion levels were estimated for this subpopulation under a defined set of assumptions. First,
Census data were used to restrict the analysis to women of childbearing age in households with
incomes below $10,000.26 That is, NF in Eq. (10.2) (i.e., number of females ages 15-44) was
redefined to include only low income women of childbearing age from each block group. State-
level fertility rate data were again used to estimate the number of low-income pregnant women
in each block group.
Second, to estimate the portion of these women that reside in angler households, results
from the NSFHWR (Pullis, 2000) were used, which found that angler participation rates among
low income individuals are on average 35 percent lower than for the general population in the
U.S. The state-level participation rate estimates were scaled down (NA,/NS in Eq. [10.3]) for the
general population by 35 percent to estimate the annual number of prenatally exposed children in
low-income angler households in each block group.
Third, it was assumed that 50 percent of these prenatally exposed children are in
households that rely on self-caught freshwater fish. This assumption was based in part on data
from NSRE 1994, which found that roughly half of the single-day freshwater fishing trips by low
income anglers were to waterbodies within 20 miles of their residence. Therefore, it was
assumed that low income individuals making single day trips to waterbodies within 20 miles
represent subsistence-type (rather than strictly recreational) fishing behaviors.
Fourth, to match these individuals in low-income, high fish consumption households with
mercury concentrations, average mercury concentrations within 20 miles of each block group
26 Poverty level for a 2-3 person household in 2000 was $11,239-513,738
(www.census.gov/hhes/poverty/threshld/threshOO.html)
10-111
-------
were calculated. To weight lake and river concentrations in this calculation, the same proportion
of lake and river fishing (c,, ck in Eqs. [10.4] and [10.5]) was assumed for low income
households as for the general population in each block group.
Fifth, to estimate mercury ingestion rates for these subpopulations, the same model
described in Eq. (10.1) was applied; however, the daily fish consumption rate (C) was allowed to
vary randomly. To create a full distribution of consumption rates that is consistent with the
assumption that subsistence fishers consume more than the 95th percentile amount offish,
consumption rates were randomly drawn from the tail beyond the 95th percentile of the log-
normal distribution defined in Section 10.5. Then, a different sampled consumption rate was
randomly assigned to each of the roughly 165,000 block groups in the study area. The
population centroid approach was then reapplied.
Table 10-35 reports detailed result for the 2001 Base Case. The annual number of
prenatally exposed children in this high risk group was estimated to be 22,400 and the mean Hgl
across block groups was 12.44 jig/day27. Average IQ decrements in this group were estimated to
be 0.32 points and the total present value of foregone earnings was estimated to be $63 million.
Table 10-36 summarizes similar model results for the 2020 Base Case with CAIR. The
annual number of prenatally exposed children in this high risk group in 2020 was estimated to be
24,100 and the mean Hgl across block groups was 10.96 jig/day, which is 12 percent lower than
in the 2001 Base Case. Average IQ decrements in this group were estimated to be 0.28 points
and the total present value of foregone earnings was estimated to be $60 million.
Table 10-37 reports estimates of beneficial changes to this income-based subsistence
population under the five emissions control scenarios, including estimates for five lag periods
and three assumed discount rates. The 2001 Utility Emissions Zero-Out and the 2020 Base Case
with CAIR are both estimated to result in per capita IQ decrements that are on average between
0.035 and 0.042 less than under the 2001 Base Case. Compared to the 2020 Base Case with
CAIR, the 2020 Utility Emissions Zero-Out is estimated to reduce per capita IQ decrements by
an average of 0.015 points. CAMR Options 1 and 2, by comparison, are estimated to reduce
them by an average of 0.003 and 0.005 points respectively.
Under CAMR Option 1, the aggregate benefits for this subsistence population assuming
10 and 20 year lag periods are estimated to be $0.6 million and $0.5 million, respectively, for
those prenatally exposed in 2020. These estimates are 21 percent as large as the corresponding
estimates for the 2020 Utility Emissions Zero-Out. For CAMR Option 2, these values are
estimated to be $0.8 million and $0.7 million, respectively, which are 31 percent as large as the
estimates for the 2020 Utility Emissions Zero-Out.
21 The average daily mercury ingestion rate given here of 12.44 ug/day from freshwater fish is two times the EPA's
Reference Dose (RfD) for mercury of 5.8 ug/day. This estimate does not account for total exposure from
consumption from other fish sources. See Section 11 of this report for a detailed discussion of the implication of the
RfD on this analysis.
10-112
-------
10.6.3 Mercury Ingestion Estimates for Two Selected Ethnic Populations
To further analyze mercury ingestion levels among potential high risk subpopulations, a
separate analysis was conducted focusing on two selected special populations that are primarily
located within a subregion of the study area:
1. Hmong in Minnesota and Wisconsin; and
2. Chippewa (Ojibwa) in Minnesota, Wisconsin, and Michigan.
These groups were selected for the following reasons. First, among ethnic groups in the
United States, Southeast Asians and Native Americans have traditionally had relative high rates
offish consumption (EPA, 1997c). The Hmong are Southeast Asians of primarily Laotian
origin, and the Chippewa are a Native American tribe from the Great Lakes area. As discussed in
more detail below, studies investigating the fishing behaviors of these two groups have found
relatively high rates of participation in freshwater fishing activities and of freshwater fish
consumption.
Second, large portions of both the Hmong and Chippewa populations are located in three
states within our 37-state study area. Other than California, the Hmong population in the United
States has primarily settled in Minnesota and Wisconsin, where they now number over 75,000.
The Chippewa are among the five most populous tribes in the United States, numbering over
100,000 in 2000 and are primarily located in Minnesota, Wisconsin, and Michigan. The
Chippewa account for a large majority of the Native American population in the study areas of
this benefit analysis, as is evident by Figure 10-19, which displays U.S. Bureau defined Native
American tracts.
Third, due to the availability of Census data and fishing behavior studies for the Hmong
and Chippewa, the exposure assessment methods developed for the recreational freshwater
angler can be adapted for these two subpopulations. In particular, the population centroid
approach described in the previous sections can be adapted to include some of the specific
conditions of these special populations
10-113
-------
Source: US Census Bureau "KM.
Figure 10-19. U.S. Census Tracts with Native American Populations
10-114
-------
Table 10-35. Summary of Estimated Mercury Exposures for Income-Based Subsistence Population, with Associated
Decrements and Foregone Earnings: Population Centroid Approach — 2001 Base Case"
IQ
Average Daily Maternal
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Annual Number
of Prenatally
Exposed Children
22,393
732
663
158
20
31
1,059
1,027
348
1,222
565
360
716
855
259
199
119
750
742
849
651
733
Ingestion
of Mercury
IQ Decrements in Prenatally
(lag/day/person)
Mean
12.44
10.00
12.96
22.21
5.04
8.84
19.64
13.86
6.21
6.33
9.04
16.98
11.46
13.07
21.59
6.53
28.78
12.09
7.60
10.68
13.29
14.32
S.D.
12.67
9.15
9.75
18.23
2.77
5.53
13.16
10.76
3.52
4.78
6.42
14.12
7.86
13.05
14.11
4.72
15.65
8.07
5.36
9.73
8.60
9.29
Mean
0.318
0.256
0.332
0.568
0.129
0.226
0.502
0.355
0.159
0.162
0.231
0.434
0.293
0.334
0.552
0.167
0.736
0.309
0.194
0.273
0.340
0.366
Children
S.D.
0.324
0.234
0.249
0.466
0.071
0.142
0.337
0.275
0.090
0.122
0.164
0.361
0.201
0.334
0.361
0.121
0.401
0.207
0.137
0.249
0.220
0.238
Exposed
Present Value
of Foregone
due to IQ Decrements (in
Total
7,126
187
220
90
3
7
532
364
55
198
131
156
210
286
143
33
88
232
144
232
221
269
Mean
$2,803
$2,254
$2,921
$5,004
$1,135
$1,993
$4,425
$3,122
$1,398
$1,426
$2,037
$3,825
$2,583
$2,946
$4,865
$1,472
$6,485
$2,725
$1,712
$2,407
$2,996
$3,227
S.D.
$2,803
$2,061
$2,197
$4,108
$625
$1,247
$2,965
$2,424
$794
$1,076
$1,448
$3,183
$1,771
$2,941
$3,179
$1,063
$3,528
$1,819
$1,207
$2,194
$1,938
$2,093
Net Earnings
2001; 1999$)
Total
$62,756,741
$1,649,964
$1,935,007
$792,822
$22,162
$62,171
$4,686,540
$3,207,259
$486,319
$1,743,592
$1,150,670
$1,378,019
$1,849,315
$2,517,166
$1,261,763
$292,363
$771,948
$2,043,439
$1,270,006
$2,042,312
$1,950,276
$2,365,451
(continued)
-------
Table 10-35. Summary of Estimated Mercury Exposures for Income-Based Subsistence Population, with Associated IQ
Decrements and Foregone Earnings; Population Centroid Approach—2001 Base Case (continued)
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TO
TX
VA
VT
WI
WV
Annual Number
of Prenatally -
Exposed Children
119
222
58
254
1,066
1,029
812
771
43
539
132
792
3,101
455
52
599
292
Average Daily Maternal
Ingestion of Mercury
(jig/day/person)
Mean
12.52
7.33
24.69
15.54
15.77
13.28
12.06
19.57
21.14
16.90
10.05
11.47
9.69
9.28
18.47
10.26
14.53
S.D.
12.12
5.16
17.20
11.42
12.04
10.01
8.15
27.90
11.23
14.54
7.39
8.36
7.08
7.86
11.99
7.53
9.51
IQ Decrements in Prenatally Exposed
Children
Mean
0.320
0.188
0.632
0.398
0.404
0.340
0.309
0.501
0.541
0.432
0.257
0.294
0.248
0.238
0.473
0.263
0.372
S.D.
0.310
0.132
0.440
0.292
0.308
0.256
0.209
0.714
0.287
0.372
0.189
0.214
0.181
0.201
0.307
0.193
0.243
Total
38
42
37
101
430
350
250
386
23
233
34
232
769
108
25
157
109
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2001; 1999$)
Mean
$2,821
$1,653
$5,563
$3,501
$3,554
$2,993
$2,717
$4,409
$4,764
$3,807
$2,264
$2,585
$2,184
$2,092
$4,162
$2,313
$3,275
S.D.
$2,731
$1,162
$3,876
$2,573
$2,712
$2,256
$1,837
$6,288
$2,530
$3,277
$1,665
$1,883
$1,596
$1,772
$2,703
$1,698
$2,144
Total
$336,550
$366,374
$325,028
$888,373
$3,790,552
$3,079,702
$2,205,353
$3,398,897
$206,753
$2,052,104
$298,184
$2,047,437
$6,771,389
$951,503
$218,402
$1,384,873
$956,703
Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table. For comparison purposes with the Base Cases in 2020, benefits presented in this table do not incorporate potential lags in fish tissue
response to a change in mercury deposition.
-------
Table 10-36. Summary of Estimated Mercury Exposures for Income-Based Subsistence Population, with Associated IQ
Decrements and Foregone Earnings: Population Centroid Approach — 2020 with CAIR"
Study Area
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KS
KY
LA
MA
MD
ME
MI
MN
MO
MS
NC
Annual Number
of Prenatally
Exposed Children
24,132
707
711
171
17
34
1,342
1,147
340
1,284
598
389
704
842
268
199
106
753
790
871
634
862
Average Daily Maternal
Ingestion of Mercury
(jig/day/person)
Mean
10.96
8.51
11.95
19.42
3.17
7.58
18.37
11.72
6.00
5.74
7.60
16.56
8.72
12.18
19.24
5.21
26.54
10.18
7.90
10.27
12.17
11.38
S.D.
9.58
9.04
8.37
19.97
1.47
4.70
11.62
9.45
3.19
3.64
5.09
12.77
6.09
11.74
12.96
4.14
14.04
6.79
5.64
9.50
7.57
7.66
IQ Decrements in Prenatally Exposed
Children
Mean
0.280
0.218
0.306
0.497
0.081
0.194
0.470
0.300
0.154
0.147
0.195
0.424
0.223
0.312
0.492
0.133
0.679
0.260
0.202
0.263
0.311
0.291
S.D.
0.245
0.231
0.214
0.511
0.038
0.120
0.297
0.242
0.082
0.093
0.130
0.327
0.156
0.300
0.332
0.106
0.359
0.174
0.144
0.243
0.194
0.196
Total
6,769
154
217
85
1
7
631
344
52
188
116
165
157
262
132
27
72
196
160
229
197
251
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2020; 1999$)
Mean
$2,470
$1,918
$2,693
$4,376
$715
$1,708
$4,139
$2,642
$1,352
$1,292
$1,714
$3,733
$1,964
$2,745
$4,335
$1,174
$5,980
$2,294
$1,780
$2,313
$2,743
$2,563
S.D.
$2,159
$2,037
$1,886
$4,500
$331
$1,059
$2,618
$2,130
$720
$821
$1,147
$2,878
$1,372
$2,645
$2,921
$934
$3,163
$1,531
$1,270
$2,140
$1,706
$1,727
Total
$59,613,908
$1,356,146
$1,914,546
$747,241
$12,454
$57,258
$5,553,735
$3,029,686
$460,380
$1,659,781
$1,024,603
$1,451,555
$1,382,855
$2,311,469
$1,163,445
$233,479
$631,624
$1,727,063
$1,405,268
$2,015,424
$1,739,298
$2,209,022
(continued)
-------
oo
Table 10-36. Summary of Estimated Mercury Exposures for Income-Based Subsistence Population, with Associated IQ
Decrements and Foregone Earnings: Population Centroid Approach—2020 with CAIR (continued)
State
ND
NE
NH
NJ
NY
OH
OK
PA
RI
SC
SD
TN
TX
VA
VT
WI
WV
Annual Number
of Prenatally
Exposed Children
115
249
59
278
1,116
1,026
869
727
47
563
135
836
3,968
462
47
609
258
Average Daily Maternal
Ingestion of Mercury
(u.g/day/person)
Mean
11.60
6.95
22.42
15.33
15.09
9.57
11.44
14.99
18.67
14.35
9.53
9.25
8.98
7.59
16.09
9.38
9.32
S.D.
10.78
4.58
15.67
10.79
10.47
6.77
7.66
19.86
9.39
11.49
8.28
6.58
6.45
6.91
9.65
6.03
9.88
IQ Decrements in Prenatally Exposed
Children
Mean
0.297
0.178
0.574
0.392
0.386
0.245
0.293
0.383
0.478
0.367
0.244
0.237
0.230
0.194
0.412
0.240
0.239
S.D.
0.276
0.117
0.401
0.276
0.268
0.173
0.196
0.508
0.240
0.294
0.212
0.168
0.165
0.177
0.247
0.154
0.253
Total
34
44
34
109
431
251
254
279
23
207
33
198
912
90
19
146
62
Present Value of Foregone Net Earnings
due to IQ Decrements (in 2020; 1999$)
Mean
$2,613
$1,566
$5,052
$3,455
$3,402
$2,157
$2,578
$3,377
$4,208
$3,233
$2,147
$2,085
$2,024
$1,711
$3,625
$2,113
$2,101
S.D.
$2,429
$1,032
$3,530
$2,432
$2,359
$1,525
$1,726
$4,474
$2,115
$2,589
$1,865
$1,484
$1,453
$1,557
$2,174
$1,359
$2,227
Total
$300,095
$389,257
$300,514
$961,970
$3,796,552
$2,211,994
$2,240,049
$2,453,711
$198,873
$1,819,762
$288,880
$1,742,925
$8,033,117
$791,208
$169,340
$1,287,089
$542,241
Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table. For comparison purposes with the Base Cases in 2001, benefits presented in this table do not incorporate potential lags in fish tissue
response to a change in mercury deposition.
-------
Table 10-37. Summary of Annual Benefit Estimates for Income-Based Subsistence Population: Population Centroid Approach
Central Estimate of Fish
Tissue Response Times
Alternative Estimate of Fish
Tissue Response Times
Annual Number of Prenatally Exposed Children
2001 Base Case
2020 Base Case (with CAIR)
Per Capita Avoided IQ Decrements
2001 Utility Emissions Zero-Out (Relative to 2001 Base Case)
2020 Base Case with CAIR (Relative to 2001 Base Case applied to 2020 demographics)
2020 Utility Emissions Zero-Out (Relative to 2020 Base Case with CAIR)
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
Total Value of Benefits (1999$s)
2001 Utility Emissions Zero-Out (Relative to 2001 Base Case)
1% Discount Rate; Present Value in 2001
3% Discount Rate; Present Value in 2001
7% Discount Rate; Present Value in 2001
2020 Base Case with CAIR (Relative to 2001 Base Case applied to 2020 demographics)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 Utility Emissions Zero-Out (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
BENEFITS OF CONTROL OPTIONS
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate- Present Value in 9020
10-Yr Lag
23,037
27,126
0.042
0.035
0.015
0.003
0.005
N/A
$6,379,897
$4,358,611
N/A
$6,139,097
$4,194,102
N/A
$2,717,338
$1,856,428
N/A
$572,354
$391,020
N/A
$840,630
SS74101
20-Yr Lag
24,310
29,661
0.041
0.034
0.015
0.003
0.005
N/A
$4,905,014
$2,289,334
N/A
$4,905,679
$2,289,645
N/A
$2,165,647
$1,010,780
N/A
$454,554
$212,155
N/A
$670,774
Sin 07"!
5-Yr Lag
22,699
25,859
0.043
0.035
0.015
0.003
0.005
N/A
$7,362,516
$6,085,461
N/A
$6,853,900
$5,665,066
N/A
$3,038,206
$2,511,218
N/A
$641,184
$529,968
N/A
$939,259
$776 141
50- Yr Lag
30,548
37,266
0.040
0.033
0.014
0.003
0.004
$6,556,561
$2,459,708
$366,063
$6,499,636
$2,438,352
$362,885
$2,851,684
$1,069,816
$159,214
$593,633
$222,703
$33,143
$885,766
$332,297
$49 4S4
-------
10,6.4 Adaptation of the Population Centroid Approach to Estimate Exposed Hmong and
Chippewa Population
The first step in applying the population centroid approach is to estimate the size and
location of the relevant populations. The approach was adapted to address limitations in the
Census data for analyzing specific ethnic subpopulations.
One of the limitations of Census data is that demographic data for specific racial groups,
such as Hmong and Chippewa, are not available at a block group level. However, they are
available at a Census tract level and at higher levels of spatial aggregation (e.g., county). To
apply the population centroid model, tract-level data were first used to approximate the number
of Hmong and Chippewa females aged 15 to 44 in each block group. First, for each tract, the
percentage of its total population residing in each block group was calculated. These same
percentages were then applied to divide the Hmong and Chippewa populations into each block
group. In effect, it was assumed that distribution of the Hmong and Chippewa populations
across block groups in a tract is the same as the distribution of the total population in the tract.
Another limitation of Census data is that, if a specific racial group numbers less than 100
in a tract, the population size for this group is not reported in Census tract-level data. Because
this analysis does not include these unreported populations, it underestimate the total size of the
relevant Hmong and Chippewa populations.
Table 10-38 summarizes the resulting block group population estimates in 2001 for
females aged 15 to 44. State-level population growth rate projections to update Census 2000
data to 2001. Using tract-level data, Hmong populations could be estimated for roughly 7
percent of the Census block groups in Minnesota and about 9 percent in Wisconsin. Chippewas
populations could be estimated in Michigan and Wisconsin for about 1 percent of block groups
and in Minnesota for 4 percent. The estimated number of females of childbearing age in each
block group ranged from 0 to 184 for the Hmong and from 0 to 662 for the Chippewa.
As shown in Table 10-38, the total number of Hmong women of childbearing age in the
two states is estimated to be roughly 13,300, which is about 16 percent of the total Hmong
population. Since women of childbearing age typically represent about 20 percent of the overall
population, it is expected that the analysis based on tract level Census data underestimated the
exposed Hmong population by about 20 percent. For Chippewa, a similar calculation suggests
10-120
-------
Table 10-38. Block Group Demographics for Hmong and Chippewa Females, Aged 15 to 44 (in 2001)
State
MI
MN
WI
Total
Total
Hmong
Population
—
45,530
36,809
82,339
Total
Chippewa
Population
32,267
39,910
16,560
88,737
Total
Number
of Block
Groups
8,450
4,082
4,388
16,920
Number of Block Groups
With Hmong
Populations'1
—
279
398
677
With
Chippewa
Populations'*
103
166
48
317
Hmong Female Population,
Aged 15-44, per Block
Group"1'
Mean S.D.
— —
28.56 29.66
13.29 13.45
19.58 22.91
Total
—
7,969
5,290
13,259
Chippewa Female Population,
Aged 15-44, per Block
Groupa'b
Mean S.D.
23.33 29.84
29.44 69.72
37.73 56.48
28.71 57.64
Total
2,403
4,888
1,811
9,101
o
H«*
K)
"For Census tracts with Hmong or Chippewa populations of less than 100, population is extrapolated from county-level data.
Estimate extrapolated from Census tract-level data.
-------
that the exposed Chippewa population was underestimated by about 50 percent. The uncounted
populations are presumably in tracts with less than 100 Hmong or Chippewa.
The second step is to estimate the number of pregnant women by Census block group for
each of the two special populations (denoted by subscript T). For the recreational freshwater
angler analysis, state-level fertility rate data were used to estimate numbers of pregnant women.
Fertility rates for broad ethnic groups are not available at a state level, but they are available at a
national level (Hamilton, Sutton, and Ventura, 2003). Therefore, state-level fertility rates for the
general population were adjusted by the ratio of (1) national-level fertility rates for each ethnic
group to (2) national-level fertility rate for the general population. For the two special
populations, the number of pregnant women in each Census block was estimated as follows.
Number of pregnant women (NPj) = NFT *fs * (fm /%) (Eq. 10.19)
where
NFT = number of females aged 15 to 44 (Census) for the given special population,
fs = state-level fertility rate for the general population (live births per 1,000 women
aged 15 to 44),
fN = national fertility rate for the general population (live births per 1,000 women
aged 15 to 44), and
fm = national fertility rate for Asians/Pacific Islander (including Hmong) or
American Indians (including Chippewa) (live births per 1,000 women aged 15
to 44).
For Minnesota, Wisconsin, and Michigan, the state-level fertility rates for the general population
in 2001 (f.) were 61.8, 59.6, and 62 live births respectively per 1,000 women aged 15 to 44. The
national fertility rate for the general population (fN) was 65.3 births per 1,000. For
Asians/Pacific Islanders and for American Indians, the national fertility rates were 64.2 and 58.1
respectively.
In the third step, the number of pregnant women in each block group who live in
households with freshwater anglers was estimated. For both groups, the size of this exposed
population of interest was estimated as:
NPAT = NPT * AHT (Eq. 10.20)
where AHT = percentage of households with freshwater anglers.
For the Hmong, results from a study by Hutchison and Kraft (1994) were used to
estimate AHT. This study examined a random sample of 125 Hmong households from Green
Bay, Wisconsin, and collected data on fishing frequency, fish consumption frequency, fishing
travel distances, types offish caught and consumed, and other related behaviors. In their sample,
57.6 percent of the households were freshwater anglers. For this analysis, it was assumed that
this same percentage applies to all Hmong households in Minnesota and Wisconsin.
For the Chippewa, this third step (estimation of AHT and NPAT) was not applied
because, as discussed below, the freshwater fish consumption rate estimate available for the
10-122
-------
Chippewa is not restricted to only angler households. Therefore, to maintain consistency with
this consumption rate estimate, the exposed Chippewa population was not restricted to only
include those in angler households.
After estimating the exposed populations of interest, the population centroid approach
was applied to estimate the average mercury concentration in consumed freshwater fish and the
rate of mercury ingestion for the exposed population so interest.
To estimate the percentage of trips to different distance categories for the Hmong, results
from Hutchison and Kraft (1994) were again used. In their study, the maximum freshwater
fishing trip distance for the Hmong was 48 miles; thus, it was assumed that all trips by Hmong
anglers are within 50 miles of their residence. Within the 50-mile radius, trips were allocated to
the three distance categories in the same proportion (23:17:23) as for the recreational freshwater
angler. Therefore, the percentage of Hmong freshwater fishing trips in the three distance
categories of 0-10 miles, >10-20 miles, and >20-50 miles were assumed to be 36.5 percent,
27 percent, and 36.5 percent, respectively.
No specific data on average distance traveled to fish are available for the Chippewa.
However, information in Peterson et al. (1994) suggests that most fish is locally caught.
Therefore, the same trip distance assumptions that were applied to the Hmong were also applied
to the Chippewa. In other words, it was assumed that all fishing trips are within 50 miles and
that, within this radius, they are distributed in the same proportion as for the recreational
freshwater angler.
To further adapt the model for the two special populations, daily consumption rate (C)
estimates based on specific data for these populations were included in the analysis. For the
Hmong, a consumption rate of 21.2 g/day was included, which was calculated using the
consumption frequency data reported by Hutchison and Kraft (1994). Using their summary data,
the average number offish meals per year (34.1) for anglers in the sample was estimated. This
estimate was then multiplied by the assumed average fish meal size of 8 oz/meal (227 g/meal)
reported in Hutchison and Kraft and divided by 365 to estimate the average daily fish
consumption rate.
For the Chippewa, a daily consumption rate of 20 g/day was used, based on
recommendations from EPA's Exposure Factor Handbook (EPA, 1997b). EPA's analysis used
data from Peterson et al. (1994) to estimate a mean freshwater fish consumption rate (1.2 meals
per week) and Pao et al. (1982) to estimate the average weight of a fish meal (117 g/meal). The
Peterson et al. (1994) study specifically examined the fish consumption habits and blood
mercury levels of Chippewa Indians from northern Wisconsin, using a random sample of 175
and a nonrandom sample of 152 tribal members. This daily consumption rate is based on a
general sample of Chippewa adults, not only on a sample of those who fish or who consume
freshwater fish. Therefore, this rate can reasonably be applied to the total estimated exposed
population of Chippewa, rather than to a subset from angler households.
Summary and Discussion of Results. Table 10-39 summarizes estimates of the size of
exposed population of interest. In 2001 the mean population of prenatally exposed children per
block group was estimated to be 0.12 and 0.08, and the total exposed population estimate was
10-123
-------
553 and 10,947 across the study area for the Hmong and Chippewa respectively. By 2021 (after
a 20 year lag), the Hmong population of prenatally exposed children is expected almost double,
and the exposed Chippewa population was projected to grow by over 60 percent.
The estimated average daily maternal ingestion rates (Hgl) across the states in 2001 are
reported in Table 10-40. For the Hmong, Hgl was estimated to be 4.46 j^g/day and per capita IQ
decrements were estimated to be 0.11 points. For the Chippewa, the Hgl estimate is 5.2 fig/day
and the per capita IQ decrements were estimated to be 0.13 points. Under baseline conditions in
2001, the present value of total IQ related losses for the Hmong was estimated to be $0.6 million.
For the Chippewa, it was estimated to be $1.3 million.
Table 10-41 reports comparable results for the 2020 Based with CAIR. For the Hmong,
Hgl was estimated to be 4.46 u,g/day and per capita IQ decrements were estimated to be 0.11
points for the two states (MN and WI) combined. These estimates are the same as for the 2001
Base Case, but they reflect higher estimated average mercury concentrations for block groups in
MN (relative to the 2001 baseline) and lower concentrations in WI. For the Chippewa, the Hgl
estimate is 4.7 u,g/day and the per capita IQ decrements were estimated to be 0.12 points. Under
baseline conditions in 2001, the present value of total IQ related losses for the Hmong was
estimated to be $1 million. For the Chippewa, it was estimated to be $1.9 million. Both of these
estimates are higher than for the 2001 Base Case due to the relatively large increase in exposed
populations from 2001 to 2020 (as reported in Table 10-39)
Tables 10-42 and 10-43 report estimates of beneficial changes to the two subpopulations
under the five emissions control scenarios, including estimates for five lag periods and three
assumed discount rates. For the Hmong, the 2001 Utility Emissions Zero-Out is estimated to
result in changes in per capita IQ decrements that range from 0.0007 less to 0.0012 more than
under the 2001 Base Case, and the 2020 Base Case with CAIR is estimated to result in per capita
IQ decrements that are between 0.0069 and 0.0078 less. Compared to the 2020 Base Case with
CAIR, the 2020 Utility Emissions Zero-Out is estimated to reduce per capita IQ decrements by
an average of 0.0075 points. CAMR Options 1 and 2, by comparison, are estimated to reduce
them by an average of 0.0004 and 0.0014 points respectively.
For the Chippewa, the 2001 Utility Emissions Zero-Out is estimated to result in per
capita IQ decrements that are approximately 0.023 less than under the 2001 Base Case, and the
2020 Base Case with CAIR is estimated to result in per capita IQ decrements that are 0.015 less.
Compared to the 2020 Base Case with CAIR, the 2020 Utility Emissions Zero-Out is estimated
to reduce per capita IQ decrements by an average of 0.013 points. CAMR Options 1 and 2, by
comparison, are estimated to reduce them by an average of 0.001 and 0.002 points respectively.
10-124
-------
Table 10-39. Estimated Annual Number of Prenatally Exposed Children from Special Populations for Selected Lag Periods:
Population Centroid Approach
LAG Periods from 2001
Base Year of Comparison
(2001)
Population
State
Hmong
MN
WI
2-State
Total
Chippewa
MI
MN
WI
3-State
Total
Per Block Group
Mean
0.136
0.102
0.119
0.051
0.172
0.061
0.079
S.D.
0.449
0.206
0.352
0.203
0.912
0.395
0.487
Total
Exposed
Population
321
232
553
400
506
189
1,094
5 Year Lag (2006)
Per Block Group
Mean S.D.
0.165 0.544
0.124 0.255
0.145 0.428
0.058 0.221
0.208 1.099
0.075 0.484
0.094 0.584
Total
Exposed
Population
390
282
672
451
613
229
1,293
10 Year Lag (2011)
Per Block
Group
Mean S.D.
0.192 0.636
0.147 0.305
0.170 0.502
0.063 0.241
0.235 1.226
0.085 0.547
0.105 0.651
Total
Exposed
Population
452
333
786
495
693
260
1,448
20 Year Lag (2021)
Per Block
Group
Mean S.D.
0.238 0.796
0.195 0.416
0.217 0.639
0.074 0.265
0.293 1.474
0.107 0.701
0.128 0.787
Total
Exposed
Population
561
445
1,006
575
862
330
1,767
50 Year Lag (2051)
Per Block
Group
Mean S.D.
0.422 1.357
0.421 0.996
0.421 1.193
0.136 0.490
0.607 3.060
0.230 1.693
0.257 1.674
Total Exposed
Population
995
958
1,953
1,064
1,788
706
3,559
LAG Periods from 2020
Base Year of Comparison
Population
State
Hmong
MN
WI
2-State
Total
Chippewa
MI
MN
WI
3-State
Tntal
(2020)
Per Block Group
Mean
0.269
0.180
0.226
0.073
0.310
0.091
0.128
S.D.
0.899
0.396
0.700
0.247
1.794
0.587
0.897
5 Year Lag (2025)
Total
Exposed
Population
634
410
1,045
569
914
280
1,763
Per Block Group
Mean S.D.
0.264 0.882
0.222 0.482
0.243 0.715
0.078 0.281
0.322 1.589
0.120 0.797
0.140 0.856
Total
Exposed
Population
621
506
1,127
611
949
370
1,930
10 Year Lag (2030)
Per Block
Group
Mean S.D.
0.297 0.993
0.261 0.576
0.280 0.815
0.090 0.316
0.375 1.836
0.142 0.951
0.162 0.994
Total
Exposed
Population
701
595
1,295
699
1,104
435
2,238
20 Year Lag (2040)
Per Block
Group
Mean S.D.
0.364 1.215
0.339 0.767
0.352 1.020
0.112 0.395
0.480 2.344
0.184 1.267
0.206 1.279
Total
Exposed
Population
859
772
1,632
875
1,413
566
2,854
50 Year Lag (2070)
Per Block
Group
Mean S.D.
0.565 1.816
0.582 1.362
0.574 1.609
0.177 0.669
0.826 4.157
0.306 2.178
0.344 2.248
Total Exposed
Population
1,333
1,325
2,657
1,386
2,432
938
4,756
-------
Table 10-40. Summary of Estimated Mercury Exposures for Special Populations in 2001, with Associated IQ Decrements and
Foregone Earnings: Population Centroid Approach—Base Case 2001
Average Daily Maternal
Ingestion of Mercury (ug/day)
Population State
Hmong
MN
WI
2-State Total
Chippewa
MI
£ MN
KJ WI
3-State Total
Mean
3.783
5.392
4.457
6.406
4.057
5.815
5.218
S.D.
0.364
1.513
1.276
1.075
0.976
1.318
1.505
IQ Decrements in Prenatally Exposed
Children
Mean
0.097
0.138
0.114
0.164
0.104
0.149
0.134
S.D.
0.009
0.039
0.033
0.028
0.025
0.034
0.039
Total
31.106
31.961
63.067
65.523
52.526
28.060
146.109
Present Value of Foregone Net Earnings
due to IQ Decrements (1999$s in 2001)
Mean
$852
$1,215
$1,004
$1,444
$914
$1,310
$1,176
S.D.
$82
$341
$288
$242
$220
$297
$339
Total
$273,959
$281,492
$555,451
$577,082
$462,613
$247,128
$1,286,823
Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table.
-------
Table 10-41. Summary of Estimated Mercury Exposures for Special Populations in 2020, with Associated IQ Decrements and
Foregone Earnings: Population Centroid Approach—Base Case 2020 with CAIR"
Average Daily Maternal
Ingestion of Mercury (ug/day)
Population State
Hmong
MN
WI
2-State Total
Chippewa
MI
£ MN
N) WI
-J
3-State Total
Mean
4.040
5.101
4.456
5.733
3.904
5.506
4.749
S.D.
0.311
1.570
1.112
1.311
0.901
1.445
1.325
IQ Decrements in Prenatally Exposed
Children
Mean
0.103
0.131
0.114
0.147
0.100
0.141
0.122
S.D.
0.008
0.040
0.028
0.034
0.023
0.037
0.034
Total
65.563
53.566
119.128
83.480
91.320
39.476
214.276
Present Value of Foregone Net Earnings
due to IQ Decrements (1999$s in 2020)
Mean
$910
$1,149
$1,004
$1,292
$880
$1,241
$1,070
S.D.
$70
$354
$251
$295
$203
$326
$299
Total
$577,428
$471,769
$1,049,197
$735,236
$804,278
$347,674
$1,887,187
Benefits analyses using the population centroid approach were conducted at a block group level, but for summary purposes the results are aggregated and reported
at a state level in this table. For comparison purposes with the Base Cases in 2001, benefits presented in this table do not incorporate potential lags in fish tissue
response to a change in mercury deposition.
-------
Table 10-42. Summary of Annual Benefit Estimates for Hmong Special Population: Population Centroid Approach
N)
OO
Central Estimate of Fish Alternative Estimate of Fish
Tissue Response Times Tissue Response Times
10-Yr Lag 20-Yr Lag
Hmong (MN and WI)
Annual Number of Prenatally Exposed Children
2001 Base Case
2020 Base Case (with CAIR)
Per Capita Avoided IQ Decrements
2001 Utility Emissions Zero-Out (Relative to 2001 Base Case)
2020 Base Case with CAIR (Relative to 2001 Base Case applied to 2020
demographics)
2020 Utility Emissions Zero-Out (Relative to 2020 Base Case with CAIR)
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
Total Value of Benefits (19995s)
2001 Utility Emissions Zero-Out (Relative to 2001 Base Case)
1% Discount Rate; Present Value in 2001
3% Discount Rate; Present Value in 2001
7% Discount Rate; Present Value in 2001
2020 Base Case with CAIR (Relative to 2001 Base Case applied to 2020
demographics)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 Utility Emissions Zero-Out (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
BENEFITS OF CONTROL OPTIONS
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% ni-sr.nnnt Rate- Present Valnp in 7070
786
1,295
0.0096
0.0006
0.0075
0.0004
0.0015
N/A
$49,613
$33,894
N/A
$5,071
$3,465
N/A
$63,540
$43,409
N/A
$3,477
$2,375
N/A
$12,968
$8 8SQ
1,006
1,632
0.0095
0.0008
0.0076
0.0004
0.0016
N/A
$46,758
$21,824
N/A
$6,539
$3,052
N/A
$60,308
$28,148
N/A
$3,311
$1,546
N/A
$12,417
$5 706
5-Yr Lag 50- Yr Lac
672
1,127
0.0097
0.0004
0.0074
0.0004
0.0015
N/A
$49,506
$40,919
N/A
$3,725
$3,079
N/A
$63,511
$52,495
N/A
$3,466
$2,865
N/A
$12,877
•c i n 64"?
1,953
2,657
0.0094
0.0012
0.0078
0.0004
0.0016
$98,779
$37,057
$5,515
$16,402
$6,153
$916
$110,358
$41,401
$6,161
$6,076
$2,280
$339
$23,043
$8,644
$1 787
-------
Table 10-43. Summary of Annual Benefit Estimates for Chippewa Special Population: Population Centroid Approach
Central Estimate of Fish
Tissue Response Times
Alternative Estimate of Fish
Tissue Response Times
ts)
10-Yr Lag 20-Yr Lag
Chippewa (MI, MN, and WI)
Annual Number of Prenatally Exposed Children
2001 Base Case
2020 Base Case (with CAIR)
Per Capita Avoided IQ Decrements
2001 Utility Emissions Zero-Out (Relative to 2001 Base Case)
2020 Base Case with CAIR (Relative to 200 1 Base Case applied to 2020
demographics)
2020 Utility Emissions Zero-Out (Relative to 2020 Base Case with CAIR)
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
Total Value of Benefits (1999$s)
2001 Utility Emissions Zero-Out (Relative to 2001 Base Case)
1% Discount Rate; Present Value in 2001
3% Discount Rate; Present Value in 2001
7% Discount Rate; Present Value in 2001
2020 Base Case with CAIR (Relative to 2001 Base Case applied to 2020
demographics)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 Utility Emissions Zero-Out (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
BENEFITS OF CONTROL OPTIONS
2020 CAMR Control Option 1 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rate; Present Value in 2020
2020 CAMR Control Option 2 (Relative to 2020 Base Case with CAIR)
1% Discount Rate; Present Value in 2020
3% Discount Rate; Present Value in 2020
7% Discount Rafe- Present Valnp in 7070
1,448
2,238
0.022
0.015
0.013
0.001
0.002
N/A
$115,066
$78,611
N/A
$130,050
$88,847
N/A
$113,612
$77,618
N/A
$6,698
$4,576
N/A
$15,331
•CIO 474
1,767
2,854
0.021
0.016
0.013
0.001
0.002
N/A
$102,720
$47,943
N/A
$123,387
$57,589
N/A
$106,942
$49,913
N/A
$6,331
$2,955
N/A
$14,543
416788
S-Yr Lag 50-Yr Lac
1,293
1,930
0.024
0.015
0.013
0.001
0.002
N/A
$120,309
$99,441
N/A
$130,029
$107,475
N/A
$114,256
$94,438
N/A
$6,716
$5,551
N/A
$15,331
$12671
3,559
4,756
0.021
0.015
0.013
0.001
0.002
$214,504
$80,472
$11,976
$220,164
$82,595
$12,292
$190,562
$71,490
$10,639
$11,266
$4,227
$629
$26,212
$9,834
SI 463
-------
For the Hmong subpopulation, the aggregate benefits of CAMR Option 1 assuming 10
and 20 year lag periods are estimated to be $3,500 and $3,300 respectively. These estimates are
5 percent as large as the corresponding estimates for the 2020 Utility Emissions Zero-Out. For
CAMR Option 2, these estimates are $14,000 and $12,400 respectively, which are 20 percent as
large as the corresponding estimates for the 2020 Utility Emissions Zero-Out.
For the Chippewa subpopulation, the aggregate benefits of CAMR Option 1 assuming 10
and 20 year lag periods are estimated to be $6,700 and $6,300 respectively. These estimates are
6 percent as large as the corresponding estimates for the 2020 Utility Emissions Zero-Out. For
CAMR Option 2, these estimates are $15,300 and $14,500 respectively, which are 14 percent as
large as the corresponding estimates for the 2020 Utility Emissions Zero-Out.
The analyses presented in this section examining distributional equity in the context of
the Flmong and Chippewa were constrained significantly by not having considered variability in
fish consumption rates for these two high fish consuming populations. Challenges in identifying
peer-reviewed data characterizing high-end percentile self-caught freshwater fish consumption
rates for these two populations prevented IQ benefits modeling from considering inter-individual
variability in fish consumption rates. While this is not problematic in the context of modeling
overall (mean) benefits for the population as a whole, it does significantly limit the utility of
considering per-capita IQ benefits in the context of distributional equity, since a key factor
determining high-end benefits levels is not considered (i.e., high-end fish consumption rates
which can result in disproportionately large IQ benefits, if matched to large mercury fish tissue
concentration changes).
In order to provide additional insights into the distributional equity issue (and address to
a certain extent this limitation in characterizing high-end fish consumption in modeling both the
Hmong and Chippewa), EPA has included a Sensitivity Analysis specifically examining the
issue of distributional equity for high fish consuming populations. As discussed below, this
sensitivity analysis uses high-end (potentially bounding) fish consumption rates identified
through NODA comments to support an analysis of distributional equity for Native American
populations in the 37 state study area."
10.6.5 Sensitivity Analysis Examining the Economic Benefit Equity Issue in the Context of
High Fish Consuming (subsistence) Populations Including Native Americans
There is the potential that, due to elevated fish consumption rates, individuals exhibiting
subsistence-like fishing behavior could experience disproportionately higher benefits from this
rule (i.e., the degree of per-capita IQ benefit for prenatally-exposed children of these fishers
could be significantly higher than the degree of IQ benefit predicted for the general recreational
angler population). Another way of viewing this issue is that EGU-attributable mercury impacts
could disproportionately impact these high fish-consuming populations and consequently, this
rule could have relatively larger benefits for these special populations by reducing their
disproportionate health impacts. If this is the case (i.e., special high-consumption populations
are found to have disproportionately larger benefits then the general recreational angler), then it
be justified to support this rule on economic equity grounds, regardless of the net economic
benefit-cost ratio.
10-130
-------
Ideally, this distributional equity issue would be examined by systematically modeling
population-level benefits for potentially high exposure populations including Native American
populations exhibiting subsistence-like behavior. Specifically, using the population centroid
model combined with (a) a probabilistic consideration offish consumption rate variability in the
study population, (b) demographic data on the distribution of these special populations across the
study area and (c) modeled delta fish tissue concentrations for the CAMR rule, we would model
the distribution of benefits (i.e., IQ loss reductions) across specific high consumption
populations as a result of CAMR. In this context, modeling exposure using fish consumption
variability distributions is critical since ultimately, we are interested in higher-end exposures for
these populations which likely reflect a combination of high fish consumption rates paired with
relatively higher mercury fish tissue concentrations changes (for the rule being modeled).
In developing the benefits model for this RIA, EPA attempted to conduct this type of
focused benefits modeling for two key potentially high-consuming populations including the
Hmong and Chippewa (the former located in Minnesota, Wisconsin and Michigan and the latter
in Minnesota and Wisconsin - see Section 10.6.1). These two populations were chosen because
they are known to have higher fish consumption rates and because mercury fish tissue
concentrations in these areas are fairly large relative to other portions of the study area.
However, several factors surfaced in modeling these two population which prevented a
comprehensive analysis of these populations in the context of distributional equity including (a)
it was not possible to establish fish consumption distributions for either populations since we
could not identify a defensible upper-end percentile consumption rate for use in fitting a
distribution and (b) ultimately, air quality modeling showed that the regions where these two
populations are located are not generally associated with the highest EGU-related mercury
deposition changes (these high deposition change areas are located primarily near the Ohio River
valley). These two factors diminished the ability of the benefits analysis, as conducted for the
Hmong and Chippewa to effectively address the potential for distributional equity concerns
regarding these two populations.
However, given the potential importance that EPA places on distributional equity as a
consideration in cost-benefit analysis, we have completed a sensitivity analysis designed to
evaluate the potential for disproportionate health benefits for high fish consumption
(subsistence) populations. The methodology used in this sensitivity analysis and the results of
the analysis are presented in this section. Specifically, we have selected a high-end (near
bounding) fish consumption rate (discussed below) for Native American populations and
combined this value with upper-bound mercury fish tissue concentration changes (deltas) for
Option 1 identified through several scenarios (described below) to estimate upper bound IQ
benefits for these populations across the 37 state study area. This sensitivity analysis, while not
allowing any enumeration of these populations, does allow us to determine whether there is the
potential for disproportionate health impacts for Native American populations and other high
fish consuming populations based on a combination of these conservative exposure assumptions.
A determination that disproportionate health benefits are experienced by a special
population is ultimately based on consideration of two factors: (a) are the levels of health
benefits (in this case IQ loss reductions for prenatally-exposed children) for the special
population relatively larger than the general recreational angler population and (b) in absolute
terms, are the magnitude of the health benefits experienced by the special population considered
10-131
-------
significant. This second point is important to emphasize. Even if a special population has
relatively larger health benefits (e.g., an order of magnitude larger than the general recreational
angler), if those health benefits are considered non-significant in absolute terms (e.g., less than 1
IQ point is saved based on conservative high-end exposure modeling), then it may be reasonable
to conclude that there is little support for an distributional equity argument since the absolute
level of health benefit involved is not considered significant.
Fish Consumption Values Used in the Sensitivity Analysis
Because this sensitivity analysis is considering high-end conditions for exposure in the
context of this distributional equity analysis for Native American populations, the fish
consumption rate selected needed to represent high-end (near bounding) behavior, if possible.
The EPA recommends a 95*% consumption rate of 170 g/day for Native American subsistence
fishing populations in its Exposure Factors Handbook (EPA, 1997). This consumption rate
would be a reasonable upper-bound for Native American populations and could arguably be used
in this sensitivity analysis. However, in comments received for the NODA to this rule, fish
consumption rates specifically for the Ojibwa in Minnesota, Wisconsin and Michigan were
identified. These values covered consumption rates for the spring (189.6-393.8 grams/day) and
fall (155.8 to 240.7 grams/day) spear fishing seasons. Because this sensitivity analysis is
attempting to determine whether an equity issue exists (i.e., are there individuals with
disproportionate health benefits), EPA decided that rather than using the EPA-recommended
value of 170 g/day, it would develop an even more conservative (near bounding) value based on
the NODA comment data. Specifically, EPA used the highest seasonal value provided by the
commentor and apply that to the full year (i.e., assume that annual-averaged daily fish
consumption is 393.8 grams/day). This value is very conservative and is not appropriate for
other portions of this analysis since it is not possible to determine the percentile of the Ojibwe
population that it represents (i.e., it could be a near max/bounding value, or it could be closer to
a 95*%). However, because this distributional equity analysis is intended to screen for potential
disparities in health benefits, use of a very conservative bounding analysis is considered
reasonable in this case.
Selection of Fish Tissue Concentration Changes (deltas') Under Option 1 for the Sensitivity
Analysis
The sensitivity analysis was implemented using the change in fish tissue concentration
(for selected locations) due to Option 1. This is the relevant metric to use in examining the
distributional equity issue since we are interested in seeing whether special high consumption
populations experience disproportionately higher health benefits because of the CAMR rule.
However, this being said, there were several different ways in which specific fish tissue
concentration values could be selected (e.g., use the Option 1 concentrations modeled in the RIA
for the Chippewa, be more conservative and select the maximum fish tissue concentrations under
Option 1 across the states where the Chippewa are located). Ultimately, EPA evaluated three
different fish tissue concentration scenarios for this sensitivity analysis, each representing a
different perspective on fish tissue concentration locations:
Scenario 1 - use fish tissue concentration values modeled in the RIA for the Chippewa: This
scenario essentially represents applying the high-end fish consumption rate described above to
10-132
-------
the Chippewa population modeled above. With this scenario, we selected the maximum Option
1 fish tissue concentration modeled for the Chippewa in the RIA (i.e., the maximum ring-
averaged fish tissue concentration estimated for the Chippewa using the Population Centroid
Model.
Scenario 2 - select the maximum Option 1-attributable fish tissue concentration change
predicted anywhere within Minnesota, Wisconsin and Michigan (i.e., identify a single maximum
fish tissue concentration value for each state): This is a more conservative scenario than
Scenario 1 since it assumes that Native Americans could be fishing anywhere in these three
states. In addition, while Scenario 1 is based on averaging of standardized fish tissue
concentrations across the 6 predicted (edible) fish species and across simulation years (which is
reasonable for the economic analysis as conducted for the primary estimate), this scenario takes
the more conservative approach of selecting the maximum fish tissue concentration change that
will reflect a specific fish species in a specific year (i.e., Scenario 2 does not average across
species or time and therefore will represent a higher-end concentration change value compared
with Scenario 1).
Scenario 3 - identify the maximum Option 1 fish tissue concentration change for locations within
the 37 state study area where Native Americans are located: This scenario expands the
sensitivity analysis to include any HUCs co-located with tribal census tracts as defined by the
US Census. Specifically, the maximum fish tissue concentration (not averaged across species or
year) for each HUC was identified and used in the sensitivity analysis (Note: only the max value
found across the entire study area is presented in the results table).
EPA believes that these three scenarios provide reasonable coverage for Native American
Populations within the 37 state study area, although it is important to note that smaller
populations of Native Americans may be located in other locations not covered by the sensitivity
analysis. It is also important to note that, this sensitivity analysis (specifically Scenario 2) also
provides coverage for the Hmong, since that scenario uses the maximum fish concentration
change that is found in three of the states where Hmong are primarily located.
Results of the Sensitivity Analysis
The results of the sensitivity analysis examining the distributional equity issue are
presented in Table 10-44. This table identifies both the mercury fish tissue concentration change
(delta) values under Option 1 and the fish species (where applicable) and provides the IQ
changes modeled for this analysis.28 As described above, all results generated for this sensitivity
analysis were based on a conservative (near bounding) consumption rate of 393.8 g/day which
EPA identified through NODA comments. This value, while appropriate for a sensitivity
analysis, was not used in other components of this analysis (e.g., primary benefits estimate)
because it is not possible to clearly identify which percentile of the subsistence population this
value represents.
28 In generating these IQ changes, EPA used the same methodology applied in the general RIA for IQ change
estimation with the exception that the consumption rate described above (393.8 g/day) was used (see Section
10.1.3.1 and 10.1.3.2 for additional details on IQ change calculation). Modeling of IQ change for the sensitivity
analysis did include application of the cooking loss factor of 1.5 in predicting exposure levels.
10-133
-------
Table 10-44. Results of the Sensitivity Analysis Examining Distributional Equity for
Native American (subsistence) Populations
Sensitivity Analysis
Scenario
Option 1-attributable
mercury fish tissue
concentration change
(ppm)
Species modeled
Predicted IQ change
for prenatally-
exposed children
Scenario 1: maximum fish concentration change for Chippewa in the RIA
Chippewa value
0.00083
average across 6 species
0.012
Scenario 2: maximum fish concentration change in MI, Wl, and MN not limited to where Native
Americans are located (and not averaged across species or year)
Michigan
Wisconsin
Minnesota
0.0212
0.0207
0.0069
Walleye
Walleye
Walleye
0.32
0.31
0.10
Scenario 3: maximum fish concentration change in HUCs -within the 37 state study area where Native
Americans are located (based on Native American Census block designation)
Max value (MS)
0.0406
Walleye
0.61
The results presented in Table 10-44 suggest that there may not be a strong argument for
distributional equity for Native American (subsistence) populations within the 37 state study
area (given the precision of the fisher exposure model used here - see discussion below). Even
the most conservative scenario evaluated for the sensitivity analysis (involving the Option 3-
attributable fish tissue concentration change of 0.0406 ppm paired with the high consumption
rate used in this analysis), only produced a IQ change of 0.61 points. Although it is likely that
high consuming Native American (subsistence) populations, as well as other high consuming
populations do experience relatively higher benefits compared with the general recreational
angler, because the absolute degree of health benefit (in terms of IQ points saved) is still
relatively low (i.e., significantly less than 1), we conclude that a compelling argument can not be
made on distributional equity grounds for this rule.29
The issue of distributional equity can also be raised in the context of high fish consuming
(near subsistence) recreational anglers (i.e., is there a relatively smaller number of recreational
anglers who consume at the high-end of the consumption distribution and experience health
benefits significantly larger than the recreational freshwater angler of recreational anglers)?
Because we conducted modeling for recreational anglers for the RIA with consideration for fish
consumption variability by this population, it is possible to examine the tail of the Option 1 -
attributable IQ change distribution to determine whether high-end consumers in this population
might experience disproportionate health benefits. The maximum modeled IQ benefit for the
recreational angler as a result of Option 1-related mercury fish tissue concentration reductions is
29 The recreational angler population modeled for the RIA has a mean IQ change due to Option 1 attributable
mercury fish tissue concentrations changes of 0.010, which is over an order of magnitude lower than the maximum
IQ change predicted in the Sensitivity Analysis.
10-134
-------
0.28 IQ points, which results in the same conclusion as drawn for the Native American
populations (i.e., this degree of health benefit may be significantly higher than the general
recreational angler population with its mean IQ change of 0.0010, but it is still considered not
significant from an absolute perspective since it is significantly less than 1 IQ point).
Note, that the conclusions presented above are based on the fish consumption exposure
model used in this RIA (with modifications as noted above) and consequently reflect the level of
precision and accuracy inherent in this model. Because this economic benefits model was not
developed to support site-specific analysis of individual waterbodies and the populations that
fish at those waterbodies, there is the potential for this analysis to have overlooked individuals
who may be subjected to higher absolute benefits because they fish at waterbodies where the
Option 1 mercury fish concentration change is significantly larger than what is captured in the
three Scenarios described above.
10.7 Discussion and Qualification of Results: Assumptions, Limitations,
and Uncertainties
The previously described methods and results provide useful insights regarding:
1. The extent of potentially harmful mercury exposures due to consumption of
noncommercial freshwater fish;
2. The size and distribution of resulting IQ losses among prenatally exposed children,;
3. The value of future earnings losses associated with these IQ decrements; and
4. The extent to which these losses could be avoided through emissions controls on
coal-fired power plants.
However, these estimates are based on modeling approaches that require simplifying
assumptions and are subject to limitations and uncertainties.
Uncertainty regarding the estimates developed in this section can arise from several
sources. Some of the uncertainty can be attributed to model uncertainty. For example, to
estimate exposures a number of different modeling approaches have been selected and
combined, each of which simplifies potentially complex processes. The results therefore depend
importantly on how these models are selected and specified.
One strategy used to evaluate the potential effects of model uncertainty has been to
define alternative estimation approaches within the exposures assessment. For instance, two
separate approaches—population centroid and angler destination—were developed to estimate
the size of the exposed populations and their relation to mercury concentrations in fish. Also, to
investigate exposures among high-risk groups three different approaches were summarized in
Section 10.6. The ranges of exposure estimates reported for these different approaches help to
demonstrate the sensitivity of results with respect to model selection.
10-135
-------
Another important source of uncertainty can be characterized as input or parameter
uncertainties. Each of the modeling components discussed in this report requires summary data
and estimates of key model parameters. All of these inputs are measured with some degree of
uncertainty and can affect, to differing degrees, the confidence range of our summary results.
The discussion below identifies and highlights some of the key model parameters, characterizes
the source and extent of uncertainties associated with them, and characterizes the potential
effects of these uncertainties on the model results.
To organize this discussion, different components of the modeling framework are
discussed separately. It first discusses issues related to estimating the mercury concentrations
and then those related to estimating the exposed population. After that, it discusses issues
related to matching these two components and then concludes by discussing the estimation of
mercury ingestion through fish consumption.
10.7.1 Mercury Concentration Estimates
As described in Section 10.1.2, the core mercury concentration estimates for the analysis
come from the fish tissue sample data in the NLFA. These estimates were then used to
approximate mercury concentrations across the study area for a specific time period (2001) and
for normalized conditions (size, species, and cut offish). Some of the key assumptions,
limitations, and uncertainties associated with these estimates are the following:
• The NLFA data themselves are subject to measurement and reporting error and
variability. The NLFA is the largest and most detailed source of data on mercury in
fish; however, the system was not centrally designed (e.g., by EPA) using a common
set of sampling and analytical methods. Rather, states collected the data primarily to
support the development of advisories, and the data are submitted voluntarily to EPA.
Each state uses different methods and criteria for sampling and allocates different
levels of resources to their monitoring programs. In addition, there are uncertainties
regarding the precise locations (lat/long coordinates) of some of the samples. The
heterogeneity and potential errors across state sampling programs can bias the results
in any direction and contribute to uncertainty.
• The NLFA sampling data were assigned as either lake or river samples, based on the
location coordinates for sampling sites in the NLFA and by mapping them with GIS
to the nearest type of waterbody. This process also involves measurement error and
may have resulted in misclassifications for some of the samples. These errors are not
expected to bias results, but they contribute to uncertainty.
• The NDMMFT statistical model described in Section 10.1.1 was applied to the NLFA
to estimate concentrations at specific locations and dates under normalized conditions
(size, species, and cut offish). These normalized estimates were then averaged
across species and dates to create a single mercury concentration estimate for each of
the roughly 5,200 sampling locations in the study area. This process of normalizing
and averaging results involves both model uncertainty and estimation error. For
example, if anglers systematically keep fish of different species or sizes than those
10-136
-------
included in the normalization and averaging process, then the modeling approach
may lead to over- or underestimates of exposures.
The normalized/averaged mercury concentration estimates (summarized in
Table 10-2) were then spatially extrapolated, as discussed in Section 10.1.2. These
spatial extrapolation processes are potentially significant sources of uncertainty and
may also overstate actual mercury levels. In the population centroid approach,
average concentrations within specified distance intervals from block group centroids
were first extrapolated to all waterbodies within the interval, and then further
averaged and extrapolated to other distance intervals as well. In the angler
destination approaches, average concentrations in a HUC were used to characterize
fish from all waterbodies in the HUC. All of these approaches assume that NLFA
mercury samples are representative of "local" conditions in similar waterbodies.
However, even though states use a variety of approaches to monitor and sample fish
tissue contaminants, in some cases, the sampling sites are selected to target areas with
high levels of angler activity and/or a high level of pollution potential. To the extent
that sample selection procedures favor areas with relatively high mercury, the spatial
extrapolation methods used in this report will tend to overstate exposures. These
approaches also implicitly assume that mercury concentration estimates are strongly
spatially correlated, such that closer sampling sites (i.e., from same HUC or distance
interval) provide more information about mercury concentrations than more distant
sites. To the extent that spatial correlation is weaker than assumed, this will increase
the degree of uncertainty in the modeling results.
To estimate mercury fish tissue concentrations under each of the emissions reductions
scenarios, it was assumed that concentrations in any given area would decline in
exactly the same proportion as modeled reductions in mercury deposition for the area.
This assumption was based on Mercury Maps, which also assumes a linear, steady-
state relationship between concentrations of methylmercury in fish and air deposition
of mercury. However, Mercury Maps also recognizes that this condition may not be
met in waterbodies that contain significant non-air sources of mercury. To assess the
potential effect that nonair sources of mercury would have on the results reported in
this section, EPA identified 56 HUCs in the 37-state study area with gold or mercury
mines or chloralkali plants. EPA recalculated aggregate benefits in the study area for
the 2001 Utility Emissions Zero-Out scenario assuming that concentrations in these
56 HUCs would not be reduced at all by the emissions controls. The resulting benefit
estimates declined by less than 3 percent. These results suggest that overestimation
of benefits due to exclusion of sources is relatively small. This analysis did not
attempt to identify or otherwise account for naturally-occurring sources of mercury.
To the extent that these are present and substantially affect concentrations in the
water bodies of interest, the benefits will be overestimated.
10-137
-------
10.7.2 Exposed Population Estimates
Section 10.1.2 describes two main approaches for estimating the annual number of
children prenatally exposed to mercury because of their mothers' consumption of
noncommercial freshwater fish. The population centroid approach addresses this objective by
focusing on where these women of childbearing age are most likely to live in relation to mercury
levels in freshwater fish. In contrast, the angler destination approach focuses on where angler
households are most likely to go fishing, in relation to where mercury concentrations in
freshwater fish are located.
Each approach has advantages and disadvantages. The main advantage of the population
centroid approach is that it uses detailed Census data and allows us to characterize and subdivide
populations with a high degree of confidence at a high level of spatial resolution (almost 165,000
block groups in the study area). A main disadvantage of this approach is that it requires strong
assumptions for identifying the portion of these populations that live in freshwater angler
households and for matching these populations to mercury levels in fish. The main advantage of
the angler destination approach is that it starts by identifying angler populations and where they
fish; however, it requires relatively strong assumptions for linking these populations to
populations of women of childbearing age who consume their catch.
In certain respects, the two approaches rely on similar data sources and use them in
similar ways; therefore, they are subject to similar uncertainties.
• Both approaches rely primarily on data from the NSFHWR to estimate state-level
freshwater angler activity levels. The NSFHWR is based on a sample of over 10,000
freshwater anglers nationwide. The population centroid approach primarily uses the
NSFHWR to estimate state-level freshwater fishing participation rates and lake-to-
river day ratios. The angler destination approach uses data on the level of lake- and
river-fishing days by state. Each of these data elements is measured with some error
in the NSFHWR, but they are based on a relatively large sample. As discussed
below, more uncertainty is generated in both approaches when these state-level
estimates are applied or extrapolated to smaller spatial scales (block groups and
HUCs).
• Both approaches also use state-level fertility rate data to approximate the rate of
pregnancy among women of childbearing age in angler households for a smaller
geographic area. The state-level fertility rates from the National Vital Statistics are
estimated with relatively little error; however, applying these rates to specific block
groups or HUCs (and specifically to women in angler households) does involve
considerably more uncertainty.
In other respects, the two approaches use very different assumptions and are subject to
different sources of uncertainty in measuring exposed populations. Moreover, as discussed
below, some of these assumptions are likely to lead to underestimates of exposed populations in
the population centroid approach and to overestimates in the other approach.
10-138
-------
• The population centroid approach assumes that, in each block group, the percentage
of women who live in freshwater angler households (i.e., households with at least one
freshwater angler) is equal to the percentage of the state adult population that fishes.
Applying the state-level participation rate to approximate the conditions at a block
level creates uncertainty. More importantly, however, using individual-based fishing
participation rates to approximate household rates is likely to underestimate the
percentage of women living in freshwater angler households.30 Unfortunately data on
household participation levels in freshwater fishing are not readily available.
• In the angler destination approach, it is assumed that the percentage of self-caught
fish from a HUC that goes to households with women of childbearing age is equal to
the sum of (1) the state-level percentage of anglers who are women of childbearing
age plus (2) the state-level percentage of anglers who are male, married, and in the
same age range as women of childbearing age. In other words, it assumes that
women who are exposed during pregnancy must either be anglers themselves or be
married to an adult male angler below the age of 45. On the one hand, this approach
may not capture women who receive noncommercial fish from other individuals
(besides husbands younger than 45), in which case it would underestimate the size of
the exposed population. On the other hand, and probably more importantly, this
approach is likely to double count women of childbearing age who meet both criteria
(anglers and married to angler men younger than 45). The extent of this double
counting is not known, but it would lead to an overestimation of the exposed
population.
• The angler destination approach also estimates exposures by estimating "angler-year
equivalents" for each HUC. Angler-year equivalents can be thought of as groups of
freshwater fishing days in a HUC, such that 16.42 fishing days in a year (U.S. annual
average for freshwater anglers) are equivalent to one angler year. This approach
assumes that the same expected total mercury ingestion by pregnant women will
result from these fishing days, regardless of whether the days are all spent by one or
by many anglers.31
10.7.3 Matching of Exposed Populations to Mercury Concentrations
Section 10.1.3 also describes how the two approaches were used to match the estimated
exposed populations in each geographic area with corresponding mercury levels in freshwater
fish. In the population centroid approach, this entails matching (1) different portions of the
estimated number of pregnant women in angler households in each block group with (2) average
mercury concentration estimates in different waterbody types and distance intervals. In the
angler destination approach, this entails matching (1) different portions of freshwater fishing
days in each state with (2) average mercury concentration estimates in each HUC in the state.
30 For example, hypothetically if one out of every three members in each household fished, the population rate would
be 33 percent, but the household rate would be 100 percent.
31 If the dose-response relationship for health effects with respect to mercury ingestion is linear (as is assumed), the
aggregate health effects resulting from ingestion should only depend on the total mercury ingestion and not on how
the ingestion is distributed among the exposed population.
10-139
-------
In the population centroid approach, subpopulations are assigned to waterbody types
based on state-level ratios of lake-to-river fishing days (from the NSFHWR). They are further
assigned to distance intervals based on observed travel distance patterns in national fishing data
(NSRE 1994). An important limitation of both methods is that neither one takes into account the
physical characteristics of the area in which the population is located. In particular, the
allocation of exposures to lakes or rivers at different distances from each block group does not
take into account the presence or number of these waterbodies in each distance interval.
Using the NSRE 1994 data to assign subpopulations to distance intervals involves
additional uncertainty because it is based on more aggregate data and on a much smaller sample
of anglers than the NSFHWR. Additional uncertainty is also introduced by using self-reported
travel distance estimates to calculate travel distance patterns3232. Moreover, the analysis does not
control for the potential effect of trip frequency on trip distance and resulting exposures. For
example, if anglers who fish more often are also more likely to fish closer to home, then mercury
concentrations within the smaller distance intervals should be weighted more heavily in their
average mercury exposure estimates. Because a constant average freshwater fish consumption
rates is assumed in the main analysis (8 g/day), variations in fishing (and therefore fish
consumption) intensity are not accounted for. If there is a negative correlation between trip
frequency and average travel distance, and if more frequent fishers are located in areas closer to
higher (lower) average mercury concentration, then the population centroid approach is likely to
underestimate (overestimate) mercury ingestion levels for frequent fishers and for the exposed
population as a whole.
The angler destination approach also combines state-level (NSFHWR) and national data
(NSRE 1994) to match anglers with mercury concentrations. However, in contrast to the
population centroid approach, it allocates state-level lake- and river-fishing days based on both
physical and demographic data. Rather than allocating fishing days in proportion to just one of
these characteristics (e.g., HUC area), data from the NSRE were used to estimate and apply a
multivariate model described in Appendix E-2. Although based on a relatively small sample of
fishing trip data and a somewhat crude estimate of angler activity in each HUC, the model results
indicate that several factors, including population size and number of lake/river miles per HUC,
have statistically significant effects in explaining variations in fishing levels across HUCs. The
intended purpose of this model for predicting the allocation of fishing days across HUCs (rather
than a simpler allocation rule) is to reduce uncertainty by including as much information as
possible about the HUCs. However, it does not eliminate uncertainty in these estimates. The
modeled coefficients inherently include statistical error, and other unmeasured factors are also
likely to contribute to variations in fishing levels across HUCs.
One potentially important factor that is not included in either model for matching
populations and mercury concentrations is the effect offish consumption advisories on fishing
behavior. Evidence summarized in Jakus, McGuinness, and Krupnick (2002) suggests that
awareness of advisories by anglers is relatively low (less than 50 percent), and even those who
32 In addition to errors in respondents' own assessments of travel distance, their estimates may be a more accurate
reflection of road distance than the straight line distance used in the exposure calculations. If so, the analysis most
likely overestimates the percentage of trips to more distant waterbodies. The effect of this potential overestimation
on the average and total mercury exposure estimates is not known.
10-140
-------
are aware do not always alter their fishing behavior. Nonetheless, anglers are less likely to fish
in areas with advisories. Unfortunately, we were not able to reliably quantify the reduction and
redistribution of fishing trips in either model to account for fish advisories. By excluding these
effects, the model estimates are likely to overstate mercury exposures.
10.7.4 Fish Consumption Estimates
Perhaps the most influential variable in both modeling approaches is the rate of
noncommercial freshwater fish consumption. Based on recommendations in EPA's EFH, we
have assumed 8 g/day for the general population in freshwater angler households.
Unfortunately, data are not available to reliably vary this rate with respect to characteristics of
the population across the entire study area. In Section 10.6, we applied alternative consumption
rate assumptions for selected Asian and Native American populations because existing studies
have estimated consumption rates specifically for these populations, but they represent a small
portion of our study population.
Uncertainty regarding the true average fish consumption rate has a direct effect on
uncertainty for the model results. Because a single consumption rate is applied uniformly across
the entire exposed population and because it is a multiplicative factor in the model (see Eq.
[10.1]), the two uncertainties are directly proportional to one another. As discussed in Section
10.1.3, the recommended 8 g/day rate is based on four studies with mean estimates ranging from
5 (37 percent less than 8) to 17 (113 percent more than 8) g/day. If it is assumed that this range
of estimates represents the uncertainty in the mean freshwater fish consumption rate for the study
population, then the resulting uncertainty range for the estimated mean mercury ingestion level
will also be between -37 percent and +113 percent of the mean mercury ingestion level.
For a number of reasons, average consumption rates of freshwater fish may well differ
from the recommended 8 g/day value. For example, the rate may be less for females than for
male anglers and even less for females in angler households who do not fish themselves. To the
extent that their consumption rates differ, the model estimates are likely to overstate mercury
exposure.
Because offish consumption advisories of various types, women of childbearing age and
pregnant women in particular may well reduce their levels of consumption. Some evidence of
these types of behavioral changes were found, for example, in a recent study by Oken et al.
(2003). To the extent that these types of changes occur, they suggest that the current model
results overstate mercury exposures in the population of interest.
A final potentially influential variable in both modeling approaches is the assumed
conversion factor for mercury concentrations between uncooked and cooked fish. Studies have
found that cooking fish tends to reduce the overall weight offish by approximately one-third
(Great Lakes Sport Fish Advisory Task Force, 1993) without affecting the overall amount of
mercury. But these conversion rates depend on cooking practices and types offish. Uncertainty
regarding this conversion factor also has a proportionate effect on the modeling results.
10-141
-------
10.7.5 Modelling and Valuation oflQ Related Effects
The models for estimating and valuing IQ effects based on (1) estimated mercury
ingestion levels and (2) the size of the exposed population involve three main steps. The first
step is translating maternal mercury ingestion rates to mercury levels in hair. The second step is
translating differences in hair mercury concentrations during pregnancy to IQ changes in
offspring. The third step is translating IQ losses into expected reductions in lifetime earnings.
As discussed below, each of these steps also involves uncertainty.
• The conversion of mercury ingestion rate to mercury concentration in hair is based on
uncertainty analysis of a toxicokinetic model for estimating reference dose (Swartout and
Rice, 2000). The conversion factor was estimated by considering the variability and
uncertainty in various inputs used in deriving the dose including body weight, hair-to-
blood mercury ratio, half-life of MeHg in blood, and others. Therefore, there is
uncertainty regarding the conversion factor between hair mercury concentration and
mercury ingestion rate. Although, the median conversion factor (0.08 jig/kg-day/hair-
ppm) is used, the ninety-percent confidence interval is from 0.037 to 0.16 jJ.g/kg-
day/hair-ppm. Any change in the conversion factor will proportionately affect the
benefits results because of the linearity of the model.
• The dose-response model used to estimate neurological effects on children because of
maternal mercury body burden is susceptible to various uncertainties, as discussed in
Section 9.4. In particular there are three main concerns. First, there are other cognitive
end-points that have stronger association with MeHg than IQ point losses. Therefore,
using IQ points as a primary endpoint in the benefits assessment may underestimate the
impacts. Second, blood-to-hair ratio for mercury is uncertain which can cause the results
from analyses based on mercury concentration in blood to be uncertain. Third,
uncertainty is associated with the design of the epidemiological studies used in deriving
the dose-response models and the differences between these studies33. More discussion
on the uncertainty in dose-response function is provided in Section 9.4.
• The valuation of IQ losses is based on a unit-value approach developed by EPA, which
estimates that the average effect of a 1 point reduction in IQ is to reduce the present value
of net future earnings by $8,807 (in 1999$). Three key assumptions of this unit-value
approach are that (1) there is a linear relationship between IQ changes and net earnings
losses and (2) the unit value applies to even very small changes in IQ, and (3) the unit
value will remain constant (in real present value terms) for several years into the future.
Each of these assumptions contributes to uncertainty in the results. The unit value
estimate is itself subject to two main sources of uncertainty. The first source is directly
related to uncertainties inherent in the statistical analysis by Salkever (1995), which
provides estimates of average reductions in future earnings and years in school as a result
of IQ changes. The average percent change estimates are subject to statistical error,
modeling uncertainties, and variability across the population. The second main source of
33 One uncertainty from the epidemiological studies is that the cognitive test were performed on young children who
continued to be exposed to mercury after birth.
10-142
-------
uncertainty are the estimates of average lifetime earnings and costs of schooling. Both of
these estimates are derived from national statistics from the early 1990s, but they are also
subject to statistical error, modeling uncertainties, and variability across the population.
10.7.6 Unqualified Benefits
In addition to the uncertainties discussed above associated with the benefit analysis of
reducing exposures to methylmercury from recreational freshwater angling, there are several
additional benefits that we are unable to quantify, which adds to the uncertainties in the final
estimate of benefits of the CAMR.
Table 10-45 displays the health and ecosystem effects associated with methylmercury
exposure that are discussed in Section 2 for which we are currently unable to quantify.
Table 10-45. Unqualified Health and Ecosystem Effects Associated with Exposure to
Mercury
Category of Health or Ecosystem Effect
Neurologic Effects
Cardiovascular Effects*
Genotoxic Effects*
Immunotoxic Effects*
Ecological Effects*
Potential Health or Ecosvstem Outcomes
Impaired cognitive development
Problems with language
Abnormal social development
Potential for fatal and non-fatal myocardial infarctions
(heart attacks) - See Appendix B for a detailed
discussion
Associations with genetic effects
Possible autoimmunity effects in antibodies
Neurological effects in wildlife (birds, fish, and
mammals) that is similar to humans
* These are potential effects and are not quantified because the literature is either contradictory
or incomplete.
For one of these health effects, cardiovascular disease, the Agency conducted a critical
review of the available literature and determined that while some studies show that the effect
may exist, it is premature to include analysis of cardiovascular effects in our benefit analysis.
Studies investigating the relationship between methylmercury exposure and cardiovascular
impacts have reached different conclusions. The findings to date and the plausible biologic
mechanisms warrant additional research in this area. Appendix B provides an in-depth
discussion of these epidemiology studies. If future scientific studies demonstrate this effect
occurs, the benefits of reduced cardiovascular effects (from fatal and non-fatal heart attacks) if
quantified could possibly be many times larger than those we are able to quantify in this section
of the report due to the potential for mortality effects (monetized with the value of a statistical
life which is much higher in value than IQ loss).
In addition to the health and ecosystem effects that we are not able to quantify, there are
exposures to other segments of the U.S. population that we are currently unable to quantify. In
Section 4 of this report, we discuss the other fish consumption pathways that lead to exposure to
10-143
-------
methylmercury, including: consumption of commercial seafood and freshwater fish (produced
domestically as well as imported from foreign sources), and consumption of recreationally
caught seafood from estuaries, coastal waters, and the deep ocean. These consumption pathways
impact additional recreational anglers who are not modeled in our benefit analysis as well as the
general U.S. population. Reductions in domestic fish-tissue concentrations can also impact the
health of foreign consumers (consuming U.S. exports). Due to technical/theoretical limitations
in the science, EPA is unable to quantify the benefits associated with several of these fish
consumption pathways. For example, reductions in U.S. power plant emissions will result in a
lowering of the global burden of elemental mercury, which will likely produce some degree of
reduction in mercury concentrations for fish sourced from the open ocean and freshwater and
estuarine waterbodies in foreign countries. In the case of mercury reductions for fish in the open
ocean, complexities associated with modeling the linkage between changes in air deposition of
mercury and reductions in biomagnification and bioaccumulation up the food chain (including
open ocean dilution and the extensive migration patterns of certain high-consumption fish such
as tuna) prevent the modeling offish obtained from the open ocean. In the case of commercial
fish obtained from foreign freshwater and estuarine waterbodies, while there are technical
challenges associated with modeling long-range transport of elemental mercury and the
subsequent impacts to fish in these distant locations, additional complexities such as accurately
modeling patterns of harvesting and their linkages to commercial consumption in the U.S.
prevent inclusion of foreign-sourced freshwater and estuarine fish in the primary benefits
analysis.
Finally, with regard to commercially-produced freshwater fish sourced in the U.S. (i.e.,
fish from catfish, bass, and trout farms), we are unable to accurately quantify effects from this
consumption pathway because many of the fish farms operating in the U.S. utilize feed that is
not part of the aquatic foodweb of the waterbody containing the fish farm (e.g., use of
agricultural-based supplemental feed). In addition, many of the farms involve artificial
"constructed" waterbody environments that are a-typical of aquatic environments found in the
regions where those farms are located, thereby limiting the applicability of Mercury Maps
assumption in linking changes to mercury deposition to changes in mercury fish tissue
concentrations (e.g., waterbodies may have restricted or absent watersheds and modified aquatic
chemistry, which can effect methylation rates and impact time-scales for reaching steady-state
mercury fish tissue concentrations following reductions in mercury deposition). Some research
indicates that the recycling of water at fish farms can magnify the mercury concentration because
the system does not remove mercury as it is recycled, while newly deposited mercury is added to
the system. Thus, additional research on aquaculture farms is necessary before a benefit analysis
can be conducted.
Exclusion of these commercial pathways means that this benefits analysis, while
covering an important source of exposure to domestic mercury emissions (recreational
freshwater anglers), excludes a large and potentially important group of individuals. As
discussed in Section 4 of this report, recreational freshwater consumption accounts for
approximately 10 -17 percent of total U.S. fish consumption, and 90 percent is derived from
commercial sources (domestic seafood, aquaculture, and imports). However, as is mentioned
throughout this report, several of the other consumption pathways will not be affected by
CAMR, or will be affected to at minimal levels.
10-144
-------
Another area of unquantified benefits is associated with the study area selected for the
benefit analysis conducted in this section, specifically the eastern 37 states of the continental
U.S.. The focus on the eastern-half of the U.S. excludes evaluation of benefits from freshwater
sources in the West as well as commercially produced fish in the Pacific (which produces 68%
of the commercial fish supply in the U.S.). However, air quality modeling has shown that the
largest change in deposition from U.S. power plants emissions of mercury will occur in the
eastern-half of the U.S. so the unquantified benefits for this portion of the U.S. is expected to be
quite small.
In conclusion, there are several unquantified benefits associated with this analysis that
add to the overall uncertainty in the estimate of total benefits of CAMR. To the extent that
CAMR will reduce mercury deposition from power plants over estuarine areas, coastal, open
ocean waters, and in waterbodies in the western-half of the U.S., there would be a subsequent
reduction in mercury fish tissue concentrations in these different waterbodies and an associated
benefit from avoided decrements in IQ and other known health and ecosystem effects.
10.8 References
Belton, T., R. Roundy, and N. Weinstein. 1986. "Urban Fisherman: Managing the Risks of
Toxic Exposure." Environment 28(9): 19-37.
Berger, Mark C., Glenn C. Blomquist, Don Kenkel, and George S. Tolley. 1987. "Valuing
Changes in Health Risks: A Comparison of Alternative Measures." Southern Economic
Journal 53(4):967-984.
Connelly, N.A., B.A. Knuth, and T.L. Brown. 1996. "Sportfish Consumption Patterns of Lake
Ontario Anglers and the Relationship to Health Advisories." North American Journal of
Fisheries Management 16:90-101.
Ebert, E., N. Harrington, K. Boyle, J. Knight, J. and R. Keenan. 1993. "Estimating
Consumption of Freshwater Fish among Maine Anglers." North American Journal of
Fisheries Management 13:737-745.
Great Lakes Sport Fish Advisory Task Force. September 1993. Protocol for a Uniform Great
Lakes Sport Fish Consumption Advisory.
Hamilton, B.E., P.O. Sutton, and S.J. Ventura. August 4, 2003. "Revised Birth and Fertility
Rates for the 1990s and New Rates for Hispanic Populations, 2000 and 2001: United
States." National Vital Statistics Reports 51(12).
Harrington, Winston, and Paul R. Portney. 1987. "Valuing the Benefits of Health and Safety
Regulation." Journal of Urban Economics 22:101-112.
Hutchison, R., and C.E. Kraft. 1994. "Hmong Fishing Activity and Fish Consumption."
Journal of Great Lakes Research 20(2):471-487.
10-145
-------
Jakus, P., M. McGuinness, and A. Krupnick. 2002. "The Benefits and Costs of Fish
Consumption Advisories for Mercury." Discussion Paper 02-55. Washington, DC:
Resources for the Future.
Morgan, J.N., M.R. Berry, and R.L. Graves. 1997. "Effects of Commonly Used Cooking
Practices on Total Mercury Concentration in Fish and Their Impact on Exposure
Assessments." Journal of Exposure Analysis and Environmental Epidemiology
7(1):119-133.
Oken E., K.P. Kleinman, W.E. Berland, S.R. Simon, J.W. Rich-Edwards, and M.W. Gillman.
2003. "Decline in Fish Consumption Among Pregnant Women After a National Mercury
Advisory." Obstetrics and Gynecology 102(2):346-351.
Pao, E.M., K.H. Fleming, P.M. Guenther, and S.J. Mickle. 1982. Foods Commonly Eaten by
Individuals: Amount per Day and per Eating Occasion. USDA Home Economics
Research Report. No. 44. Washington, DC, Human Nutrition Information Service.
Citations as referenced in EPA 1997b.
Peterson, D.E., M.S. Kanarek, M.A. Kuykendall, J.M. Diedrich, H.A. Anderson, P.L.
Remington, and T.B. Sheffy. 1994. "Fish Consumption Patterns and Blood Mercury
Levels in Wisconsin Chippewa Indians." Archives of Environmental Health 49(l):53-58.
Pullis, Genevieve. 2000. Participation and Expenditure Patterns of African-American,
Hispanic, and Women Hunters and Anglers: Addendum to the 1996 National Survey of
Fishing, Hunting and Wildlife-Associated Recreation. Report 96-6. Washington, DC:
U.S. Fish and Wildlife Services.
Salkever, David. 1995. "Updated Estimates of Earnings Benefits from Reduced Lead Exposure
of Children to Environmental Lead." Environmental Research 70:1-6.
Swartout, J., and G. Rice. 2000. "Uncertainty Analysis of the Estimated Ingestion Rates Used
to Derive the Methylmercury Reference Dose." Drug and Chemical Toxicology
23(1):293-306. 11-41
U.S. Department of the Interior (DOI), Fish and Wildlife Service and U.S. Department of
Commerce, Bureau of the Census. 1992. 1991 National Survey of Fishing, Hunting, and
Wildlife-Associated Recreation. Washington, DC: U.S. Government Printing Office.
U.S. Department of the Interior (DOI), Fish and Wildlife Service and U.S. Department of
Commerce, Bureau of the Census. 1997. 1996 National Survey of Fishing, Hunting, and
Wildlife-Associated Recreation. Washington, DC: U.S. Government Printing Office.
U.S. Department of the Interior (DOI), Fish and Wildlife Service and U.S. Department of
Commerce, Bureau of the Census. 2002. 2001 National Survey of Fishing, Hunting, and
Wildlife-Associated Recreation. Washington, DC: U.S. Government Printing Office.
10-146
-------
U.S. Environmental Protection Agency (EPA). 1997a. Exposure Factors Handbook. Volume 1:
General Factors. EPA/600/P-95/002Fa. Washington, DC: Office of Research and
Development, National Center for Environmental Assessment.
U.S. Environmental Protection Agency (EPA). 1997b. Exposure Factors Handbook. Volume 2:
FoodIngestion Factors. EPA/600/P-95/002Fa. Washington, DC: Office of Research
and Development, National Center for Environmental Assessment.
U.S. Environmental Protection Agency (EPA). 1997c. Mercury Study Report to Congress.
EPA-452/R-97-003. Washington, DC: U.S. Environmental Protection Agency.
U.S. Environmental Protection Agency (EPA). 2000a. Economic Analysis of Toxic Substances
Control Act Section 403: Lead-Based Paint Hazard Standards. Washington, DC: U.S.
Environmental Protection Agency.
U.S. Environmental Protection Agency (EPA). 2000b. Fish Consumption and Environmental
Justice. Washington, DC: U.S. Environmental Protection Agency.
U.S. Environmental Protection Agency (EPA). 2000c. Methodology for Deriving Ambient
Water Quality Criteria for the Protection of Human Health (2000). Washington, DC:
U.S. Environmental Protection Agency.
U.S. Environmental Protection Agency (EPA). 2002a. Estimated Per Capita Fish Consumption
in the United States. EPA-821-C-02-003. Washington, DC: U.S. Environmental
Protection Agency.
U.S. Environmental Protection Agency (EPA). 2002b. Guidelines for Preparing Economic
Analyses. Washington, DC: U.S. Environmental Protection Agency.
U.S. Environmental Protection Agency (EPA). 2002c. Mercury Neurotoxicity Workshop Notes.
Washington, DC. November 4, 2002. http://www.epa.gov/ttn/ecas/regdata/
Benefits/mercuryworkshop.pdf.
Wente, S.P. 2004. A Statistical Model and National Data Set for Partitioning Fish-Tissue
Mercury Concentration Variation between Spatiotemporal and Sample Characteristic
Effects: U.S. Geological Survey Scientific Investigations Report 2004-5199.
West, P.C., M.J. Fly, R. Marans, and F. Larkin. 1989. Michigan Sport Anglers Fish
Consumption Survey. A report to the Michigan Toxic Substance Control Commission.
Michigan Department of Management and Budget Contract No. 87-20141.
West, P.C., J.M. Fly, R. Marans, F. Larkin, and D. Rosenblatt. May 1993. 1991-92 Michigan
Sport Anglers Fish Consumption study. Prepared by the University of Michigan, School
of Natural Resources for the Michigan Department of Natural Resources, Arm Arbor, MI.
Technical Report No. 6.
10-147
-------
West, Patrick C. 1992. "Invitation to Poison? Detroit Minorities and Toxic Fish Consumption
from the Detroit River." In Race and the Incidence of Environmental Hazards: A Time
for Discourse. Bryant Bunyan and Paul Mohai (eds.). Boulder: Westview Press.
Woods & Poole Economics, Inc. 2001. Population by Single Year of Age CD. CD-ROM.
Woods & Poole Economics, Inc.
10-148
-------
SECTION 11 BENEFITS OF MERCURY REDUCTION CONSIDERING ESTABLISHED
HEALTH-BASED BENCHMARKS AND OVERALL BENEFITS
CONCLUSIONS 11-1
11.1 Introduction 11-1
11.2 The Mercury - IQ-loss Paradigm 11-1
11.3 Quantifying IQ Benefits Associated with Mercury Emission Reductions ... 11-3
11.4 Data Element (3) - The Level of the Threshold 11-4
11.5 Date Element (4) - The Baseline Levels of Exposure for Consumers of
Recreational-Caught Fish 11-4
11.6 Overview of Benefits Methodology 11-5
11.7 Freshwater Fish Mercury Exposure 11-5
11.8 Deriving Baseline Mercury Exposures from All Sources of Mercury 11-8
11.9 Deriving Scaling Factors 11-12
11.10 Monetization and Scaling of IQ Benefits 11-13
11.11 Uncertainties 11-15
11.12 Conclusions 11-15
11.13 References 11-16
Tables
Table 11-1. Freshwater Fish IQ Loss and Hair Mercury from the 2020 Baseline 11-6
Table 11-2. Change in IQ Loss and Hair Mercury from the 2020 Zero Out (No Threshold)
Compared to the Baseline, and the Relative Probability of each Change Category .. 11-7
Table 11-3. Change in IQ Loss and Hair Mercury from the CAMR Option 1 (No Threshold)
Compared to the Baseline, and the Relative Probability of each Change Category .. 11-8
Table 11-4. Change in IQ Loss and Hair Mercury from the CAMR Option 2 (No Threshold)
Compared to the Baseline, and the Relative Probability of each Change Category .. 11-8
Table 11-5. Joint Distribution of Mercury Exposure from Freshwater Fish and Total Mercury
Exposure 11-10
Table 11-6. Scaling Factors 11-13
Table 11-7. IQ Benefits for CAMR Option 1 under Established Health-Based Benchmarks
11-14
Table 11-8. IQ Benefits for CAMR Option 2 under Established Health-Based Benchmarks
11-14
-------
SECTION 11
BENEFITS OF MERCURY REDUCTION CONSIDERING ESTABLISHED
HEALTH-BASED BENCHMARKS AND OVERALL BENEFITS CONCLUSIONS
11.1 Introduction
Chapter 10 presents our estimates of the reductions in methylmercury exposure from
recreationally-caught fish for our regulatory options. That chapter also presents our calculations
of IQ loss for the case of a nonthreshold, linear dose-response curve (i.e., the analysis assumes a
linear dose-response curve and no threshold).1 Under this combination of assumptions, the risk
assessor need not worry about background concentrations - all mercury exposure reductions
have the same effect on IQ, regardless of the initial level of exposure from all sources of
mercury. This set of assumptions is convenient but not consistent with EPA's mercury
Reference Dose (RfD).2 In fact, as discussed below, the effect of methylmercury on IQ is less
likely below the benchmark dose (BMD) and increasingly more uncertain as exposures approach
zero.
This chapter first presents some important summary material about EPA's mercury RfD
and discusses the implications of the RfD for risk assessment and benefit analysis. Next, the
exposure reduction and IQ loss estimates for the no-threshold case presented in Chapter 10 are
scaled to quantify the IQ loss due to fetal mercury exposure under a variety of scenarios
depicting our understanding of possible thresholds and background concentrations. The
scenarios in this chapter represent the Agency's interpretation of benefits under the current RfD.
The IRIS assessment (Methylmercury (MeHg), EPA's Integrated Risk Information System,
CASRN 22967-92-6) USEPA, www.epa.gov/iris/subst/0073.htm. EPA 2001) reflects the
Agency's formal position on the nature of risk of Mercury at low doses.
The chapter concludes with a presentation of EPA's final estimates of IQ-related
benefits. To generate IQ-related benefits from reduced mercury exposure, we use three different
models - two reflect health-based thresholds for mercury and the other assumes no threshold
exist. The benefit estimates are arrayed in a heirarchy from most certain to less certain benefits.
11.2 The Mercury - IQ-loss Paradigm
In general, a risk threshold for any particular substance suggests that there is an exposure
level that is without appreciable risk of adverse health effects. In the absence of data or
compelling biological rationale indicating the contrary, EPA's default paradigm assumes such a
1 IQ was chosen as the endpoint for this analysis because the monetary implications associated with this endpoint
are the most straightforward to model and provide a reasonable method for developing a concentration-response
relationship. (EPA 2002)
2 This linear, no threshold case serves a very useful purpose. In this chapter, we explain that as we empirically
apply thresholds to the exposed population, we can scale the "no threshold" IQ loss by the change in exposed
populations (weighted by change in exposure) to arrive at benefits for threshold scenarios
11-1
-------
threshold for all noncancer health effects. This threshold is reflected in the derivation of the
Reference Dose (RfD). "In general, the RfD is an estimate (with uncertainty spanning perhaps
an order of magnitude) of a daily exposure to the human population (including sensitive
subgroups) that is likely to be without an appreciable risk of deleterious effects during a
lifetime" (EPA 2001, p. 2) "This oral reference dose is based on the assumption that thresholds
exist for certain toxic effects ..." 3 (EPA 2001, p. 2) The RfD is typically expressed in units of
mg/kg bw/day.
The RfD paradigm is appropriate for methylmercury. We lack a biological rationale for
how methylmercury causes neurobehavioral loss and other neurological effects4. Hence we
cannot use our understanding of how methylmercury causes these effects to provide suggestive
theories of the lack of a threshold. Further, for the most part, the data from the major studies
examining developmental and neurological effects from mercury exposure involve exposure
levels that are well above the exposure levels in the U.S. Hence, the underlying data do not
provide an ability to empirically test whether a threshold does or does not exist at levels of
exposure experienced by US citizens.
As described in Section 2.3 of this document, in updating the RfD, EPA considered a
BMD analyses completed by NRC involving endpoints of neuropsychological development from
the Faroe Islands cohort (including results for the Boston Naming Test), the New Zealand
cohort, and the NRC's integrative analysis of all three studies. The BMDLs for these endpoints,
measured as concentrations of mercury in umbilical cord blood, were considered. For the
purposes of calculating the RfD, EPA converted these BMDLs to maternal daily dietary intake in
mg/kg bw/day using a one-compartment model. The BMDLs for these analyses (measured in
3 Note that it is conceptually possible to have a de minimus risk at the RfD, rather than an absolute threshold.
4 There is limited evidence directly linking IQ and methylmercury exposure in the three large epidemiological
studies that were evaluated by the NAS and EPA. Based on its evaluation of the three studies, EPA believes that
children who are prenatally exposed to low concentrations of methylmercury may be at increased risk of poor
performance on neurobehavioral tests, such as those measuring attention, fine motor function, language skills,
visual-spatial abilities (like drawing), and verbal memory. For this analysis, EPA is adopting IQ as a surrogate for
the neurobehavioral endpoints that NAS and EPA relied upon for the RfD.
In the Faroes Island Study, a full scale IQ evaluation was not conducted. However, two core subtests were evaluated
(similarities and block design) and one supplementary test was conducted (Digit Span). The similarities and block
design tests are reported to be well correlated with the full WISC-R battery (0.885, see Bellinger (2005) paper), but
how the Digit Span test relates is not reported. In the EPA analysis, we assume that it relates similarly. In the
Faroes study, performance scores on the similarities and block design tests were not shown to be statistically related
to cord blood or maternal mercury levels; the digit span test did show a statistical relationship with cord blood
mercury.
Both the New Zealand and Seychelles study administered the WISC IQ test (WISC III in Seychelles, WISC R in
New Zealand). A reanalysis of the New Zealand data found a positive association, but it was not statistically
significant. No significant associations were seen in the Seychelles study. As displayed in Figure 5 of Ryan (2005),
the confidence intervals for full scale IQ in both these studies include zero. However, Ryan conducted an integrative
analysis, combining results from all three studies. When combined, the statistical power of the analysis increases.
While the size of the dose-response relationship declined relative to past studies with a statistically significant
finding, Ryan found a statistically significant relationship between IQ aad mercury. The confidence interval did not
include zero.
11-2
-------
terms of mercury in cord blood) were all observed to be within a relatively close range, and the
calculated RfDs converge at about 0.0001 mg/kg bw/day. These exposures were converted to
dietary exposures of about 0.0005 mg/kg bw/day to 0.0019 mg/kg bw/day, with most dietary
exposures estimated to be about 0.001 mg/kg bw/day. The integrative BMDL (taking into
account data from all three studies) was calculated by NRC to be 32 ppb mercury in cord blood,
or an exposure of about 0.6 ug/kg-day. EPA used the various BMDLs and then applied an
uncertainty factor of 10 to account for interindividual toxicokinetic variability and
pharmacodynamic variability and uncertainty. On this basis, EPA defined the updated RfD of
0.0001 mg/kg bw/day in 2001. Although derived from a more complete data set and with a
somewhat different methodology, the current RfD is the same as the previous (1995) RfD. The
threshold is likely to be at or above the RfD. We discuss our assumptions of where the threshold
falls for benefit analysis later in this section.
Its worth noting that other risk assessors and regulatory agencies from around the world
use similar paradigms for benchmarking risks of noncancer health effects. Agencies use
reference dose (RfD), acceptable daily intake (ADI), tolerable daily intake (TDI) and minimal
risk level (MRL) as descriptors of their exposure benchmarks indicating acceptable exposure
levels without appreciable risks. These concepts, in general, assume a threshold for effect, a
level with no effect or acceptable effect.
How does EPA's RfD compare to other regulatory agencies benchmark for mercury risk?
Health Canada established its Tolerable Daily Intake (TDI) level at twice the EPA's RfD. Their
benchmark is .2 ug/kg bw/day. The Agency for Toxic Substances and Disease Registry
(ATSDR) has set a Minimal Risk Level (MRI) of .3 ug/kg bw/day - three times EPA's RfD
level. The World Health Organization's (WHO) benchmark is set at .23 ug/kg bw/day. Of these
major agencies, EPA's RfD has established the lowest risk benchmark to define levels of
exposure that are without appreciable risks.
11.3 Quantifying IQ Benefits Associated with Mercury Emission Reductions
To assess the IQ benefits from reduced mercury exposure, we must have four pieces of
information or data elements:
(1) The reduction in exposure from the regulatory option under investigation
(2) Assuming a linear dose-response curve, we need the slope of that curve.
(3) The level (if any) of the threshold where the dose-response curve intersects the x axis.
(4) If a threshold does exist, we also need to know the baseline exposure levels from all
sources.5
5 This baseline exposure level is necessary to know who is above the threshold. Exposure reductions for those
individuals already below the threshold would not generate benefits. Hence, we need to know each fish consumer's
methymercury exposure from all sources to ascertain whether they are above or below the threshold.
11-3
-------
Exposure reductions [data element (1)] are derived in Chapter 10. The dose-response curve
[data element (2)] is also discussed in a previous chapter. Our core analysis will use a primary
does-response curve that implies an individual will avoid an IQ point decrement for every 1
ppm hair mercury reduction.
Data elements (3) and (4) require more discussion.
11.4 Data Element (3) - The Level of the Threshold
The preceding discussion of EPA's RfD does suggest we should model a threshold for
calculating IQ benefits. However, as noted above, the RfD is a level of exposure without
appreciable risks. It is plausible that the actual threshold could be higher than the RfD, it is
unlikely to be lower. In the absence of data or a compelling biological rationale to suggest a
threshold level, we will estimate IQ benefits using a variety of threshold assumptions. We will
use some of the benchmark levels of exposure established by regulatory agencies (discussed
above) as possible thresholds. Specifically, we will simulate benefits assuming: (1) a threshold
equal to EPA's RfD and (2) a threshold in the neighborhood of the WHO and Health Canada
benchmarks of .23 and .2 ug/kg bw/day respectively.
11.5 Date Element (4) - The Baseline Levels of Exposure for Consumers of
Recreational-Caught Fish
To quantify the IQ benefits from the methyl mercury exposure reductions, we need to
know the initial methylmercury exposure levels. If the mercury exposure of a woman eating
recreationally-caught fish is below the proposed threshold, then further reductions would not
yield IQ improvements. If the fish consumers were above the RfD, then IQ improvements would
be expected from reductions in methylmercury exposure. Threshold models require information
on the joint distribution of exposure from all sources and much more precise information about
fish consumption patterns.
We also must account for the people currently above the RfD (due to all sources of
methylmercury exposure) that would be taken below the RfD due to the rule. This would be
very difficult to do because we would need the background exposure from all sources to
determine who is currently exposed at a level above the RfD. For example, assume that there is
an individual who is currently exposed at a level just over the RfD, but only a small part of this
is due to the power plants. Looking at the change due to the rule would show a very small
change in mercury exposure for this individual, but it could actually push him below the RfD
threshold.
Unfortunately, EPA does not have the empirical information necessary to statistically
link recreational fishing populations to baseline methylmercury levels. EPA's Notice of Data
Availability (EPA 2004) discussed our interest in obtaining additional information on this issue
and asked for any available information. We did not receive information that would help us
estimate the joint probability distribution of recreational fish consumption and commercial fish
consumption sufficient to quantify baseline mercury exposures. Although we have some
information on the population at large, we do not have similar information for recreational
anglers.
11-4
-------
Hence, we are left without sufficient data to address this issue in an ideal manner. The
technical approach for dealing with this data gap is addressed in this chapter below.
The remainder of this chapter derives and presents our estimates of IQ benefits.
11.6 Overview of Benefits Methodology
We make full use of the exposure estimates and IQ loss benefits calculated for the linear,
no threshold case for each of our regulatory options presented in Chapter 10 to derive benefit
estimates for our core threshold scenarios. The analytic steps we take are:
(1) Using the results from Chapter 10 for the Zero Out scenario, place each block group
(described in Chapter 10) into categories or bins depending on the size of their exposure
reduction (IQ reduction). By establishing bins or discrete categories, the computations,
particularly for the uncertainty analysis are simplified greatly. The zero out scenario
provides us with an estimate of the total mercury exposure due solely to electric utilities.
(2) Assign an average exposure reduction to and IQ losses for each bin (discrete category).
(3) Perform similar classifications and calculations for regulatory options.
(4) Assign each bin a "background" exposure based on blood mercury distribution from the
1999-2002 National Health and Nutrition Examination Survey (NHANES), assuming of
course that the mercury from eating recreationally-caught food is a lower bound on total
mercury exposure for any individual. This provides a distribution of total mercury
exposure for the recreational fishing population.
(5) Impose thresholds (discussed above)
(6) Assess exposure reductions for each of the regulatory options for all bins above
threshold.
(7) Scale benefits from the no-threshold case (presented in Chapter 10) to reflect only those
exposure reductions that occur to people above the threshold.
11.7 Freshwater Fish Mercury Exposure
The model in Chapter 10 used 165,000 block groups to produce estimates of mercury
exposure and the associated monetized benefits of avoided IQ decrements due to the rule.
While, in principle, these block groups could be used to estimate the benefits for individuals
above a threshold, a simpler approach can be used relying on a small number discrete interval
categories or "bins" of data. In other words, we use the number of individuals experiencing an
IQ decrement within a small range, as was illustrated in [Figure 11-11 from the draft Chapter 11
on 2-17-05]. The advantage to using a small number of bins is that an uncertainty analysis can
be conducted to produce a range of estimates which would be much more difficult to conduct
using the full sample.
11-5
-------
The 2020 baseline results from the Population Centroid Model described in Chapter 10
provides an estimate of the average mercury concentration in maternal hair (ppm) from
freshwater fish consumption for 481,987 individuals in the baseline. For purposes of producing
bins, mercury concentrations are obtained using the mid point of IQ decrements in 0.1 intervals
and the inverse of equation [EQ 11-18 from the draft Chapter 11 on 2-17-05]. That is
CHgHj =d!Qj 70.131
(HEq. 1)
where
CHgH = average mercury concentration in maternal hair (ppm) for bin j;
j = mid point of the IQ decrement interval for bin j.
These data are describe in Table 1 1-1.
Table 11-1. Freshwater Fish IQ Loss and Hair Mercury from the 2020 Baseline
Number of Individuals
416,844.83
45,666.60
11,157.90
4,095.02
1,870.15
899.31
432.62
339.36
206.28
176.41
74.69
68.73
15.76
23.49
31.21
22.41
9.77
11.20
17.96
1.56
0.92
3.84
5.07
1.35
2.17
1.39
1.00
3.97
2.44
IQ Loss Range
0-0.1
0.1-0.2
0.2 - 0.3
0.3 - 0.4
0.4 - 0.5
0.5 - 0.6
0.6-0.7
0.7-0.8
0.8-0.9
0.9-1
1-1.1
.1- .2
.2- .3
.3- .4
.4- .5
.5- .6
.6- .7
.7- .8
.8- .9
1.9-2
2-2.1
2.1-2.2
2.2 - 2.3
2.3 - 2.4
2.5-2.6
3-3.1
4.9-5
5.1-5.2
5.7-5.8
IQ Loss Mid Point
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
.05
.15
.25
.35
.45
.55
.65
.75
1.85
1.95
2.05
2.15
2.25
2.35
2.55
3.05
4.95
5.15
5.75
Hair Mercury (ppm)
0.385
1.154
1.923
2.692
3.462
4.231
5.000
5.769
6.538
7.308
8.077
8.846
9.615
10.385
11.154
11.923
12.692
13.462
14.231
15.000
15.769
16.538
17.308
18.077
19.615
23.462
38.077
39.615
44.231
The 2020 baseline results can be compared to three scenarios
11-6
-------
1. A hypothetical "zero out" scenario, in which mercury emissions from coal-fired utilities
are completely eliminated in 2020. This provide an estimate of the full range of potential
IQ-related benefits from eliminating EGU-attributable mercury contamination of
freshwater fish.
2. Option 1, in which mercury emissions from coal-fired utilities are capped at 38 tons per
year in 2010 and capped at 15 tons per year in 2018.
3. Option 2, in which mercury emissions from coal-fired utilities are capped at 38 tons per
year in 2010 and capped at 15 tons per year in 2015.
Due to the fact that the analysis is being done using data in bins, these three scenarios are
reported as change in mercury levels. These changes are generated by comparing the IQ
decrements from the baseline and the particular scenario under consideration. The difference in
the IQ decrements are then place in a small number discrete interval categories, or bins, with the
number of individuals in each change category. These changes can then be applied to the
number of individuals in each exposure category to obtain the estimated monetized benefit of
avoided IQ decrements. These data are reported in Table 11-2, 11-3, and 11-4 for the three
scenarios.
Table 11-2. Change in IQ Loss and Hair Mercury from the 2020 Zero Out (No Threshold)
Compared to the Baseline, and the Relative Probability of each Change Category
Number of Individuals
IQ Loss Range
Hair Mercury Range Probability
477,526
3,640
537
186
53
16
10
9
2
1
4
2
0 - 0.025
0.025 - 0.05
0.05 - 0.075
0.075 - 0.1
0.1-0.125
0.125-0.15
0.15-0.175
0.175-0.2
0.2 - 0.225
0.225 - 0.25
0.4 - 0.425
0.475 - 0.5
0-0.192
0.192-0.385
0.385 - 0.577
0.577 - 0.769
0.769 - 0.962
0.962-1.154
1.154-1.346
1.346-1.538
1.538-1.731
1.731 - 1.923
3.077 - 3.269
3.654 - 3.846
99.0744%
0.7552%
0.1115%
0.0386%
0.0111%
0.0033%
0.0021%
0.0019%
0.0004%
0.0002%
0.0008%
0.0005%
11-7
-------
Table 11-3. Change in IQ Loss and Hair Mercury from the CAMR Option 1 (No
Threshold) Compared to the Baseline, and the Relative Probability of each Change
Category
Number of Individuals IQ Loss Range Hair Mercury Range
481,555
361
48
12
4
1
2
4
0 - 0.025
0.025 - 0.05
0.05 - 0.075
0.075 - 0.1
0.1-0.125
0.125-0.15
0.15-0.175
0.275 - 0.3
0-0.192
0.192-0.385
0.385 - 0.577
0.577 - 0.769
0.769 - 0.962
0.962-1.154
1.154-1.346
2.115-2.308
Probability
99.910%
0.075%
0.010%
0.002%
0.001%
0.000%
0.000%
0.001%
Table 11-4. Change in IQ Loss and Hair Mercury from the CAMR Option 2 (No
Threshold) Compared to the Baseline, and the Relative Probability of each Change
Category
Number of Individuals
IQ Loss Range
Hair Mercury Range Probability
481,324
558
72
22
2
4
2
4
0 - 0.025
0.025 - 0.05
0.05 - 0.075
0.075-0.1
0.1-0.125
0.125-0.15
0.15-0.175
0.3 - 0.325
0-0.192
0.192-0.385
0.385-0.577
0.577 - 0.769
0.769 - 0.962
0.962-1.154
1.154-1.346
2.308-2.5
99.862%
0.116%
0.015%
0.004%
0.000%
0.001%
0.000%
0.001%
11.8 Deriving Baseline Mercury Exposures from All Sources of Mercury
For this analysis, we assumed that the NHANES data provided coverage for recreational
freshwater anglers (e.g., sampling of individuals from the general US population for inclusion
within NHANES, include some fraction of recreational freshwater anglers). However, we are
unable to identify the specific location of recreational anglers within the full NHANES
distribution for purposes of establishing total mercury exposure for those modeled individuals.
Therefore, we had to make certain assumptions in "relating" the recreational anglers modeled for
the RIA to the NHANES distribution for purposes of predicting total mercury exposure for this
population. Specifically, we assumed that the least exposed recreational angler would at least
have total mercury exposure equal to the mercury exposure level they receive from the
self-caught freshwater pathway (i.e., assuming they have zero background or commercial fish
exposure). This is a reasonable lower-bound estimate of total mercury exposure for the
recreational anglers modeled for the RIA and therefore represents a reasonable basis for
establishing the lower bound of total recreational angler exposure within the NHANES
distribution. The next step is to determine how recreational anglers are distributed across
percentiles of the NHANES distribution above that lower bound percentile
11-8
-------
The NHANES data used was provided in percentiles, with an average blood mercury
concentration for each 1 percent of the exposed population. The blood mercury level (in ppb)
was converted to a hair mercury level (in ppm) using the hairblood ratio of 200, as described in
Chapter 9. This implies that hair mercury levels (in ppm) are one-fifth of blood mercury levels
(in ppb). The lowest hair mercury level in Table 11-1 is 0.385 ppm. This represents the mercury
exposure from freshwater fish alone. Since the total exposure must be above this level, these
individuals must be associated with a NHANES percentile with a mercury concentration higher
than 0.385 ppm. The lowest NHANES percentile above this level is the 77th percentile, with a
value of 0.4 ppm. Therefore, the 481,987 individuals in the 2020 baseline are assumed to be
evenly distributed over the 77th through the 100th percentile (24 bins) in the NHANES data.6 The
exact number of individuals in table 11-1 is 481,987.41. Divided by 24 implies 20,082.81 in
each NHANES percentile.7
The mercury exposure from freshwater fish and total mercury exposure is estimated by
combining the data from Table 11-1 and the NHANES data. These are listed in Table 11-5. The
sum of individual for any freshwater fish exposure level must equal the number of individuals in
Table 11-1 for that exposure. The sum of individuals for any total exposure level must equal the
number of individuals in the NHANES percentile. In the upper tail of the freshwater fish
consumption, the freshwater fish exposure exceeds the top hair mercury level from the NHANES
data. For these individuals, the freshwater fish mercury exposure is also taken to be their total
exposure value.
6 These results suggest that the lowest modeled recreational fisher has total mercury exposure matching the 77th%
US resident. In reality, if the threshold analysis had been conducted using the fully disaggregation set of 481,987
modeled individuals described in Chapter 10, the least exposed recreational angler would have been matched to a
much lower NHANES percentile (probably less than the median). Note, however, that the mean for the recreational
anglers would likely be larger than the NHANES mean and probably close to the 77th%, which is plausible given
that recreational anglers consume self-caught fish in addition to their commercial fish exposure and therefore may
reasonably be expected to have total fish consumption above the general population. The use of aggregated
(clustered) data in the threshold analysis, while making the analysis tractable, did remove much of the
inter-individual variability in modeled exposure from self-caught fish, which adds additional uncertainty into the
threshold analysis both in terms of (a) predicting the number of individuals exceeding the threshold of concern due
to total exposure and (b) estimating the fraction of total benefits exceeding those same thresholds. It is also
important to note that the assumption that recreational anglers are evenly distributed across the 77th-100th
percentiles of the NHANES distribution is also subject to uncertainty. Compelling arguments can be made that this
assumption is either over- and under-conservative. Ultimately, without identifying data specifically defining the
relationship between self-caught fish and commercial fish consumption for the recreational angler population, any
assumptions regarding this relationship is subject to considerable uncertainty.
7 We note that the conversion from blood mercury levels (in ppb) to hair mercury levels (in ppm) using a 5:1 ratio
might produce higher values for hair mercury in the upper tail of the distribution than is found in the actual
NHANES data on hair mercury levels. Since the NHANES data use for this analysis included only blood mercury
levels, however, this type of conversion was necessary and is a common technique. If we were to try to account for
this potential over-estimate, we might assume that individuals are at some lower percentile (say, the 50th) of the
NHANES distribution and then impose the shape of the distribution beyond this point for total exposure. This would
likely force more individuals into the lower exposure range and reduce the subsequent scaling factor. This type of
sensitivity analysis, however, was not done.
11-9
-------
Table 11-5. Joint Distribution of Mercury Exposure from Freshwater Fish and Total
Mercury Exposure
Number of Individuals
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
20082.81
15188.66
4894.15
20082.81
20082.81
606.83
11157.90
4095.02
1870.15
899.31
432.62
339.36
206.28
176.41
74.69
68.73
15.76
23.49
31.21
22.41
9.77
11.20
17.96
1.56
0.92
3.84
5.07
Freshwater Fish Exposure
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
0.385
1.154
1.154
1.154
1.154
1.923
2.692
3.462
4.231
5.000
5.769
6.538
7.308
8.077
8.846
9.615
10.385
11.154
11.923
12.692
13.462
14.231
15.000
15.769
16.538
17.308
Total Exposure
0.400
0.420
0.420
0.440
0.460
0.480
0.520
0.540
0.560
0.600
0.640
0.680
0.740
0.780
0.860
0.920
0.980
1.100
1.220
1.360
1.540
1.540
1.980
2.460
7.780
7.780
7.780
7.780
7.780
7.780
7.780
7.780
7.780
8.077
8.846
9.615
10.385
11.154
11.923
12.692
13.462
14.231
15.000
15.769
16.538
17.308
11-10
-------
Number of Individuals
1.35
2.17
1.39
1.00
3.97
2.44
Freshwater Fish Exposure
18.077
19.615
23.462
38.077
39.615
44.231
Total Exposure
18.077
19.615
23.462
38.077
39.615
44.231
Thresholds
The introductory sections of this chapter present background on our RfD, other agencies'
risk benchmarks and the nature of thresholds in risk assessment. As discussed in these sections,
we use the two benchmark levels of exposure as possible thresholds.
a. 0.1 ng/kg bw/day - A threshold equal to EPA's RfD
b. 0.2 ng/kg-day - A threshold in the neighborhood of the WHO and Health
Canada benchmarks of .23 and .2 ng/kg bw/day respectively.
These were converted to ppm of hair mercury based on Eq. 2 (11 - Eq. 11.17 from the draft
Chapter 11 on 2-17-05] in Chapter 10. Specifically,
CHgH = (0.08)'1 * Hglkg (11 Eq. 2)
where
CHgH = average mercury concentration in maternal hair (ppm)
Hglkg = average daily mercury ingestion rate (mg/kg bw/day);
This conversion rate between average daily ingestion rate and maternal hair concentration is
based on the one compartment model used by Swartout and Rice (2000) as described in Chapter
10. This implies the two thresholds of 1.25 ppm and 2.5 ppm.
As discussed above, there are other potential thresholds that could be considered. For
example, from the Agency for Toxic Substances and Disease. These are not considered here
because of the nature of the discrete nature of the NHANES data in percentile form. The 99th
percentile of hair mercury in the NHANES data is 2.46 ppm. The 100th percentile data is 7.78
ppm. Because of this analysis is being conducted using data bins, the choice of any threshold
between 2.46 and 7.78 ppm would produce the same results. Therefore, consideration of the
benchmark values from these other agencies would produce the same results as our second
threshold above.
11.9 Deriving Scaling Factors
Using the data from Table 11-5; the changes in Tables 11-2, 11-3, and 11-4 and the
thresholds described above, it is possible to evaluate the effect of our two thresholds. Because of
11-11
-------
a number of uncertainties, including the level of total exposure, this evaluation is best conducted
using a Monte Carlo simulation.
In a Monte Carlo simulation a model is run many times using different values for
uncertain input parameters. Prior to the simulation, each uncertain parameter is assigned a
probability distribution. For each iteration of the simulation, a random input value is chosen for
each uncertain parameter, using the assigned distribution and a realization of the output value is
calculated. The output value is recorded and another iteration with a different random choice of
input values is conducted. After a very large number of iterations, a probability distribution can
be developed for the output value.
The analysis in Chapter 10 calculates the change in mercury exposure due to this rule.
What is unknown is the total exposure level of the individuals experiencing this change. To
address this uncertainty, we conduct a Monte Carlo simulation. In this case, we treat the change
in exposure as the uncertain input variable. We assume that the distribution in the change in
exposure is equal to the change in exposure from the Population Centroid Model results from
Chapter 10. For example, for the results from CAMR, a separate change in exposure is
randomly chosen for each data bin using the probabilities in 11-3. This change is then applied to
data bin listed in Table 11-5. For the scenarios in which a threshold is applied, change in
exposure is assumed to not occur for any data bin in which the total exposure is below the
threshold. The resulting distribution of the decision variable can then be used to determine the
scaling factors for each scenario.
This approach assumes zero correlation between the total mercury exposure and the
change in mercury exposure as a consequence of the regulatory option. However, the individual
level results from Chapter 10 suggest a high degree of correlation between the mercury exposure
self-caught fish and the changes in mercury exposure. This means that the zero correlation
assumption above could understate the scaling factor because the individuals experiencing the
highest change are likely above the thresholds. To account for this potential underestimate, a
second scaling factor is developed assuming perfect correlation between total mercury exposure
and the changes in mercury exposure. Since the changes listed in Tables 11-2, 11-3, and 11-4
are given in ranges, a uniform distribution was assumed, so that each data bin in Table 11-5 is
associated with a progressively larger change in mercury exposure from the regulatory option.
The scaling factors are listed in Table 11-6. The No Threshold scenario is the reference
value, and has a scaling value of 100%. The lower bound is based on the mean value from the
Monte Carlo simulation. The upper bound is based on the perfect correlation scenario. Due to
the fact that over 99% of the changes for all three scenarios occur in the lowest change category
(a change of 0.0 - 0.192 ppm in hair mercury), there is no difference in the scaling factors
between regulatory scenarios. These values can be used to scale the detailed benefits estimates
from Chapter 10.
Table 11-6. Scaling Factors
Zero Out Option 1 Option 2
rru u i^EP^iRfD/u u IA ^ 21-34% 21-34% 21-34%
(Threshold = 0.1 jig/kg bw/day)
11-12
-------
WHO/Health Canada „„
(Threshold = 0.2- 0.23 ng/kgbw/day) 4'8/° 4'8/° 4'8/°
No Threshold 100% 100% 100%
11.10 Monetization and Scaling of IQ Benefits
The IQ decrements associated with a change in mercury exposure are derived from the
Dose-Response curve described in Chapter 9. The slope of that dose response curve is 0.13 1, so
each 1 ppm in hair mercury is assumed to cause a decrement of 0.131 of an IQ point. The
monetary value of losses resulting from IQ decrements are assessed in terms of foregone future
earnings for the affected individuals and described in Chapter 10. These monetary value of these
losses were estimated to be $8,807 (in 1999$) per IQ point lost. Therefore, the monetized
benefit from a decrease in hair mercury is calculated as follows:
j * VIQ * (0.131 * dCHgHj) (1 1 - Eq.-3)
where
VN = Value of avoiding IQ loss for N; individuals in bin i;
N; = Number of individuals in bin i;
VIQ = Value per change in IQ point decrement (= $8,807 in 1999$);
dCHgH = Change in mercury concentration in maternal hair (ppm) for bin i.
The dose-response slope coefficient of 0.131 was modeled assuming a no threshold case,
in which the dose-response line passes through the origin, and a zero IQ response only occurs
with a zero dose. If there were a threshold, then ideally, we would reestimate the dose-response
curve with the threshold assumption incorporated into the estimation procedure. If this was
done, the new linear dose response curve would have a steeper slope than the one estimated
under a no threshold case. This would affect the monetized benefits value from equation (11
Eq.-3). Unfortunately, we do not have the original data to reestimate the dose-response with a
threshold assumption. However, since the coefficient enters this equation linearly, the benefits
would simply be increased by the same proportion as the increase in the dose-response
coefficient.
The benefits from Chapter 10 are used as inputs to our model to estimate benefits under
our threshold cases. The monetary benefits and reduction in IQ point decrements can be
obtained for the two thresholds cases by multiplying the scaling factor from Table 11-6 times
these benefits. Tables 1 1-7 and 11-8 present our results for the two threshold models and for the
no threshold case. The benefits are presented in order of increasing uncertainty. We are most
certain of the benefits at high doses, given that the original data from the developmental studies
fall mostly at high doses. Accordingly, the benefits for the highest threshold are most certain,
followed by the lower threshold equal to EPA's RiD. The benefits estimated at exposure levels
below the RfD are the most uncertain. The derivation of the unit value for each IQ point is
discussed in Chapter 10. The monetized benefit of CAMR Option 1 over the baseline range
from $.07 million to $2 million a year. For Option 2, the benefits range from $.1 million to $4.6
million per year, depending on the dose-response model and the choice of discount rates. The
11-13
-------
tables also present "discounted" IQ points (IQ points gained discounted to present values at 3
and 7 percent.
Table 11-7. IQ Benefits for CAMR Option 1 under Established Health-Based Benchmarks
Uncertainty
Regarding
Threshold
More Certain
Less Certain
Benchmark
Source
WHO/
Health
Canada
EPARfD
No
Threshold
Level of Threshold
0.2 - 0.23 jig/kg
bw/day
0.1 ng/kg bw/day
N/A
Discount
Rate
3%
7%
3%
7%
3%
7%
Scaling
4%
8%
4%
8%
21%
34%
21%
34%
100%
100%
Benefits (millions
1999$)
$0.07 - $0.12
$0.14 -$0.24
$0.03 - $0.08
$0.06 -$0.16
$0.36 - $0.63
$0.58 -$1.0
$0.17 -$0.42
$0.27 - $0.68
$1.7 -$3.0
$0.8 - $2.0
Discounted
IQ Points
8-14
15-27
4-9
7-18
41-72
66- 116
19-48
31 -77
193-341
91 - 277
Table 11-8. IQ Benefits for CAMR Option 2 under Established Health-Based Benchmarks
Uncertainty
Regarding
Threshold
More Certain
Less Certain
Benchmark
Source
WHO/
Health
Canada
EPARfD
No
Threshold
Level of Threshold
0.2 - 0.23 jig/kg
bw/day
0.1 ng/kg bw/day
N/A
Discount
Rate
3%
7%
3%
7%
3%
7%
Scaling
4%
8%
4%
8%
21%
34%
21%
34%
100%
100%
Benefits (millions
1999$)
$0.1 -$0.18
$0.2 - $0.37
$0.05 -$0.1 2
$0.1 -$0.25
$0.53 - $0.97
$0.85 -$1.56
$0.25 - $0.65
$0.41 -$1.05
$2.5 - $4.6
$1.2 -$3.1
Discounted
IQ Points
11-21
23-42
5-14
11-28
60-110
97 - 178
29-74
46 - 120
284 - 522
136-352
11-14
-------
11.11 Uncertainties
This benchmark analysis is subject to uncertainty reflecting a number of factors. The
first is the choice of the threshold value. We simply do not have enough data to test whether a
threshold does or does not exist. The use of the RfD represents a best estimate of the population
threshold. Another area of uncertainty is that the lack of data characterizing the correlation
between self-caught fish consumption and commercial fish consumption in determining total fish
consumption for recreational freshwater anglers, prevents a truly representative analysis of total
mercury exposure for this population. The use of aggregated (clustered) exposure estimates for
recreational freshwater anglers rather than the full set of block group-level results generated for
the RIA introduces uncertainty into (a) the estimate of the number of individuals exceeding a
given benchmark due to total mercury exposure and (b) the fraction of total Option 1 and Option
2 benefits exceeding that threshold. Finally, failure to consider an upward adjustment to the IQ
loss function under assumptions of various thresholds considered in this analysis resulted in an
under-prediction of benefits above the threshold.
11.12 Conclusions
This chapter presents modeled estimates of IQ benefits for three scenarios. (1) A
threshold equal to EPA's Rfd of 0.1 ug/kg bw/day; (2) a threshold at a level near the Health
Canada and World Health Organization's benchmark levels of 0.2 ug/kg bw/day and 0.23 ug/kg
bw/day respectively; and (3) a no threshold benchmark. The chapter presents IQ benefit
estimates for the CAMR Option 1 and CAMR Option 2 that are incremental to the expected
reduction in mercury emissions under CAIR.
We are more confident of the likelihood of effects at higher exposures (i.e., above the
WHO/ Health Canada value) because these exposure levels are closer to the observed level of
exposures in the three underlying studies. However, if the threshold were set above the RfD
level, at a level equivalent to the WHO level, we would be ignoring the potential health effects
that might be gained from exposures above the EPA RfD. On the other hand, we are less certain
of the benefits below the RfD in the no threshold analysis since the RfD identifies a level below
which there is no appreciable risk of deleterious effects. Based on the NAS review of the science
and their endorsement of the RfD value as being scientifically justifiable for the protection of
public health, EPA has adopted the RfD as an appropriate construct for expressing the level of
daily exposure which is likely to be without an appreciable risk of deleterious effects during a
lifetime.
Under the EPA's RfD threshold scenario, we found that under CAMR Option 1, total IQ
points gained were between 19 and 116 IQ points for children of recreationally-caught fish
consumers. These IQ points were valued at a total benefit of $0.25 - $1.56 million.
We also simulated a threshold at a level near the Health Canada and World Health
Organization benchmark levels of .2 ug/kg bw/day and .23 ug/kg bw/day respectively. The
table indicates that benefits are indeed sensitive to the modeled thresholds. Going from the EPA
threshold to the higher threshold suggested by the WHO and Health Canada reduced benefits by
19 to 24 percent.
11-15
-------
Finally, not surprisingly, the no threshold case - the more uncertain of the estimates - has
the largest benefits - roughly four times the benefits as the EPA RfD threshold case.
11.13 References
Bellinger DC (2005). Neurobehavioral Assessments Conducted in the New Zealand, Faroe
Islands, and Seychelles Islands Studies of Methylmercury Neurotoxicity in Children.
Report to the U.S. Environmental Protection Agency.
NRC (2000). Toxicological Effects of Methylmercury, National Research Council. Washington,
DC. National Academies Press.
Ryan, LM (2005). Effects of Prenatal Methylmercury on Childhood IQ: A Synthesis of Three
Studies. Report to the U.S. Environmental Protection Agency
Swartout, J., and G. Rice. 2000. "Uncertainty Analysis of the Estimated Ingestion Rates Used to
Derive the Methylmercury Reference Dose." Drug and Chemical Toxicology 23(1):293-
306.11-41
US EPA. 2001. Integrated Risk Information System: Methylmercury (MeHg) (CASRN
22967-92-6). Downloaded on 2/25/05 from http://www.epa.gov/iris/subst/0073.htm
U.S. Environmental Protection Agency (EPA). 2002. Mercury Neurotoxicity Workshop Notes.
Washington, DC. November 4, 2002. http://www.epa.gov/ttn/ecas/regdata/
Benefits/mercuryworkshop.pdf.
US EPA. 2004. Proposed National Emission Standards for Hazardous Air Pollutants; and, in
the Alternative, Proposed Standards of Performance for New and Existing Stationary
Sources, Electric Utility Steam Generating Units: Notice of Data Availability. Federal
Register / Vol. 69, No. 230 / Wednesday, December 1, 2004 / Proposed Rules.
11-16
-------
SECTION 12 CO-BENEFITS RESULTING FROM REDUCTIONS IN
EMISSIONS OF PM2.5 12-1
12.1 Introduction 12-1
12.2 Emissions Modeling 12-3
12.3 Air Quality Modeling and Population-Level Exposure Estimation 12-4
12.4 Modeling Changes in Health Endpoint (Mortality) Incidence 12-5
12.5 Valuation of Benefits 12-7
12.6 Presentation of Results 12-8
12.7 Discussion of Uncertainties 12-8
12.8 References 12-11
Tables
Table 12-1. PM2.5 Co-Benefits Associated with CAMR Regulatory Options 1 and 2
in 2020 12-8
-------
SECTION 12
CO-BENEFITS RESULTING FROM REDUCTIONS IN EMISSIONS OF PM2.5
12.1 Introduction
Emissions control strategies adopted by power plants to meet cap-and-trade regulations
implemented under CAMR are likely to result in co-benefits including reductions in direct
emissions of PM2.5. These PM emissions reductions could result in decreased population-level
exposure to PM2.5, which, in turn will produce reductions in adverse health effects including
both morbidity and premature mortality for the U.S. population. Because of limitations
associated with the Integrated Planning Model (IPM), EPA was not able to conduct a
comprehensive assessment of health benefits associated with reductions in directly-emitted
PM2.5 from coal-fired power plants (the IPM model, as currently configured, cannot project
changes in directly emitted PM2.5 for the technology configurations relevant to this regulation).
Instead, EPA conducted a illustrative analysis focused on direct PM2.5 and on the key health
endpoint; premature adult mortality. Despite the use of a simplified modeling approach, the
illustrative methodology used in evaluating the adult mortality endpoint for this rule is still
considered robust enough to provide perspective on the likely magnitude of PM-related co-
benefits of CAMR if Activated Carbon Injection with the addition of a polishing baghouse
(TOXECON"") is used. In conducting the illustrative analysis, EPA has used standard methods,
assumptions, and monetary values that are used in other EPA regulatory analyses and have
undergone significant peer review. See the CAIR for a detailed discussion of the derivation of
assumptions used in PM benefit analyses conducted by EPA (EPA, 2005).
It is important to note that this analysis does not consider a number of health endpoints
and welfare effects that would add to the overall co-benefits if modeled, including: (a) PM-
related morbidity (e.g., chronic bronchitis, hospital admissions for respiratory and cardiovascular
events, non-fatal myocardial infarctions), (b) infant mortality, and (c) welfare effects including
visibility improvements. In other analyses of PM2.5 benefits, the mortality endpoint has
typically accounted for 85-95% of the total quantified benefits.
In addition, this analysis does not consider potential co-benefits resulting from reductions
inprecursors to PM2.5, including S02. In the case of SO2, given the SO2 caps under CAIR, we
do not expect SO2 reductions beyond the CAIR levels under CAMR. There is also the potential
that compliance with mercury regulations will simply shift the time-course of SO2 emissions
changes, but not result in absolute changes in the magnitude of those emissions. Consequently,
co-benefits resulting from reductions in the secondary formation of PM2.5 were not modeled.
Options Evaluated for Co-Benefits
As is discussed in Section 7 of this report, two scenarios were considered in the
emissions modeling and thus modeled for the PM2.5 co-benefits analysis including (a) Option 1
(38 ton cap in 2010 and a 15 ton cap in 2018), and (b) Option 2 (38 ton cap in 2010 and aJ5 ton
cap in 2015). We also examined the effects of the possibility that advanced sorbents (i.e.,
advanced carbon injection (ACI)) technology are available earlier than anticipated (assumed
12-1
-------
available in 2013) with resulting reductions in PM2.5 co-benefits (see Section 7.5 for details on
projected development/application of ACI technology and EPA, 2005). Without advanced
sorbents, Option 2 yields the greatest PM2.5 co-benefits (5671 tons in 2020) due to less
opportunity for banking of mercury allowances by power plants which reduces the opportunity
to delay installation of mercury control equipment. Option 1 produces lower PM2.5 co-benefits
(1920 tons in 2020) due to greater potential for banking which increases the potential for plants
to delay installation of control equipment. With advanced sorbents, there are noticeably less
PM2.5 emissions reductions (67 tons in 2020) due to the assumption that advanced sorbent
technology will not require installation of additional bag houses (the source of the majority of
the PM2.5 co-benefits predicted for these scenarios).1
Overview of Methodology
The PM co-benefits illustrative analysis was conducting using the following step-wise
procedure:
The Integrated Planning Model (IPM) was used to estimate the fraction of coal-fired
power plants expected to install bag houses (and the time course for those installations) for each
of the regulatory scenarios under consideration. This information was then combined with (a)
information on the relevant characteristics of key coal types used by these facilities (i.e., heating
values and ash content) and (b) emissions factors for bag houses and electrostatic percipitators
installed at power plants, to generate estimates of direct PM2.5 emissions reductions for each
scenario.
The Source-Receptor Matrix (SR-MATRIX) model (E. H. Pechan and Associates, 1997),
was used to predict population-level changes in PM2.5 exposure given emissions reduction
estimates for direct PM2.5 from coal-fired power plants. The SR-MATRIX model uses a
simplified methodology in conducting both air quality modeling and exposure modeling. In a
full-scale benefits analysis, these two steps would have been conducted using more rigorous and
sophisticated models such as CMAQ (for air quality modeling) and BenMAP (for exposure
modeling).
Changes in population-level exposure to PM2.5 were then combined with baseline
incidence data for adult mortality and used together with the concentration-response function for
mortality derived from Pope et al., 2002, to estimate reductions in premature adult mortality.
The Pope et al., 2002 -based concentration response function has been used in other benefits
analysis including CAIR.
Adult mortality incidence reductions were then monetized using the value of statistical
life (VSL) metric of 5.5 million dollars per statistical life saved that is used in other EPA
regulatory analyses. We also incorporate other elements of the benefit methodology from CAIR
(EPA, 2005), including consideration for: (a) a mortality reduction lag of 20 years (i.e., the
1 Because of time and resource considerations, and because Option 1 was the preferred option for CAMR, we only
ran the advanced sorbents sensitivity analysis for Option 1, but we believe the results for Option 2 would be similar
to what is reported here for Option 1 (e.g., approximately 67 tons in 2020).
12-2
-------
decrease in mortality incidence is distributed over 20 years after the emissions change), (b)
income elasticity associated with the valuation function (i.e., degree to which willingness-to-pay
(WTP) for a reduction in mortality changes as income grows), (c) real growth in per-capita
income and (d) growth in the adult population cohort experiencing the mortality reduction.
Organization of this Section
The remainder of this section is organized as follows. Section 12.2 provides an overview
of emissions modeling conducted using IPM and other data sources in order to generate
estimates of emissions reductions in direct PM2.5 from power plants. Section 12.3 provides a
brief overview of the SR-MATRIX model and its role in air quality modeling and predicting
population-level changes in exposure for the study population. Section 12.4 describes the use of
the Pope et al., 2002 study in modeling long-term exposure-based mortality reductions linked to
reduced PM2.5 exposure. This section also describes the lag period used in modeling mortality
reductions. Section 12.5 describes the valuation of benefits including the VSL metric used, the
income elasticity and related factors. Section 12.6 presents the results of the illustrative analysis
of PM-related co-benefits. Finally, Section 12.7 describes some of the uncertainties related to
this analysis.
12.2 Emissions Modeling
Potentially, several different technologies could be selected by coal-fired power plants to
reduce mercury emissions including: (a) optimized existing emission controls (no direct PM2.5
co-benefit), (b) injection of halogenated or standard (non-halogenated) powdered activated
carbon (PAC) in front of an electrostatic precipitator (no direct PM co-benefit)2, (c) injection of
standard PAC in front of a retrofit polishing baghouse - TOXECON™ application (direct PM2.5
co-benefit), and (d) other technology (may or may not have an impact on direct PM2.5
emissions). In option "c" described above, a polishing baghouse is retrofitted in after an existing
electrostatic precipitator. Since the baghouse is added as an additional PM control and is
generally a more efficient collector of PM2.5 than an ESP, a direct PM2.5 co-benefit is likely in
this option. Therefore, the first step in estimating possible co-benefits-associated with reductions
in emissions of direct PM2.5, was to estimate the power plant capacity likely to select option
"c", which involves the addition of polishing baghouses with associated reductions in direct
PM2.5 emissions. Because different coal types (e.g., bituminous, subbituminous) are used by
power plants and these coal have different characteristics (e.g., ash content) that can impact
direct PM2.5 emissions, it is important to consider the distribution of these coal types across the
subset of power plants projected to install baghouses to control mercury. The IPM model, which
is a linear programming model used to predict the behavior of the U.S. electric utility sector in
response to air-pollution-control-related regulatory scenarios, was used to determine the coal-
specific capacities of coal-fired power plants likely to install baghouses in response to the two
regulatory scenarios under consideration
2 Concerns have been raised (EPA, 2005) that this option (injection of PAC in front of an electrostatic precipitator
(ESP) could actually result in dis-benefits by producing increased arcing in the ESP which can degrade ESP PM
capture performance. However, at this point it is unclear whether the rates of PAC injection likely to be utilized
under this scenario would produced sustained increases in arcing such that PM removal is significantly reduced.
Research is currently on-going looking at this issue in the context of longer-term facility performance (EPA, 2005).
12-3
-------
Once the capacities of power plants projected to install baghouses under the two
regulatory scenarios were known, the next step was to estimate reductions in direct PM2.5
emissions associated with these capacities. Calculation of reductions in direct PM2.5 emissions
involves combining data on (a) the utilization of the four coal types at power plants projected to
install baghouses (annual heat input values for each coal type obtained from IPM), (b) the
relevant attributes of those coal types, including heating values and ash content (from the
Information Collection Request (ICR) database), and (c) the PM2.5 emissions factors for
electrostatic precipitators and baghouses (from AP-42)3. For each regulatory scenario, the
annual heat input and the average heating values for a specific coal type were combined to
determine the amount of that coal type used by power plants of interest. This usage value was
then combined with the ash content value and emissions factors to generate the emissions
reduction estimate for that coal type. Such estimates are obtained for each coal type. These
estimates were then summed across the coal types to generate a single direct PM2.5 emissions
reduction value for coal-fired power plants under the specific regulatory scenario.
EPA used the currently available information on mercury'controls to develop the
estimates of PM2.5 co-benefits described above (Note: EPA projects that, under Option 1,13
units will have installed ACI by 2020 - see Section 7.5 for projections regarding installation of
ACI devices). It is recognized, however, that mercury control technologies are under vigorous
development and, therefore, control approaches other than those described above may be
implemented in the future. Such actions may result in different PM2.5 co-benefits compared to
estimates provided here.
12.3 Air Quality Modeling and Population-Level Exposure Estimation
The SR-MATRIX model (E. H. Pechan and Associates, 1997), uses county-to-county
transfer factors to predict changes in PM2.5 air concentrations resulting from reductions in
emissions of directly-emitted PM2.5 from (user-specified) source categories (Levi et al., 2003).
These transfer factors are generated using an adjusted version of the Climatological Regional
Dispersion Model (CRDM), which uses a sector-averaged dispersion model combined with
summaries of 1990 meteorological data to produce these transfer factors. SR-MATRIX can
model secondary formation of PM2.5 including nitrates and sulfates, however this functionality
was not used in this illustrative analysis. All sources in the SR-MATRIX model are divided into
four categories including three categories based on effective stack height and a forth category for
all area sources. The SR-MATRIX model has been shown to generate reasonable predictions of
population-level changes in exposure resulting from changes in directly emitted PM and
3 The PM2.5 emissions factors for electrostatic precipitators and baghouses are taken from Table 1.1.6 in AP-42,
Fifth Edition, Volume 1 Chapter 1: External Combustion Sources, available at
http://www.epa.gov/ttri/chiefyap42/ch01/final/c01s01.pdf. As is well known, emission factors are simply averages of
all available data of acceptable quality, and are generally assumed to be representative of long-term averages for all
facilities in the source category (i. e., a population average). In the absence of availability of specific data on direct
PM2.5 co-benefit associated with TOXECON applications, EPA chose to use PM2.5 emission factors. It is noted
however, that baghouse emission factor in AP-42 does not take into account any mercury-specific design changes
that may take place. For example, if the baghouse to capture mercury-impregnated sorbent, with particles larger than
PM2.5, was designed to use a fabric with a coarser weave (to keep pressure drop low), capture of PM2.5 in such a
baghouse may be lower than that indicated in the emission factor. This, in turn, will result in reduced direct PM2.5
co-benefit.
12-4
-------
consequently, this model was applied in modeling these direct PM-related co-benefits (Levy et
al., 2003).
As applied in this analysis, the SR-MATRIX model was used for converting the
estimates of reductions in direct PM2.5 emissions into changes in ambient PM2.5
concentrations. In modeling population-level exposure reductions, the SR-MATRIX generates
results in the form of population-weighted PM2.5-related exposure reductions. The model
provides estimates by region throughout the U.S. As with the benefit analysis of mercury
reductions presented in Section 10, the eastern-half of the U.S. is analyzed, which is identified
by the Midwest/Northeast (ME) and Southeast (SE) regions in the SR Matrix. We then averaged
the results in these regions to derive total PM benefits. A more comprehensive analysis of
power-plant related emissions reductions (and associated exposure reductions) conducted using
BenMAP, would have used a more spatially-refined geographic grid that tracked both ambient
PM2.5 reductions and associated exposure reductions with greater precision and specificity.
Note, however, that it is not known whether the simplified modeling approach used here results
in a net over- or under-prediction of population-level exposure since sources of uncertainty
associated with this illustrative approach can result in both over- and under-predictions of
ambient PM2.5 levels and population-level exposures. However, the more generalized approach
used here is considered reasonable for a illustrative analysis of potential benefits. The SR-
MATRIX model can be used for health effects incidence estimation and valuation, but the
version used in this analysis did not have the latest mortality functions (Pope et al., 2002) or the
latest valuation functions and consequently, both mortality incidence changes and valuation of
those incidence reductions were estimated outside of the model (as discussed in detail below).
12.4 Modeling Changes in Health Endpoint (Mortality) Incidence
Exposure to PM2.5 has been linked to a variety of morbidity endpoints (e.g., respiratory
and cardiovascular hospital admissions, chronic bronchitis events, asthma exacerbations) as well
as mortality (both child and adult) (NRC, 2002). However, the majority of monetized benefits
linked to health endpoints are associated with chronic (long-term exposure-related) mortality in
adults. Consequently, this illustrative analysis focuses on this endpoint exclusively.
Both long- and short-term exposure to PM2.5 has been associated with mortality in adults
(NRC, 2002). However, long-term exposure cohort studies, which are better able to capture the
full public health impact of PM2.5 exposure over time and likely cover some of the shorter-term
exposure mortality signal together with the longer-term exposure signal, have typically found
higher levels of mortality risk than shorter-duration studies (Kunzli et al., 2001, NRC, 2002).
Consequently, this illustrative analysis evaluated chronic exposure-related mortality in adults.
Long-term mortality studies examine the relationship between community-level exposure
to PM2.5 over multiple years and annual mortality rates, also recorded at the community-level
(NRC, 2002). More recently-conducted prospective cohort studies allow for better control for
key confounders associated with mortality including individual-level information on key risk-
related factors such as diet, occupational exposure and smoking. The EPA's Science Advisory
Board recommends the use of long-term prospective cohort studies in estimating PM-related
mortality (EPA-SAB-Council-ADV-99-005, 1999), a finding that was confirmed by the recent
National Research Council Report (NRC, 2002).
12-5
-------
The prospective cohort study selected for this illustrative analysis (Pope et al, 2002), has
the broadest geographical coverage of the available prospective cohort studies examining PM-
related mortality and uses the American Cancer Society's (ACS) dataset which has been
subjected to extensive reexamination and reanalysis, and which has served to strengthen the
confidence associated with its findings. The latest reanalysis of the ACS study data conducted by
Pope (2002) provides additional refinements including (a) an extended follow-up period that
triples the size of the mortality data set, (b) significant increase in the exposure dataset, including
consideration for cohort exposure following implementation of the PM2.5 standard, (c) greater
control for possible confounders (diet and workplace exposure) and (d) use of advanced
statistical techniques to address concerns over spatial autocorrelation of survival times in
communities located near each other. Both the NRC and SAB-HES recommended the use the
Pope et al., 2002 study as the basis for modeling adult mortality linked to PM2.5 exposure as
part of economic benefits analysis. The "all-cause" mortality category was selected as the
endpoint modeled using this study.
Addressing Possible Mortality Reduction Lag
It is expected that reduction in ambient PM2.5 levels will produce a decrease in mortality
that does not occur immediately, but is distributed over some number of years (i.e., lagged over
time following the PM reduction) (NRC, 2002). However, the exact nature of the lag associated
with PM-related mortality reductions is not currently known. Consideration of mortality-related
lag periods is important since delays in reductions in mortality rates following PM reductions
will result in discounting (lowering) of overall monetary benefits associated with those mortality
reductions. Given limited information available for establishing a lag structure for PM-related
mortality reductions, the SAB-HES has recommended the following provisional 20-year lag
structure: 30% in the first year, 50% distributed over years 2-5 and 20% distributed over years 6-
20. This structure reflects the following perspective towards PM-related mortality reduction:
short-term exposure related reductions are assumed to occur in the first year, cardiovascular-
related mortality reductions are assumed to occur in years 2-5 and longer-term respiratory as
well as lung cancer mortality reductions are assumed to occur in years 5-20.
Consideration of Population Growth
The regulatory analysis (simulation) year for this illustrative analysis is 2020 for both
Option 1 and Option 2 and for the Sensitivity Analysis scenario. Demographic change (growth
in the adult-age population) needs to be considered in estimating mortality reductions for this
future simulation year. Typically, in a comprehensive benefits analysis conducted using
BenMAP, demographic growth would be projected separately for each geographic unit of
analysis (e.g., county, or US Census block) for which mortality reductions are being projected.
However, this illustrative analysis, does not include this level of spatial resolution and
consequently, a simple demographic scaling ratio was being used to adjust mortality estimates to
reflect potential growth in the adult-age cohort over the years leading up to the simulation year.
This demographic scaling factor was simply the ratio of US adult population (>29yrs) predicted
for the year 2020 divided by the adult population in 1999 (the year modeled in SR-MATRIX).
Note, that because both Options being considered are modeled for the same simulation year, the
same demographic scaling ratio was used in modeling each regulatory option.
12-6
-------
Addressing Potential Uncertainty in the Mortality Function
The EPA is currently investigating uncertainty associated with the concentration-
response function for chronic-duration PM2.5 mortality. A number of methods are being
considered, including the use of expert elicitation to develop quantitative assessments of overall
uncertainty associated with PM2.5-related mortality estimates. Because this analysis was
conducted as an illustrative analysis, consideration of uncertainty in the mortality estimate was
not quantitatively modeled. Section 12.6 provides a summary of the potential magnitude of
uncertainty surrounding the mortality estimate based on benefit analysis of CAIR. The reader is
referred to the CAIR RIA (EPA, 2005) for an in-depth discussion of uncertainty associated with
PM2.5-related mortality estimates and ongoing efforts to characterize that uncertainty.
12.5 Valuation of Benefits
Valuation of benefits involved monetizing of the reduction in adult premature mortality
associated with PM2.5 using a value of statistical life (VSL) of 5.5$ million. This value is based
on several published meta-analyses examining value of statistical life (VSL) studies from the
wage-risk literature (Mrozek and Taylor, 2002 and Viscusi and Aldy, 2003). See the Final Non-
Road Diesel Rule (EPA, 2004) for a detailed discussion of the derivation of this value.
Monetized values for avoided mortality are subject to several additional adjustments
reflecting factors relevant to health effects valuation including (a) discounting over the lag
period established for the mortality effect, (b) consideration of real-world growth in per-capita
income over the period leading up to the regulatory analysis year (i.e., 2020) and (c) income
elasticity for the mortality endpoint (i.e., the degree to which WTP to reduce mortality will
match real-world income growth). Following EPA and OMB guidelines for preparing economic
analyses (EPA 2000b, OMB, 2003), discounting over the mortality lag period employed both 3%
and 7% discount rates. Modeling of real-world growth in per-capita income is based on a
projection of GDP growth (obtained from Kleckner and Neuman, 1999 and Standard and Poor's,
2000) as well as population growth for the adult cohort. With regard to income elasticity (for the
mortality endpoint), research has shown that elasticity is related to the severity of the health
endpoint. This has lead to the development of elasticities for different categories of health effect
including minor, severe/chronic and mortality (Kleckner and Neuman, 1999). For this
illustrative analysis, an income elasticity value of 1.2008 (in 2020) for mortality has been
selected and applied in adjusting monetized estimates.
12.6 Presentation of Results
This illustrative analysis generated mortality incidence reduction estimates for the adult
cohort for each regulatory option and provided two monetized benefits estimates for each
regulatory option (a 3% discount-based estimate and a 7% discount-based estimate). The
discount values are applied in valuing mortality results that are distributed across the lag period
with the higher discount rate of 7% resulting in a lower overall monetized value for aggregated
mortality incidence reductions compared with the 3% rate. Overall results for both options are
presented in Table 12-1. As presented in Table 12-1, benefits for Option 1 range from $1.5
million to $44 million depending on the availability of advanced sorbents technology. Potential
12-7
-------
benefits for Option 2 range from $1.5 million to $130 million, again depending on the status of
advanced sorbent technology.
Table 12-1. PM2.5 Co-Benefits Associated with CAMR Regulatory Options 1 and 2 in
2020
Regulatory Option
Option 1
advanced sorbents not available
advanced sorbents available
Option 2
advanced sorbents not available
advanced sorbents available
Annual Mortality
Incidence
Reduction
(adults)
7
<1
21
<1
Benefits in 2020 (Millions of 1999
dollars)
3% discount rate
44
1.5
130
1.5
7% discount rate
40
1.4
117
1.4
12.7 Discussion of Uncertainties
Characterization of health-related benefits associated with PM reductions is a complex
process which is subject to a variety of potential sources of uncertainty. Key assumptions
underlying the estimate of avoided premature mortality include the following:
• Inhalation of fine particles is causally associated with premature death at concentrations
near those experienced by most Americans on a daily basis. Although biological
mechanisms for this effect have not yet been established, the weight of the available
epidemiological and experimental evidence supports an assumption of causality.
• All fine particles, regardless of their chemical composition, are equally potent in causing
premature mortality. This is an important assumption, because PM produced via
transported precursors emitted from EGUs may differ significantly from direct PM
released from automotive engines and other industrial sources. However, no clear
scientific grounds exist for supporting differential effects estimates by particle type.
• The C-R function for fine particles is approximately linear within the range of ambient
concentrations under consideration. Thus, the estimates include health benefits from
reducing fine particles in areas with varied concentrations of PM including both regions
that are in attainment with the fine particle standards and those that do not meet the
standard.
• The forecasts for future emissions and associated air quality modeling are valid.
Although recognizing the difficulties, assumptions, and inherent uncertainties in the
overall enterprise, these analyses are based on peer-reviewed scientific literature and up-
to-date assessment tools, and we believe the results are highly useful in assessing this
rule.
12-8
-------
Overall uncertainty in mortality-related benefits can be increased when a simplified
methodology such as the illustrative analysis described here is employed due to potential
limitations in capturing important relationships and inter-dependencies between key factors (e.g.,
use of coarse geographic scale which may miss important spatial gradients in emissions, air
quality impacts and the location and density of exposed populations). Key sources of potential
uncertainty associated with this illustrative analysis are briefly described below (Note: this
discussion begins with a list of categories of potential benefits not considered quantitatively in
this illustrative analysis):
• Predicting power plant (sector) behavior in relation to mercury controls: As shown in
table 12-1, the benefits are highly dependent on the assumptions about what mercury-
specific technologies will be chosen. For example, in the control option chosen in the
final rule (option 1), the benefits range from $1.5 million (assume ACI works without an
additional baghouse) to $42 million (assume ACI requires a baghouse). Note that costs
and benefits may co-vary; if H-PAC is used then costs and benefits are lower and if ACI
requires a baghouse then costs and benefits are higher. PM2.5 cobenefits presented in this
section are ultimately dependent on decisions made (collectively) by industry in
controlling mercury emissions under different cap-and-trade scenarios.
• Unqualified benefits: the illustrative approach used here, in focusing on the mortality
endpoint, is likely to capture the majority of health-related benefits and the majority of
total benefits (i.e., health plus welfare). However, important categories of potential PM-
related benefits have been excluded to simplify modeling including: morbidity (e.g.,
hospital admissions for respiratory and cardiovascular endpoints, chronic bronchitis
episodes, asthma exacerbations, non-fatal myocardial infarctions) and visibility benefits.
Had these additional benefits been evaluated, overall benefits would be higher.
• SR-MATRIX model: In modeling population-level exposure reductions, the SR-MATRIX
results in the form of population-weighted PM2.5-related exposure reductions generated
for the Midwest/Northeast (ME) and Southeast (SE) regions were selected and averaged
together, since these regions are where power plant-based emissions reductions are
primarily expected to occur. A more comprehensive analysis of power-plant related
emissions reductions (and associated exposure reductions) conducted using BenMAP,
would have used a more spatially-refined geographic grid that tracked both ambient
PM2.5 reductions and associated exposure reductions with greater precision and
specificity. However, the more generalized approach used here is considered reasonable
for a illustrative analysis of potential benefits. While the SR-MATRIX model has been
shown to produce reasonably accurate predictions of direct PM emissions-related
changes in ambient PM2.5 concentrations, the decision to average together population-
weighted exposure changes (for ambient PM2.5) generated at the highly-aggregated ME
and SE levels to produce a single value for use in this analysis does introduce
considerable uncertainty. It is not known whether this simplifying approach biases the
results in a more or less conservative direction.
• Long-term cohort-based mortality function: Use of the Pope et al., 2002-derived
mortality function to support this analysis is associated with uncertainty resulting from:
12-9
-------
(a) potential of the study to incompletely capture short-term exposure-related mortality
effects, (b) potential mis-match between study and analysis populations which introduces
various forms of bias into the results, and (c) failure to identify all key confounders and
effects modifiers, which could result in incorrect effects estimates relating morality to
PM2.5 exposure. EPA is researching methods to characterize all elements of uncertainty
in the dose-response function for mortality. As is discussed in detail in the CAIR RIA
(EPA, 2005), EPA has used two methods to quantify uncertainties in the mortality
function, including: the statistical uncertainty derived from the standard errors reported in
the Pope et al., 2002 study, and the use of results of a pilot expert elicitation conducted in
2004 to investigate other uncertainties in the mortality estimate. Because this analysis is
a illustrative analysis, we do not quantify uncertainty with these two methods in this
report. In the CAIR benefit analysis, the statistical uncertainty from the standard error of
the Pope et al, 2002 study was twice the mean benefit estimate at the 95th percentile and
one-fourth of the mean at the 5th percentile, while the expert elicitation provided mean
estimates that ranged in value from less than one-third of the mean estimate from the
Pope et al, 2002 study-based estimate to two and one-half times the Pope et al., 2002-
based estimate. The confidence intervals from the pilot elicitation applied to the CAIR
benefit analysis ranged in value from zero at the 5th percentile to a value at the 95th
percentile that is seven times higher than the Pope et al., 2002-based estimate. These
results are highly dependent on the air quality scenarios applied to the concentration-
response functions of the Pope et all, 2002 study and the pilot expert elicitation. Thus,
the characterization of uncertainty discussed in the CAIR RIA could differ greatly from
what would be observed for CAMR due to differences in population-weighted changes in
concentrations of PM2.5 (i.e., the location of populations exposure relative to the changes
in air quality). EPA is continuing its research of methods to characterize uncertainty in
total benefits estimates, and is conducting a full-scale expert elicitation. The full-scale
expert elicitation is scheduled to be completed by the end of 2005.
• Lag period for mortality reductions: Failure to accurately capture the true lag period in
mortality following reductions in ambient PM2.5 concentrations. Although the
distributed lag approach described in Section 12.4 is reasonable given our current
understanding of disease endpoint behavior including lung cancer, cardiovascular disease
and respiratory disease (all of which are associated with PM2.5 exposure), the actual lag
period for specific mortality causes linked to PM2.5 exposure could differ from that used
in this analysis. In the CAIR RIA, we conducted sensitivity analyses to investigate how
different lag structures could influence the total benefits, and concluded that substitution
of the most plausible alternative lag structures had little overall impact on the estimate of
total benefits (reductions are on the order of 5 to 15 percent).
• WTP-based valuation for mortality: The WTP-based value used in this analysis to
monetize mortality incidence reductions is enhanced by being based on a variety of
wage-risk studies evaluated using meta-analysis techniques. However, this WTP-based
value may misrepresent the actual societal value for reductions in PM-related mortality if
societal perception of mortality risk related to PM exposure differs significantly from that
associated with job-related hazards.
12.8 References
12-10
-------
E.H. Pechan & Associates, Inc. (1997b), "Integrated Ozone Particulate Matter and Regional
Haze Cost Analysis: Methodology and Results." Prepared for Innovative Strategies and
Economics Group, Office of Air Quality Planning and Standards, U.S. EPA. Research
Triangle Park, NC. July 1997.
EPA-SAB-COUNCIL-ADV-99-005. February 1999. "An SAB Advisory on the Health and
Ecological Effects Initial Studies of the Section 812 Prospective Study: Report to
Congress: Advisory by the Health and Ecological Effects Subcommittee."
Kunzli N., S. Medina, R. Kaiser, P. Quenel, F. Horak Jr, and M. Studnicka. 2001. "Assessment
of Deaths Attributable to Air Pollution: Should We Use Risk Estimates Based on Time
Series or on Cohort Studies?" American Journal of Epidemiology 153(11):1050-55.
Levi J.I., A.M. Wilson, J.S. Evans, J.D. Spengler , 2003, "Estimation of Primary and Secondary
Particulate Matter Intake Fractions for Power Plants in Georgia", Environmental Science
and Technology, Vol. 37, pp. 5528-5536.
Mrozek J.R., and L.O. Taylor. 2002. "What Determines the Value of Life? A Meta-Analysis."
Journal of Policy Analysis and Management 21(2):253-270.
National Research Council (NRC). 2002. Estimating the Public Health Benefits of Proposed
Air Pollution Regulations. Washington, DC: The National Academies Press.
Pope, C.A., III, R.T. Burnett, M.J. Thun, E.E. Calle, D. Krewski, K. Ito, and G.D. Thurston.
2002. "Lung Cancer, Cardiopulmonary Mortality, and Long-term Exposure to Fine
Particulate Air Pollution." Journal of the American Medical Association 287:1132-1141.
U.S. Environmental Protection Agency, 2005, "Regulatory Impact Analysis for the Final Clean
Air Interstate Rule".
U.S. Environmental Protection Agency, February 2005, Control of Mercury Emissions from
Coal Fired Electric Utility Boilers: An Update. Air Pollution Prevention and Control
Division, National Risk Management Research Laboratory, Office of Research and
Development, US EPA, Research Triangle Park, NC.
U.S. Environmental Protection Agency. September 2000b. Guidelines for Preparing Economic
Analyses. EPA 240-R-00-003.
U.S. Office of Management and Budget (OMB). 2003. Circulate A-4 Guidance to Federal
Agencies on Preparation of Regulatory Analysis.
Viscusi, V.K., and I.E. Aldy. 2003. "The Value of a Statistical Life: A Critical Review of
Market Estimates Throughout the World." Journal of Risk and Uncertainty 27(l):5-76.
12-11
-------
APPENDIX A-1 MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR
EAGLE BUTTE (SOUTH DAKOTA) Al-1
Al.l Introduction Al-1
A 1.2 Site Characteristics and Model Parameterization Al-4
Al .3 SERAFM Simulations of Eagle Butte, Lee Dam Al-5
A1.4 WASP7 Simulations of Eagle Butte, Lee Dam Al-6
A1.5 BASS Model Simulations of Methylmercury and Fish Dynamics in Lee Dam,
Eagle Butte, SD—Response to Changes in Mercury Loading Al-13
APPENDIX A-2 MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR
PAWTUCKAWAY LAKE (NEW HAMPSHIRE) A2-1
A2.1 Introduction A2-1
A2.2 SERAFM Application A2-2
A2.3 WASP Model Calibration A2-3
A2.4 References A2-10
APPENDIX A-3 MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR
LAKE WACCAMAW (NORTH CAROLINA) A3-1
A3.1 Introduction A3-1
A3.2 Empirical Data from Lake Waccamaw A3-2
A3.3 SERAFM Application: Lake Waccamaw A3-4
A3.4 Lake Waccamaw WASP Model Calibration A3-6
A3.5 References A3-13
APPENDIX A-4 MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR THE
BRIER CREEK WATERSHED (LOCATED IN THE SAVANNAH RIVER BASIN,
GEORGIA) A4-1
A4.1 Background A4-1
A4.2 Mercury Deposition Network A4-3
A4.3 Watershed Hydrologic and Sediment Loading Model A4-4
A4.4 Water Quality Fate and Transport Model A4-5
A4.5 Model Results A4-6
A4.5.1 Water Quality Model A4-6
A4.6 Brier Creek Watershed Results A4-8
A4.6.1 Brier Creek Soil Mercury Calibration A4-9
A4.6.2 Mercury Loading Fluxes A4-10
A4.6.3 Future Projections A4-11
A4.6.4 Sensitivity of Temporal Response A4-13
A4.7 Brier Creek Water Body Results A4-16
A4.7.1 Phase 1: Long Term Buildup A4-16
A4.7.2 Phase 2: Response to 2002 Flows A4-17
A4.7.3 Phase 3: Future Attenuation A4-20
A4.7.4 Sensitivity of Time Response A4-22
APPENDIX A-5 MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR
LAKE BARCO (FLORIDA) A5-1
A5.1 Introduction A5-1
-------
A5.2 Empirical Data from Lake Barco A5-1
A5.3 SERAFM Application: Lake Barco A5-1
A5.4 References A5-3
Tables
Table Al-1. Observed Mercury Concentrations in Northern Pike from Lee Dam (DMA-80
results) Al-4
Table Al-2. Statistical summary of northern pike length normalized BAF (4 years) for Eagle
Butte (used in SERAFM) Al-5
Table Al-3. SERAFM Parameter Values for Eagle Butte Al-5
Table Al-4. Calibrated SERAFM Rate Constants for Eagle Butte Al-6
Table Al-5. SERAFM 50% Load Reduction Scenario for Eagle Butte Al-6
Table Al-6. SERAFM Zero-Out Scenario for Eagle Butte (Removal of Deposition attributed to
coal-fired utilities) in the CMAQ and REMSAD Models Al-6
Table A1-7. WASP Forecasted Mercury Concentrations in Eagle Butte Sediments in Response
to 50% Loading Reduction Scenario Al-12
Table A2-1. Summary of Yellow Perch Mercury Data from Pawtuckaway Lake A2-1
Table A2-2. Pawtuckaway Lake Parameter Values A2-2
Table A2-3. A Comparison of Measured and Baseline Steady State Values for Pawtuckaway
Lake A2-2
Table A2-4. Lake Pawtuckaway SERAFM Calibrated Rate Constants A2-3
Table A2-5. Time to Reach 90% Steady State After 50% Reduction in Atmospheric
Deposition A2-3
Table A2-6. SERAFM Model Forecasts with Zero-Out Scenario for Coal-Fired Power Plants
(Medium Response Time Scenario) A2-3
Table A2-7. Mercury Response Times for Lake Pawtuckaway, in years A2-8
Table A3-1. Observational Data from Lake Waccamaw A3-2
Table A3-2. Raw Fish Tissue Data Collected from Lake Waccamaw A3-3
Table A3-3. Annual Wet Deposition of Mercury at Waccamaw 1998-2000 A3-4
Table A3-4. Model Parameter Values A3-5
Table A3-5. Measured and Baseline Steady State Values for Lake Waccamaw A3-5
Table A3-6. SERAFM Calibrated Rate Constants for Lake Waccamaw A3-6
Table A3-7. SERAFM 50% Load Reduction Scenario for Lake Waccamaw A3-6
Table A3-8. SERAFM Zero-Out Scenario for Lake Waccamaw (Removal of Deposition
Attributed to Coal-fired Utilities) in the CMAQ and REMSAD Models A3-6
Table A3-9. WASP Response Time Estimates for Lake Waccamaw A3-11
Table A4-1. Average Mercury Deposition Hg Concentrations and Depositions Rates .... A4-4
Table A4-2. Specified and Calculated Reaction Rates and Coefficients A4-6
Table A4-3. Flows, Depths, Length and Volumes used in WASP Model A4-7
Table A4-4. Measured vs. Predicted for Sediment Components A4-7
Table A4-5. Predicted and Observed Mercury Concentrations under Annual Average Load and
Flow A4-8
Table A4-6. Soil Mercury Data in Local Region A4-8
Table A4-7. June 2003 Survey vs WASP Predictions for Mercury A4-18
Table A5-1. Lake Barco Parameter Values A5-1
Table A5-2. Measured and Baseline Steady State Values for Lake Barco A5-2
Table A5-3. Lake Barco SERAFM Calibrated Rate Constants AS-2
Table A5-4. Time to Reach 90% Steady State After 50% Reduction in Atmospheric
-------
Deposition A5-2
Table A5-5. SERAFM Model Forecasts with Zero-Out Scenario for Coal-Fired Power
Plants A5-3
Figures
Figure Al-1. Location of Lee Dam (lower left quadrant) on La Plant SW quadrangle .... Al-3
Figure Al-2. WASP Water Column Solids Calibration Al-7
Figure Al-3. WASP Upper Sediment Solids Calibration Al-8
Figure Al-4. WASP Burial Rate Calibration Al-8
Figure Al-5. WASP Total Mercury Buildup in Water Al-9
Figure Al-6. WASP Methyl Mercury Buildup in Water Al-9
Figure Al-7. WASP Total Mercury Buildup in Sediment Al-10
Figure Al-8. WASP Methyl Mercury Buildup in Sediment Al-10
Figure Al-9. WASP Total Mercury Attenuation in Water Al-11
Figure Al-10. WASP Total Mercury Attenuation in Surface Sediment Al-11
Figure Al-11. WASP Attenuation Sensitivity in Water Al-12
Figure Al-12. WASP Attenuation Sensitivity in Water Al-13
Figure Al-13. Base Case Response of Northern Pike to Methylmercury Exposure Al-14
Figure Al-14. Base case response of yellow perch to methylmercury exposure (0.5ng/L) Al-14
Figure Al-15. Attenuation of Methylmercury in Northern Pike after Load Reduction ... Al-15
Figure Al-16. Attenuation of Methylmercury in Yellow Perch after Load Reduction ... Al-15
Figure A2-1. WASP Water Column Solids Calculation A2-4
Figure A2-2. WASP Solids Simulation for Surface Sediment A2-5
Figure A2-3. WASP Simulation of Burial Velocity A2-5
Figure A2-4. WASP Total Mercury Buildup in Water A2-6
Figure A2-5. WASP Methyl Mercury Buildup in Water A2-7
Figure A2-6. WASP Total Mercury Buildup in Sediment A2-7
Figure A2-7. WASP Methyl Mercury Buildup in Sediment A2-8
Figure A2-8. WASP Total Mercury Attenuation in Epilimnion A2-9
Figure A2-9. WASP Total Mercury Attenuation in Hypolimnion A2-9
Figure A2-10. WASP Total Mercury Attenuation in Surface Sediment A2-10
Figure A3-1. Southeastern North Carolina and Lake Waccamaw A3-1
Figure A3-2. WASP Water Column Solids Calculation A3-7
Figure A3-3. WASP Solids Simulation for Surface Sediment A3-8
Figure A3-4. WASP Simulation of Burial Velocity A3-8
Figure A3-5. WASP Total Mercury Buildup in Water A3-9
Figure A3-6. WASP Methyl Mercury Buildup in Water A3-10
Figure A3-7. WASP Total Mercury Buildup in Sediment A3-10
Figure A3-8. WASP Methyl Mercury Buildup in Sediment A3-11
Figure A3-9. WASP Total Mercury Attenuation in Epilimnion A3-12
Figure A3-10. WASP Total Mercury Attenuation in Surface Sediment A3-12
Figure A4-1. Brier Creek Watershed A4-1
Figure A4-2. Brier Creek Subwatersheds for Hg Loadings A4-2
Figure A4-3. Brier Creek Watershed Landuses A4-3
Figure A4-4. Mercury Deposition Network Sampling Locations A4-4
Figure A4-5. Brier Creek Soil Mercury Buildup A4-9
Figure A4-6. Brier Creek Loading Flux Buildup A4-10
-------
Figure A4-7. Upper Brier Creek Soil Mercury Attenuation A4-11
Figure A4-8. Brier Creek Loading Flux Attenuation A4-12
Figure A4-9. Upper Brier Creek Soil Mercury Attenuation A4-13
Figure A4-10. Upper Brier Creek Loading Flux Attenuation A4-14
Figure A4-11. Watershed Loading Flux Attenuation considering Landuse Change A4-15
Figure A4-12 Base Case Water Column Mercury Concentration for Brier Creek A4-16
Figure A4-13. Base Case Sediment Mercury Concentration for Brier Creek A4-17
Figure A4-14. Brier Creek Total Mercury Water Column Concentration A4-18
Figure A4-15. Brier Creek Methyl Mercury Water Column Concentration A4-19
Figure A4-16. Mercury Attenuation over Time in Water Column A4-20
Figure A4-17. Mercury Attenuation over Time in Sediments A4-21
Figure A4-18. Sensitivity Range for Upstream Waters A4-23
Figure A4-19. Sensitivity Range for Downstream Waters A4-24
-------
APPENDIX A-l
MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR EAGLE BUTTE
(SOUTH DAKOTA)
Al.l Introduction
This appendix contains technical details of input parameters, model calibration and
scenario projection in response to atmospheric mercury loading reduction in Eagle Butte, South
Dakota.
Eagle Butte is located in north/central portion of South Dakota on the Cheyenne River
Sioux Tribal (CRST) Lands. The USGS Hydrologic Unit Code (HUC) for this watershed is:
01010302 (Upper Lake Oahe). The CRST inhabits the Cheyenne River Sioux Reservation,
which is located within the former Dakota Territory, in an area that is known today as central
South Dakota. The Reservation is over 2.8 million acres, comparable in size and shape to the
State of Connecticut. The Reservation encompasses Dewey and Ziebach counties and is
comprised of a portion of the aboriginal territory of the Great Sioux Nation and the United
States. The boundaries of Cheyenne River Reservation are set forth in Section 5 of the Act of
March 2, 1889, 25 Stat. 888 (1889). The Cheyenne River (Wakpa Waste) and the Missouri River
(Mni Sose') form the southern and eastern Reservation boundaries, respectively. Bordering the
Cheyenne River Sioux Reservation on the north is the Standing Rock Sioux Reservation. The
world's largest earthen dam was constructed by the U.S. Army Corps of Engineers (USAGE) on
the Missouri River downstream of the Cheyenne River Sioux Reservation in 1958 and formed
Lake Oahe.
The climate of the Cheyenne River Basin is characterized as semi-arid continental, with
large variations in precipitation and temperature. The high plains of eastern Wyoming are
relatively dry. During the period from 1931 to 1990, the mean annual temperature was about 46
degrees Fahrenheit (°F). Orographic effects induced by the Black Hills are responsible for
dramatic spatial changes in climate within western South Dakota. As elevation increases,
precipitation generally increases and temperature generally decreases. Precipitation also varies
spatially within the Black Hills, from a mean annual value of less than 12 inches in the
southwest, to a maximum of more than 29 inches in the north.
The Cheyenne and Belle Fourche Rivers both originate in east-central Wyoming in areas
of Eocene and Paleocene-age sedimentary rock exposures. The Paleocene-age rocks host thick,
continuous coal seams. The two rivers around the domal Black Hills uplift where mostly
metasedimentary rocks of precambrain age, carbonate rocks of Paleozoic age, and mixed
sandstones and shales of early Mesozoic age are exposed. The two rivers stay within the outcrop
belts of Late Cretaceous-age Pierre Shale, which is a marine shale that contains high
concentrations of iron, manganese, and limestone concretions. The Pierre shale also is
characterized by an abundance of low permeability bentonite clay. As a result, exposures of this
unit are prone to high runoff during periods of intense or extended rainfall
Al-1
-------
The site modeled (Lee Dam), is a shallow, well-mixed system and has a water surface
area of 0.2 km2 and a watershed:lake area ratio of 22.6. There are power plants in the vicinity of
this site, although the prevailing meteorology likely transports most emissions east of the site.
Currently, consumption advisories are in place on reservation lands due to high levels of
mercury in piscivorous fish. Atmospheric deposition data, total and methylmercury
concentrations in sediments, water and biota have all been collected as part of this research. To
mode this system, we used the empirical data collected at this site to parameterize both the
SERAFM and WASP models.
Al-2
-------
Chiyinne River Sioui Tribal
Reservation
D BO 120 240 360 4BC
Figure Al-1. Location of Lee Dam (lower left quadrant) on La Plant SW quadrangle
Top map shows tribal lands in EPA Region 8 (highlighted in yellow)
Al-3
-------
A1.2 Site Characteristics and Model Parameterization
Mercury dynamics at this site are currently being studied as part of an EPA Regional
Office RARE Grant awarded in 2003 (http://www.epa.gov/osp/regions/RARE_Region8.pdf).
Data used to parameterize ecosystem and food web models were obtained from this study
(Unpublished Data, John Johnston, EPA/ORD ERD-Athens).
Table A1-1 presents Northern Pike sample data collected at the Lee Dam site that were
used to characterize empirically calibrated BAFs for this site. Data were normalized to correct
for length/age variability, long transformed and summarized in Table A1-2.
Table Al-1. Observed Mercury Concentrations in Northern Pike from Lee Dam (DMA-80
results)
Sample ID Weight Height [TOTHg] Mass Fish g Length Fish mm
ppb
fpsd!50401
fpsdl 50402
fpsd!50403a
fpsd!50403b
fpsdl 50404
fpsdl 50405
fpsd!50406
fpsd!50407
fpsd!50408
fpsdl 50409
fpsd!50412
fpsd!50410
fpsdl 504 11
0.1986
0.1943
0.1520
0.1363
0.2250
0.1683
0.1885
0.1524
0.2157
0.2378
0.2898
0.2066
0.2616
0.1829
0.2779
0.1695
0.1350
0.2356
0.1427
0.2024
0.1947
0.2175
0.1690
0.0803
0.0843
0.0749
683.96
1094.07
824.79
725.34
790.33
622.06
801.88
951.94
756.57
525.43
199.89
294.42
206.19
1900
1980
2460
2460
1272
1758
1184
1858
708
550
n/a
n/a
n/a
680
854
730
730
590
669
580
662
500
468
n/a
n/a
n/a
BAFs of both benthic macroinvertebrates and zooplankton were calculated directly from
field data and used in the model. The expression used for mercury exposures under atmospheric
loading reduction (dissolved organic mean concentration from field data used was 0.5 ng L"1) is
as follows:
Cwater (ng L-1) = 0.5 exp(-0.001 l*t(day))
This is consistent with the predictions of both SERAFM and WASP, such that after two years
epilimnion concentrations were half of the starting concentration and after five years
concentration reached a 90% reduction.
Annual water temperature was simulated by the following function:
T = 15 + 10*sin[(0.0172*T(days)-0.280)], (T0 = April 1)
This is intended to capture the annual variability in temperature (which affects metabolic
demand), matching minimum and maximum temperatures collected in the field data, as well as
the lowest temperatures reached in January.
Al-4
-------
Table Al-2. Statistical summary of northern pike length normalized BAF (4 years) for
Eagle Butte (used in SERAFM)
Northern Pike
Value
BAF-MeHg water (kg L-1)
Stdev BAF (kg I/1)
N samples
N of age 4 pike for length correction
Normalized HgT (ug g"1)
Stdev normalized Hg (|ig g'1)
8.9x105
2.65xl05
42
10
0.89
0.27
A1.3 SERAFM Simulations of Eagle Butte, Lee Dam
The following tables and parameter values summarize the application of the SERAFM
model to simulate mercury dynamics in Eagle Butte, Lee Dam. The model was first run to
steady state after being calibrated to the observed data and then used for scenario projections of
1) a 50% decline in atmospheric deposition and 2) removal of coal fired utilities from overall
projected deposition using the CMAQ and REMSAD models.
Table Al-3. SERAFM Parameter Values for Eagle Butte
Parameter
Water Column MeHg Unfiltered (ng L/1)
Water Column HgT Unfiltered (ng L'1)
Sediment MeHg (ng g"1 dry)
Sediment HgT (ng g"1 dry)
Age 4 Northern Pike Tissue Hg (ng/g)
Observed BAF: FishHg/MeHg Water
Measured
Range
0.4-2.9
0.5 - 100
0.062-1.74
28.1-95
0.5-2
Mean
1.0
6.9
0.40
44.1
0.89
8.9 xlO5
Predicted
0.82
10.2
0.29
63.9
0.97
Table Al-4. Calibrated SERAFM Rate Constants for Eagle Butte
Eagle Butte: SERAFM Calibrated Rate Constants
Process
Methylation*
Demethylation
Biotic Reduction
Photo-Degradation
Photo-Reduction (Vis)
Photo-Reduction (UV-B)
Photo-Oxidation (UV-B)
Dark Oxidation
Media
Epilimnion
Sediment
Epilimnion
Sediment
Water Column
Water Column
Water Column
Water Column
Water Column
Water Column
Value
0.04
0.001
0.1
0.40
0.03
0.002
0.003
2.825
5.885
1.44
Units
per day
per day
per day
per day
per day
Per day per E/m2-day
Per day per E/m2-day
Per day per E/m2-day
Per day per E/m2-day
per day
"Note: Methylation rate constant in Lee Dam is increased to account for the reservoir effect of repeated flooding and
drying, and the increased zone of redox potential where methylation is found to be increased.
Al-5
-------
Table Al-5. SERAFM 50% Load Reduction Scenario for Eagle Butte
Lee Dam, Eagle Butte 50% Loading Scenario
Slow Medium Fast
Water 224
Sediment 346
Fish 2 3 4
Note: Fast = 1 cm active sediment layer, D (macro-dispersion coefficient) = 1&4 cm2 s'1; Medium
= 2 cm active sediment layer, D = 10'4 cm2 s'1; Slow -3 cm sediment, D = 5x10~5 cm2 s'}.
Table A1-6. SERAFM Zero-Out Scenario for Eagle Butte (Removal of Deposition
attributed to coal-fired utilities) in the CMAQ and REMSAD Models
Eagle Butte - Lee Dam
Epilimnion
Hypolimnion
Sediment
Fish
Slow
2
-
3
2
Med
3
—
4
3
Fast
4
—
6
4
A1.4 WASP7 Simulations of Eagle Butte, Lee Dam
Simulations were set up with the basic parameters from the SERAFM model of Lee Dam.
This water body receives watershed loadings of solids and mercury, but has no significant
outflow. Solids loadings from the watershed are balanced by in-lake mineralization and burial.
Mercury loadings from direct deposition and from the watershed are subject to volatilization and
burial losses.
WASP7 simulations were run for a total of 200 years. The first 100 years represent the
buildup of mercury to steady-state levels using present loadings. Mercury loadings were then
cut 50%, and the attenuation period was tracked for 100 years.
The solids balance is represented in the following figures (Figure 2-Figure 3). The water
column equilibrated at a silt concentration of just over 2 mg L"1. The organic matter (OM)
represents biotic solids (including phytoplankton, periphyton, and macrophytes) and detritus.
Al-6
-------
2500
20.00
c" 15.00
§
§ 1000
0
1
i
5.00
0.00
(
Water Column Solids
) 10 20 30 40 50 60 70 80 90
Time, years
— Epilimnion Silt
Epilimnion OM
Figure Al-2. WASP Water Column Solids Calibration.
Sediment solids are balanced by the composition of the erosion load, the in-lake biotic
production, and the mineralization of OM. The sediment composition is primarily abiotic silt (70%),
with a sand fraction just under 20% and an organic fraction just over 10%.
Upper Sediment Solids Composition
80.0
70.0
60.0
•
£ 50.0
I «•»
I 30.0
* 20.0
10.0
0.0
"D
%sand
%OH
Observed %OM
10 20 30 40 50 60
Time, years
70 80 90
Figure Al-3. WASP Upper Sediment Solids Calibration
Al-7
-------
The burial rate is calculated internally from the solids balance. Given the specified loads,
production, and mineralization rates, burial stabilized at about 0.24 cm/yr.
Burial from Lower Sediment
n »K
f
•S1 n,is I
I
I 0,
1
m
0.05
0
(
) 10 20 30 4O SO 60 70 8O 90
Time, years
Calculated
Q Observed
Figure Al-4. WASP Burial Rate Calibration
Total mercury built up over the first 100 years to 9.5 ng L"1 in the water, and 180 ng g"1 in
the sediment. The sediment levels are higher than observations, indicating an additional loss
mechanism (e.g., faster burial or uptake by macrophytes). Methyl mercury levels built up to
reasonable concentrations in the water and the sediment.
Al-8
-------
Base Case Mercury Buildup
in
HgT Concentration, ng/L
g
R
7
6
4 .
/
/ A
g L-
I
.(
2
1
0 .
0 10 20 30 40 50 60 70 80 90
Time, years
A Observed Epilimnion, ng/L
Figure Al-5. WASP Total Mercury Buildup in Water
Base Case Mercury Buildup
1 2
1 .
c 08
.1
1 0.6-
01
0
§
0 0.4 -
01
i
0.2
0
• .: • ' 1
f
\
'
Epilimnion, ng/L
A Observed Epilimnion, ng/l
0 10 20 30 40 50 60 70 80 90
Time, years
Figure Al-6. WASP Methyl Mercury Buildup in Water
Al-9
-------
200
180
160
™ 140
O)
c
c 120
o
I 100
0)
80
o
H 60
£
40
20
0
Base Case Mercury Buildu
/"
/ ^"""^
/ z
Z "i
/
/
/
) 10 20 30 40 50 60 70 80 90
Time, years
P
Upper Sediment, ng/g
O Observed Sediment, ng/g
Figure Al-7. WASP Total Mercury Buildup in Sediment
0.7
0.6
S 05
MeHg Concentration, ng
0000
O •* M <* «*
Base Case Mercury Buildup
/"-" ~
/ ^_^^~^*-
\ -Z
I z
I /
/ L
) 10 20 30 40 50 60 70 80 90
Time, years
1
1
Hypollmnion, ng/g
Upper Sediment, ng/g
rj Observed Sediment, ng/g
1
Figure Al-8. WASP Methyl Mercury Buildup in Sediment
After external mercury loads were reduced 50%, the mercury levels in the water column
and surface sediment declined rapidly, on the order of years. Mercury levels in the lower
sediment layer declined slowly over the following decades due to burial.
Al-10
-------
Base Case Mercury Attenuation
10
9
8
t7
.1 6
c 4
* 3
2
1
0
I
~|
T
V
\_
) 10 20 30 40 50 60 70 80 90
Time, years
Epllimnion, ng/L
90% Response, Epilimnion
Figure Al-9. WASP Total Mercury Attenuation in Water
200
180
160
tl40
§ 12"
£ 100
•
80
o
o
I- 60
s
40
20
0
Base Case Mercury Attenuatio
~~L
"T^x.
\ ^^
\
) 10 20 30 40 50 60 70 80 90
Time, years
1
Upper Sediment, ng/g
Lower Sediment, ng/g
Sediment
Figure Al-10. WASP Total Mercury Attenuation in Surface Sediment
Three scenarios were simulated representing fast, medium, and slow estimates of
recovery. The upper sediment layer thickness was varied from 1 cm, 2 cm, and 3 cm. For the
slow scenario, the sediment-water dispersion coefficient was reduced by half from 10"" cm2 sec'1.
Response times are summarized in the following table, and presented graphically for water and
sediment.
Al-11
-------
Table Al-7. WASP Forecasted Mercury Concentrations in Eagle Butte Sediments in
Response to 50% Loading Reduction Scenario
Mercury Response Times
Compartment
Epilimnion
Surface Sediment
for Lee Dam,
Fast
5
6
in years
Medium
9
11
Slow
14
16
Lee Dam Water Column Mercury Attenuation
icentration, ng/L
b. Ol O> -J 00 ID <
O
0
a
1
— '
L
) 10 20 30 40 50 60 70 80 90
Time, years
Fast Response
Slow Response
90% Response, Epilimnion
Figure Al-11. WASP Attenuation Sensitivity in Water
Al-12
-------
Lee Dam Sediment Mercury Attenuation
200
180
160
r Concentration, ng/g
•*•*•*
Soo o to ^
O 0 0 0
01
40
20
0
1
I
1
¥
) 10 20 30 40 50 60 70 80 90
Time, years
Fast Response
Medium Response
___90% Response, Epilimnlon
Figure Al-12. WASP Attenuation Sensitivity in Water
A1.5 BASS Model Simulations of Methylmercury and Fish Dynamics in Lee Dam, Eagle
Butte, SD—Response to Changes in Mercury Loading
The following figures illustrate the response time to both loadings and load reductions of
mercury from atmospheric sources. BASS is run in FGETS mode, i.e., without population
dynamics, to simulate whole-body organic mercury chemical residues in weight and length
classes of community members. BASS is calibrated by adjusting species growth rates, dietary
composition and concentrations and bioaccumulation factors of benthic macroinvertebrates and
zooplankton. Field data for length, weight and ages of species are corroborated by the BASS
model. Predicted model concentrations for northern pike, yellow perch, black bass and black
crappie agree with observed field concentrations for these species. Exposure to mercury occurs
through food and gill uptake only. The macroinvertebrate BAF used is 1.14X105, and the
zooplankton BAF used is 1.67"105, both of which are calculated from field data at Lee Dam.
Lee Dam is predominantly a northern pike/yellow perch community, but also includes
subdominant black crappie, black bass and sportail shiner, all of which have been sampled for
residue analysis over a two year period using a combination of gill nets and seine nets. An
abundance of submerged aquatic vegetation provides suitable cover for these species. Northern
pike is the top predator (SERAFM Level 4) in this lake system, feeding primarily on yellow
perch.
Dissolved methylmercury concentrations are set at O.Sng L"1 for the base case, and
equilibrium is reached in 1500 days for length class 3 pike and 3500 days for length class 4 pike.
Al-13
-------
tot tax
Figure Al-13. Base Case Response of Northern Pike to Methylmercury Exposure
Dynamics of the community co-dominants to reductions in atmospheric loading were
simulated by decreasing epilimnion dissolved methylmercury concentrations from O.Sng L"' at
time zero to approximately 0.25ng L"' at year 2 and approximately 0.05ng L"1 at year 5 to be
consistent with predicted water column dynamics from SERAFM and WASP for this lake
system (reaching a 90% reduction on the order of 5 years).
"/
J ' ' I /. L
L.L i L
L L L '. 1 (
Figure Al-14. Base case response of yellow perch to methylmercury exposure (0.5ng/L)
Al-14
-------
The response time of the concentration of mercury in fish is roughly twice the duration of
the simulated water concentration response. The largest northern pike length classes reach the
action limit concentration of 0.3 ppm in approximately 5-10 years. The largest yellow perch
reach the 0.3 ppm level in a span of 5 to 8 years. Note that a roughly equivalent time lag is
predicted for the biotic response in this dynamic and relatively rapidly responding system (see
SERAFM and WASP forecasts).
Figure Al-15. Attenuation of Methylmercury in Northern Pike after Load Reduction
I
.. .
Figure Al-16. Attenuation of Methylmercury in Yellow Perch after Load Reduction
Al-15
-------
APPENDIX A-2
MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR PAWTUCKAWAY
LAKE (NEW HAMPSHIRE)
A2.1 Introduction
Pawtuckaway Lake is a medium sized, seepage lake in Nottiham, New Hamsphire. This
lake a water surface area of 3.6 km2 and a watershedrlake area ratio of 13.7. Lakes in this region
are characteristic of undisturbed lakes within the Northeastern Highlands Ecoregion (Omernik,
1987; US EPA, 2000). This lake was part of a recent study of mercury dynamics across a
number of Vermont and New Hampshire Lakes funded by EPA's Office of Research and
Development under the Regional Environmental Monitoring and Assessment Program (Kamman
et al., 2004). Site specific data used to parameterize the mercury cycling models developed for
Pawtuckaway Lake were from Kamman and Engstrom (Kamman and Engstrom, 2002) and
Kamman (2004) or http://www.vtwaterquality.org/lakes/docs/lp remap-datareport.pdf. An
excerpt from Kamman and Engstrom (2002) describing the region is given below:
"Most lakes in this area occupy undisturbed forested catchments, which are a mix of
deciduous or coniferous vegetation overlying soils ranging from stony to silty loams. Bedrock
geology is largely schistose or granitic, and most watersheds are poorly buffered. Some shales
and slates are in evidence near High Pond in Vermont, and the buffering capacity of this
watershed is enhanced accordingly. These watersheds have experienced varying degrees of
deforestation during settlement, but have regrown to forest in the past 75-150 years."
Fish data were normalized for age/length variability. Details of data transformations can
be found in Kamman (2004). SERAFM was calibrated to forecast fish tissue concentrations in 4
year old yellow perch from this system. These data are summarized in Table 1.
Table A2-1. Summary of Yellow Perch Mercury Data from Pawtuckaway Lake
Pawtuckaway Lake Value
Mean normalized Yellow Perch Hg (ug/g) 0.21
Range Hg Yellow Perch (ug/g) 0.12-0.40
Mean MeHg Water Concentration Epilimnion 0.19
(ng/L)
Range MeHg Epilimnion (ng/L) 0.14-0.24
HgT Water Concentration (ng/L)
Mean 1.65
Log Mean 0.18
Range HgT water (ng/L) 0.7-3.8
A2-1
-------
A2.2 SERAFM Application
Table A2-2. Pawtuckaway Lake Parameter Values
Parameter
Lake Pawtuckaway
Watershed Area
Percent Impervious
Percent Forest
Percent Riparian
Percent Upland
Lake Area
Catchment/Lake Ratio
Epilimnion Depth
Hypolimnion Depth
Hypolimnion Anoxia
Hydraulic Residence Time
Inflow/Outflow
Water pH
Epilimnion DOC
Hypolimnion DOC
Trophic Status
Annual Precipitation
Hgll Cone, in Precip
Wet Deposition (Hgll)
Dry Deposition (Hgll)
Wet Deposition (Hgll)
Dry Deposition (Hgll)
50,007,406 m2
1%
88%
10%
1%
3,6412,200 m2
13.73
2m
3m
Yes
0.45 yrs
4.05x10' m3/yr
6.45
5.46 mg/L
5.55 mg/L
Dystrophic
102 cm/yr
lOng/L
10.2 ug/m2/yr
10.2 ug/m2/yr
0.153 ug/m2/yr
0.153 ug/m2/yr
Table A2-3. A Comparison of Measured and Baseline Steady State Values for
Pawtuckaway Lake
Parameter
EPI MeHg Unfiltered
EPI HgT Unfiltered
HYP MeHg Unfiltered
HYP HgT Unfiltered
Sediment MeHg
Sediment HgT
Perch Tissue Hg
BAF: FishHg/MeHg
Measured
Range
0.14- 0.24 ng/L
0.71- 3.8 ng/L
2.38 - 3.44 ng/L
6.94 -34.54 ng/L
0.12- 0.4 ug/g
l.llxlO6
Mean
0.19 ng/L
2.26 ng/L
2.91 ng/L
20.74 ng/L
7ng/g
290 ng/g
0.21 ug/g
Predicted
0.35 ng/L
3.57 ng/L
0.38 ng/L
5.63 ng/L
6 ng/g
237 ng/g
0.23 ug/g
(0.21-0.26)
A2-2
-------
Table A2-4. Lake Pawtuckaway SERAFM Calibrated Rate Constants
Process
Methylation
Demethylation
Biotic Reduction
Photo-Degradation
Photo-Reduction (Vis)
Photo-Reduction (UV-B)
Photo-Oxidation (UV-B)
Dark Oxidation
Media
Epilimnion
Hypolimnion
Sediment
Epilimnion
Hypolimnion
Sediment
Water Column
Water Column
Water Column
Water Column
Water Column
Water Column
Value
0
0.01
0.01
0.0001
0.001
0.4
0.03
0.002
0.003
2.825
5.885
1.44
Units
per day
per day
per day
per day
per day
per day
per day
per day per E/m2-day
per day per E/m2-day
per day per E/m2-day
per day per E/m2-day
per day
Table A2-5. Time to Reach 90% Steady State After 50% Reduction in Atmospheric
Deposition
Lake Pawtuckaway Fast Medium Slow
Epiliminion 59 115 179
Hypolimnion 79 154 >180
Sediments 80 125 >180
Fish 34 56 64
Fast = 1 cm active sediment layer, D (macro-dispersion coefficient) = 10"4 cm2/s
Medium = 2 cm active sediment layer, D = 10"4 cm2/s
Slow - Slow Responding Sediment: 3 cm sediment, D = 5x10~5 cm2/s
Table A2-6. SERAFM Model Forecasts with Zero-Out Scenario for Coal-Fired Power
Plants (Medium Response Time Scenario)
Lake Pawtuckaway
Epilimnion HgT (ng/L)
Epilimnion MeHg (ng/L)
Hypolimnion (ng/L)
Sediment MeHg (ng/g)
Fish (ng/g)
Time
95
122
125
47
A2.3 WASP Model Calibration
WASP simulations of Lake Pawtuckaway were set up with the basic parameters from the
SERAFM model. In addition to direct deposition of mercury, this water body receives
watershed loadings of solids and mercury. Solids loadings from the watershed and internal
production of organic matter are balanced by outflow, in-lake mineralization, and burial.
Mercury loadings from direct deposition and from the watershed are subject to outflow,
volatilization and burial losses. WASP simulations were run for a total of 200 years. The first
100 years represent the buildup of mercury to steady-state levels using present loadings.
Mercury loadings were then cut 50%, and the attenuation period was tracked for 100 years.
A2-3
-------
WASP calibration was conducted for a base case scenario with an active sediment depth
of 2 cm and a sediment macro-dispersion coefficient (E) = 10"4 cmVsec. This calibration
employed the same deposition and resuspension velocities for silts and organic matter in the
SERAFM model. Solids were calibrated to match the observed porosity of 0.93, OM fraction of
35%, and burial rate of 0.05 cm/yr
The solids balance is represented in Figure 1 and Figure 4. The epilimnion solids are
mostly organic, including runoff and plankton. Hypolimnion organic matter includes living
periphyton and macrophytes, as well as detritus. Sediment solids are balanced by the
composition of the erosion load, the in-lake biotic production, and the mineralization of OM.
The sediment composition is a balance between silt and organic matter (35% each), and sand
(30%). The burial rate is calculated internally from the solids balance. Given the specified loads,
production and mineralization rates, burial stabilized at 0.04 cm/yr, close to observations.
Water Column Solids
3500
30.00
=J 25.00
O)
° 2000
i
g 1500
0
O
in
— 10.00
o
-------
450
400
35.0
g. 300
£
225.0
"> 200
I 150
100
5.0 -
00
Upper Sediment Solids Composition
^
-
) 10 20 30 40 50 60 70 80 90
Time, years
%silt
%OM
G Observed %OM
Figure A2-2. WASP Solids Simulation for Surface Sediment
0.2
0.18
0.16
O.U
•S" 0.12
£ 0.1
0
|
> 0.08
!
m 0.06
0.04
0.02
0
(
Burial from Lower Sediment
[
1
r
10 20 30 40 50 60 70 80 90
Time, years
Calculated
Q Observed
Figure A2-3. WASP Simulation of Burial Velocity
WASP was calibrated to better match total mercury levels by adjusting the fractions of
sand and silt in the watershed erosion load resulting in a silt/sand/OM erosion composition of
50/45/5 percent. The hypolimnion methylation and demethylation rate constants were adjusted
to try to obtain reasonable a MeHg concentration using reasonable coefficients. The final
A2-5
-------
hypolimnion rate constants were the same as in the active sediment layer, reasoning that the
hypolimnion of this lake has a lot of plant substrate and organic matter.
Total mercury built up over the first 100 years to 8 ng/L in the epilimnion, 15 ng/L in the
hypolimion, and 300 ng/g in the upper sediment. Methyl mercury levels built up to 0.5 - 1 ng/L
in the water, and 6 ng/g in the sediment. Although HgT is overpredicted in the epilimnion and
MeHg is underpredicted in the hypolimnion, sediment levels match the data very well.
Note that while the model calculates high OM in the hypolimnion, much of it
conceptually is macrophyte biomass. The epilimnion HgT concentration is high, while the
hypolimnion HgT is reasonably good. The calculated sediment HgT for the upper 2 cm and the
lower 10 cm bracket the observed value.
Base Case Mercury Buildup
25
20
|>
E 1S
HgT Concentratio
O Ol O
[i5
/""_ _
/
/^--^
r
0 10 20 30 40 50 60 70 80 90
Time, years
^^_Epilimnion, ng/L
^^—Hypolimnion, ng/L
A Observed Epilimnion, ng/l
Q Observed Hypolimnion,
ng/L
Figure A2-4. WASP Total Mercury Buildup in Water
A2-6
-------
3.5
3
d 25
t
= 2
8 1.5
o
0
2 1
£
0.5
0
(
Base Case Mercury Buildup
1
^— — -* ""
^ *"^™
~ ^
) 10 20 30 40 50 60 70 80 90
Time, years
J
k
Epllimnlon. ng/L
Hypnlimiiinn. ng/1
^ Observed Epilimnion, ng/L
Q Observed Hypolimnion,
ng/L
Figure A2-5. WASP Methyl Mercury Buildup in Water
350
Base Case Mercury Buildup
Upper Sediment, ng/g
Lower Sediment, ng/g
Observed Sediment, ng/g
0 10 20 30 40 50 60 70 80 90
Time, years
Figure A2-6. WASP Total Mercury Buildup in Sediment
A2-7
-------
Base Case Mercury Bulldu
8
7
0) 6
= 5 .
°
1 1
1
o 3
o -1 '
01
X
o 2 .
S
1 .
o .
|_
/"•""^
/
/
/
' — — - -"~~~~^
P
]
Upper Sediment, ng/g
D Observed Sediment, ng/g
0 10 20 30 40 SO 60 70 80 90
Time, years
Figure A2-7. WASP Methyl Mercury Buildup in Sediment
After external mercury loads were reduced 50%, the mercury levels in the water column
and surface sediment declined at moderate rates. Mercury attenuation seems to be controlled by
the slow burial and by resuspension of silt and organic matter. Three scenarios were simulated
representing fast, medium, and slow estimates of recovery. The upper sediment layer thickness
was varied from 1 cm, 2 cm, and 3 cm. For the slow scenario, the sediment-water dispersion
coefficient was reduced by half from 10~4 cm2/sec. Response times are summarized in Table 7,
and presented graphically for water and sediment in Figures A2-8 and A2-10.
Table A2-7. Mercury Response Times for Lake Pawtuckaway, in years
Compartment
Epilimnion
Hypolimnion
Surface Sediment
Fast
20
23
24
Medium
36
43
44
Slow
55
66
69
WASP is 2-3 times faster in recovery than SERAFM, probably due to the treatment of
sediment solids balance. In particular, WASP resuspends silt and OM, which have high Kd's
rather than bulk sediment (influenced by sand), which has a lower Kd.
A2-8
-------
Lake Pawtuckaway Epilimnion Mercury Attenuation
.Fast Response
.Medium Response
.Slow Response
.90% Response, Epilimnion
0 10 20 30 40 50 60 70 80 90
Time, years
Figure A2-8. WASP Total Mercury Attenuation in Epilimnion
Lake Pawtuckaway Hypolimnion Mercury Attenuation
-Fast Response
.Medium Response
.Slow Response
.90% Response,
Hypolimnion
0 10 20 30 40 50 60 70 80 90
Time, years
Figure A2-9. WASP Total Mercury Attenuation in Hypolimnion
A2-9
-------
Lake Pawtuckaway Upper Sediment Mercury Attenuation
-Fast Response
-Medium Response
.Slow Response
.90% Response, Upper
Sediment
0 10 20 30 40 50 60 70 80 90
Time, years
Figure A2-10. WASP Total Mercury Attenuation in Surface Sediment
A2.4 References
Kamman, N., Driscoll, C.T., Estabrook, B., Evers, D.C. and Miller, E.K., 2004. Biogeochemistry
of Mercury in Vermont and New Hampshire Lakes An Assessment of Mercury in Water,
Sediment and Biota of Vermont and New Hampshire Lakes Comprehensive Final Project
Report May, 2004, Project Funding Provided by United States Environmental Protection
Agency Office of Research and Development under the Regional Environmental
Monitoring and Assessment Program.
Kamman, N.C. and Engstrom, D.R., 2002. Historical and present fluxes of mercury to Vermont
and New Hampshire lakes inferred from Pb-210 dated sediment cores. Atmospheric
Environment, 36: 1599-1609.
Omernik, J.M., 1987. Ecoregions of the conterminous United States. Map (scale 1:7,500,000).
Annals of the Association ofAmerican Geographers, 77: 119-125.
US EPA, 2000. Level III ecoregions ofthe continental United States (revision of Omernik, 1987).
US Environmental Protection Agency National Health and Environmental Effects
Research Laboratory, Western Ecology Division, Corvallis, OR.
A2-10
-------
APPENDIX A-3
MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR LAKE
WACCAMAW (NORTH CAROLINA)
A3.1 Introduction
Lake Waccamaw is a Large Bay lake in southeastern North Carolina. Lake Waccamaw
has a water surface area of almost 35 km2 and a catchment to lake area ratio of slightly more than
six. In 1992, a survey offish mercury concentrations North Carolina's Department of
Environment, Health and Natural Resources in this region revealed that fish mercury
concentrations exceeded 1 ppm in over 60% of the samples.
Waccamaw is a popular destination for recreational fishing and a fish consumption
advisory is currently in place. The area surrounding Lake Waccamaw is typical of the region:
flat terrain with ubiquitous wetlands and waterways. Very little commercial or industrial activity
takes place in the area immediately surrounding the park, population density is relatively low
and roadways are lightly traveled.
The nearest town is Whiteville, NC, located approximately 15 kilometers to the west-
northwest of Lake Waccamaw. A variety of mercury sources are located in this region including
at least two coal-fired electric utility boilers, a large municipal waste incinerator, several large
coal or oil-fired industrial boilers, and a pulp and paper mill. By far the largest source of historic
mercury emissions was the HoltraChem mercury cell chlor-alkali operation located in
Riegelwood, NC, approximately 25 kilometers east-northeast of Lake Waccamaw.
.
Figure A3-1. Southeastern North Carolina and Lake Waccamaw
A3-1
-------
A3.2 Empirical Data from Lake Waccamaw
Data in Tables 1 and 2 were provided by Debra A. Owen, Environmental Biologist,
North Carolina Department of the Environment and Natural Resources, Department of Water
Quality and Riggs et al., (Riggs et al., 2000). Fish data were normalized to two year old large
mouth bass assuming a length of 10 inches
(http://animaldiversity.ummz.umich.edu/site/accounts/information/Micropterus_salmoides.html)
Table A3-1. Observational Data from Lake Waccamaw
Quarter
1-fall
2-winter
3-spring
4-summer
1-fall
2-winter
3-spring
4-summer
1-fall
2-winter
3-spring
4-summer
Date
11/20/02
01/22/03
05/13/03
07/24/03
11/20/02
01/22/03
05/13/03
07/24/03
1 1/20/02
01/22/03
05/13/03
07/24/03
Hg,
ng/L
4.380
2.740
18.400
10.100
5.990
2.940
18.100
9.32
1.060
3.890
2.680
1.990
MeHg,
ng/L
0.317
0.150
4.040
1.480
0.568
0.152
4.990
1.42
0.132
0.230
0.233
0.132
Diss Hg,
ng/L
7.350
3.78
1.210
Diss
MeHg,
ng/L
1.680
1.8
0.120
DOC,
mg/L
31.90
21.10
49.80
36.30
28.20
19.50
50.40
29.60
8.63
9.96
14.00
12.00
Sulfate,
mg/L
13.40
32.60
1.37
1.41
9.90
28.80
1.99
1.73
5.40
8.40
4.60
4.11
A3-2
-------
Table A3-2. Raw Fish Tissue Data Collected from Lake Waccamaw
DATE
SAMPLED
10/7/2003
. LENGOT
\(cml '
27.6
^I^EIGHTJ^'
s '•',!, &\;.(g) . ''-
286
Hg fclgg'1)
0.22
v- -'..- ^ ,,v* , ^:;
,. Common name
BROWN
BULLHEAD
-*,'; • .,4 • ,-:! ^j
Scientific Name ;,*
ICTALURUS
NEBULOSUS
% ! Trophic
~ Status
Omnivore
' -: ..• '' •"•¥>• * • '(mi? ^ BROWNJBULLHEAB Average' ;•<*• • :• " ' '!
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
•3
28.3
28.3
29.6
29.8
30.4
31.9
32.2
32.3
36.5
39.5
27
29.5
40.5
40.5
43.5
52
:'*l,34:4875
23.2
25.5
28.5
24.3
25.4
.*,-«,. *»>i
285
310
333
349
388
550
421
399
619
1055
295.3
428
1029
1229
1419
2685
?lt,737.1438 "•
272.5
385
421
376
453
•. *?**. "'""'' " '
0.52
0.52
0.68
0.5
0.59
0.75
0.73
0.88
0.95
0.87
0.77
0.84
1
1.1
0.82
2
',', 0.845'
0.11
0.2
0.48
0.31
0.29
'; 0.278-. •--
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
LARGEMOUTH
BASS
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
MICROPTERUS
SALMOIDES
. ""LARGEMOUTH BASSfAveragc ; - "^
REDEAR SUNFISH
REDEAR SUNFISH
REDEAR SUNFISH
REDEAR SUNFISH
REDEAR SUNFISH
LEPOMIS
MICROLOPHUS
LEPOMIS
MICROLOPHUS
LEPOMIS
MICROLOPHUS
LEPOMIS
MICROLOPHUS
LEPOMIS
MICROLOPHUS
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
Piscivore
: '. ' ^S ' T-
Insectivore
Insectivore
Insectivore
Insectivore
Insectivore
'**RED"EA'R'SUNEISH.Averase , .-.;- - ' < • '' - •*"
A3-3
-------
DATE
SAMPLED
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
10/7/2003
LENGTH
(cm)
43.5
38.7
41
43.5
44
45.2
24
18.5
WEIGHT
(e)
1189
752
925
1072
1339
1228
165.3
87.5
•?•• * *
Hg Oige"')
0.12
0.18
0.41
0.4
0.24
0.4
0.291667
0.35
0.81
0.58
• >, >• :
Common name Scientific Name
MINYTREMA
SPOTTED SUCKER MELANOPS
MINYTREMA
SPOTTED SUCKER MELANOPS
MINYTREMA
SPOTTED SUCKER MELANOPS
MINYTREMA
SPOTTED SUCKER MELANOPS
MINYTREMA
SPOTTED SUCKER MELANOPS
MINYTREMA
SPOTTED SUCKER MELANOPS
SPOTTED SUCKER Average
YELLOW PERCH PERCA FLAVESCENS
YELLOW PERCH PERCA FLAVESCENS
YELLOW PERCH Average
Trophic
Status
Insectivore
Insectivore
Insectivore
Insectivore
Insectivore
Insectivore
Piscivore
Piscivore
Atmospheric Deposition Data
The Mercury Deposition Network Monitoring (MDN) Site NCOS corresponds to Lake
Waccamaw. We averaged cumulative wet deposition between 1998 and 2000 to get wet
deposition to Lake Waccamaw (North Carolina Division of Air Quality, 2002) and assumed that
dry deposition is approximately half of total deposition.
Table A3-3. Annual Wet Deposition of Mercury at Waccamaw 1998-2000
Year
1998
1999
2000
Cumulative Wet
Deposition (ug/m2/yr)
15.8
14.8
12.6
Precipitation (cm/yr)
126.6
185.4
133.8
Volume weighted
concentration
(Hg ng/L)
11.6
7.8
9.4
A3.3 SERAFM Application: Lake Waccamaw
A3-4
-------
Table A3-4. Model Parameter Values
Parameter
Watershed Area
Percent Impervious
Percent Forest
Percent Riparian
Percent Upland
Lake Area
Catchment/Lake Ratio
Epilimnion Depth
Hydraulic Residence Time
Inflow/Outflow
Parameter
Water pH
Epilimnion DOC
Hypolimnion DOC
Trophic Status
Annual Precipitation
Hgll Cone, in Precip
Wet Deposition (Hgll)
Dry Deposition (Hgll)
Wet Deposition (Hgll)
Dry Deposition (Hgll)
Lake Waccamaw
216,789,552m2
1%
72%
0%
27%
34,706,448 m2
6.25
2.3m
0.66 yrs
1.20xl08m3/yr
Lake Waccamaw
4.3
25.95 mg/L
n/a
Mesotrophic
120.4 cm/yr
12.0 ng/L
14.4 ug/m2/yr
14.4 ug/m2/yr
0.22 ug/m2/yr
0.22 ug/m2/yr
Table A3-5. Measured and Baseline Steady State Values for Lake Waccamaw
Parameter
Water Column MeHg
Unfiltered
Water Column HgT
Unfiltered
Sediment MeHg
Sediment HgT
Largemouth Bass Tissue
Hg
Observed BAF:
FishHg/MeHg
Measured
Range
0.132 -4.99 ng/L
1.06 -18.4 ng/L
0.033- 0.2 ng/g
1.66- 36.4 ng/g
0.5 - 2 ug/g
Mean
0.483 ng/L
4.79 ng/L
0.13 ng/g
22.8 ng/g
2 yr old Igmouth bass:
0.60 ug/g
Predicted
0.17 ng/L
1.95 ng/L
0.2 ng/g
1.5 ng/g
0.21 ug/g
1.24xl06
A3-5
-------
Table A3-6. SERAFM Calibrated Rate Constants for Lake Waccamaw
Process
Methylation
Demethylation
Biotic Reduction
Photo-Degradation
Photo-Reduction (Vis)
Photo-Reduction (UV-B)
Photo-Oxidation (UV-B)
Dark Oxidation
Media
Epilimnion
Sediment
Epilimnion
Sediment
Water Column
Water Column
Water Column
Water Column
Water Column
Water Column
Value
0
0.07
0.0001
0.40
0.03
0.002
0.03
2.825
5.885
1.44
Units
per day
per day
per day
per day
per day
per day per E/m2-day
per day per E/m2-day
per day per E/m2-day
per day per E/m2-day
per day
Table A3-7. SERAFM 50% Load Reduction Scenario for Lake Waccamaw
Compartment
Lake Waccamaw
Epilimnion
Sediment
Fish
Fast
1
3
1
Medium
2
6
1
Slow
1
12
2
Note: Fast = 1 cm active sediment layer, D (macro-dispersion coefficient) = Idr4 cm2 s'1; Medium = 2 cm active
sediment layer, £> = 70~* en? s~'; Slow -3 cm sediment, D = 5x1 ft5 cm2 s'
Table A3-8. SERAFM Zero-Out Scenario for Lake Waccamaw (Removal of Deposition
Attributed to Coal-fired Utilities) in the CMAQ and REMSAD Models
Compartment
Response Time
(yrs)
Epilimnion HgT (ng/L) 2.8
Epilimnion MeHg (ng/L)
Sediment (ng/g) 5.8
Fish(ug/g) 3.8
A3.4 Lake Waccamaw WASP Model Calibration
WASP simulations of Lake Waccamaw were set up with the basic parameters from the
SERAFM model. In addition to direct deposition of mercury, this water body receives
watershed loadings of solids and mercury. Solids loadings from the watershed and internal
production of organic matter are balanced by outflow, in-lake mineralization, and burial.
Mercury loadings from direct deposition and from the watershed are subject to outflow,
volatilization and burial losses. WASP simulations were run for a total of 200 years. The first
100 years represent the buildup of mercury to steady-state levels using present loadings.
Mercury loadings were then cut 50%, and the attenuation period was tracked for 100 years.
WASP calibration was conducted for a base case scenario with an active sediment depth
of 2 cm and a sediment macro-dispersion coefficient (E) = 10"4 cmVsec. This calibration
employed the same deposition and resuspension velocities for silts and organic matter in the
A3-6
-------
SERAFM model. Solids were calibrated to match the observed porosity of 0.93, OM fraction of
35%, and burial rate of 0.05 cm/yr
The solids balance is represented in Figure A3-2- Figure A3-4. The epilimnion solids are
mostly organic, including plankton, periphyton and macrophytes, as well as detritus. Sediment
solids are balanced by the composition of the erosion load, the in-lake biotic production, and the
mineralization of OM. The predicted sediment composition is primarily sand (90%), with a
residual of silt and organic matter (10% and 5%). The burial rate is calculated internally from the
solids balance. Given the specified loads, production and mineralization rates, burial rates
averaged about 0.04 cm/yr.
Water Column Solids
20 00
18 00
1600
=5, 1400
E
c 12 00
o '*-"u
~ 1000
«
u
o 8 00
O
-------
Percent Solid Type
90.0 -,
800 -
700
60.0 -
50.0 -
40.0 -
30.0 -
20 0
0.0 -
Upper Sediment Solids Composition
0 10 20 30 40 50 60 70 80 90
Time, years
%silt
% canH
%OM
Figure A3-3. WASP Solids Simulation for Surface Sediment
Burial from Lower Sediment
10 20 30 40 50 60 70 80
Time, years
Figure A3-4. WASP Simulation of Burial Velocity
A3-8
-------
The WASP MeHg photo-degradation rate constant was adjusted to try to obtain more
reasonable MeHg concentrations. Total mercury built up over the first 100 years to 8 ng/L in the
epilimnion, 15 ng/L in the hypolimion, and 300 ng/g in the upper sediment. Methyl mercury
levels built up to 0.5 - 1 ng/L in the water, and 6 ng/g in the sediment. Although HgT is
overpredicted in the epilimnion and MeHg is underpredicted in the hypolimnion, sediment levels
match the data very well.
Note that while the model calculates high OM in the hypolimnion, much of it
conceptually is macrophyte biomass. The epilimnion HgT concentration is high, while the
hypolimnion HgT is reasonably good. The calculated sediment HgT for the upper 2 cm and the
lower 10 cm bracket the observed value.
Base Case Mercury Buildup
4 5
4
3 5
!> i
I 2.5 -
£ 2 -
u
1 15
0 1.5-
01
I 1
0 5
0
£
.X^ a
/
/
^
Epilimnion, ng/L
A Observed Epilimnion, ng/L
0 10 20 30 40 50 60 70 80 90
Time, years
Figure A3-5. WASP Total Mercury Buildup in Water
A3-9
-------
Base Case Mercury Buildup
04
035
0.3 -
1
,-- 0.25 -
o
I 0.2-
c
8
o 0.15 -
"
0)
i 0.1 .
£
0.05 -
0
f
/
/
/
^^—Epllimnion, ng/L
A Observed Epilimnion, ng/L
^
0 10 20 30 40 50 60 70 80 90
Time, years
Figure A3-6. WASP Methyl Mercury Buildup in Water
Base Case Mercury Buildup
35
30 -
o) 25 -
|,
§ 20 -
I
'c
-------
Base Case Mercury Buildup
.Upper Sediment, ng/g
.Lower Sediment, ng/g
a Observed Sediment, ng/g
10 20 30 40 50 60
Time, years
70 80 90
Figure A3-8. WASP Methyl Mercury Buildup in Sediment
After external mercury loads were reduced 50%, the mercury levels in the water column
and surface sediment declined at relatively rapid rates. Mercury attenuation seems to be
controlled by resuspension of silt and organic matter, followed by volatilization and export.
Three scenarios were simulated representing fast, medium, and slow estimates of recovery. The
upper sediment layer thickness was varied from 1 cm, 2 cm, and 3 cm. For the slow scenario,
the sediment-water dispersion coefficient was reduced by half from 10~4 cmVsec. Response
times are summarized in Table A3-9, and presented graphically for water and sediment in Figure
A3-9-Figure A3-10.
Table A3-9. WASP Response Time Estimates for Lake Waccamaw
Mercury Response Times
Compartment
Epilimnion
Surface Sediment
for Waccamaw (years)
Fast
8
10
Medium
11
15
Slow
11
17
SERAFM is 2-3 times faster in recovery than WASP, probably due to the treatment of
sediment solids balance. In particular, WASP resuspends silt and OM, which have high Kd's
rather than bulk sediment (influenced by sand), which has a lower Kd.
A3-11
-------
Lake Waccamaw Water Column Mercury Attenuation
4
4 e
3
0)
C- 25
c
o
2 9
c
0)
u
C 4 K
0 1.5
0
S 1
0.5 -
0 .
\
^V^*— ^
LT^^r-^^
,^_ &_~—
•>-,„.__ __
^^^_Fast Response
..„. Slow Response
90% Water Response,
Medium
90% Water Response,
Fast
90% Water Response,
Slow
0 5 10 15 20 25 30 35 40 45
Time, years
Figure A3-9. WASP Total Mercury Attenuation in Epilimnion
Lake Waccamaw Sediment Mercury Attenuation
25
20
O)
115
o
o
10
o>
.Fast Response
.Medium Response
.Slow Response
. 90% Water Response,
Medium
. 90% Sediment
Response, Fast
. 90% Sediment
Response, Slow
10 15 20 25 30 35 40 45
Time, years
Figure A3-10. WASP Total Mercury Attenuation in Surface Sediment
A3-12
-------
A3.5 References
North Carolina Division of Air Quality, 2002. Waccamaw Atmospheric Mercury Study. Final
Report to the United States Environmental Protection Agency Persistent Bioaccumulative
and Toxic Chemical Program. GRANT AGREEMENT # X98493600-1
3/19/2002. http://daq.state.nc.us/toxics/studies/waccamaw/PBT FINAL.pdf.
Riggs, S.R., Ames, D.V., Brant, D.R. and Sager, E.D., 2000. The Waccamaw Drainage System:
Geology and Dynamics of a Coastal Wetland, Southeastern North Carolina, East Carolina
University. Submitted to North Carolina Department of Environment and Natural
Resources, Dvision of Water Resources. September 2000.
A3-13
-------
APPENDIX A-4
MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR THE BRIER
CREEK WATERSHED (LOCATED IN THE SAVANNAH RIVER BASIN, GEORGIA)
A4.1 Background
The Brier Creek watershed is located in central/eastern portion of Georgia. The USGS
Hydrologic Unit Code (HUC) for this watershed is: 03060108 (Brier). The Brier Creek
watershed is presented in Figure A4-1.
Brier Creek Watershed
r~icat
TV Brier Creek
y Penult Compliance System
Populated Places
30
30
Figure A4-1. Brier Creek Watershed
The Brier Creek watershed has been divided into 11 subwatersheds for this analysis
(Figure A4-3), representing all of the major tributaries to Brier Creek. A total mercury load will
be determined for each of these subwatersheds to determine the impact of atmospheric
deposition on the Brier Creek.
A4-1
-------
30
30
Figure A4-2. Brier Creek Subwatersheds for Hg Loadings
The watershed contains several different types of landuses. The landuses for the Brier
Creek watershed are given in Figure A4-3. Different landuses collect and distribute mercury at
different rates as a function of runoff and erosion.
Figure A4-3. Brier Creek Watershed Landuses
This analysis covers all waterbodies in the Brier Creek watershed. Because the spatial
distribution of mercury contamination is not completely known in the streams and creeks
throughout the watershed, and fish move throughout the watershed, this analysis is developed to
protect all streams and creeks in the entire watershed from unacceptable accumulations of
mercury in fish tissue. As discussed in previous sections of this document, the State of Georgia
has issued a Fish Consumption Guideline for a segment of the Brier Creek watershed. This
guideline was issued due to elevated levels of mercury found in fish flesh collected in the
watershed.
A4-2
-------
A4.2 Mercury Deposition Network
The objective of the Mercury Deposition Network (MDN) is to develop a national
database of weekly concentrations of total mercury in precipitation and the seasonal and annual
flux of total mercury in wet deposition. The data will be used to develop information on spatial
and seasonal trends in mercury deposited to surface waters, forested watersheds, and other
sensitive receptors. Locations of the MDN sampling stations are shown on Figure A4-4.
The sampling stations with the average mercury concentration in precipitation and mercury
deposition are presented in Table A4-1.
Table A4-1. Average Mercury Deposition Hg Concentrations and Depositions Rates
Station Number
GA09
GA22
GA40
SC03
SC19
Station Name
Okefenokee National
Wildlife Refuge
Jefferson Street
Yorkville
Savannah River
Congaree Swamp
Average
Hg Cone in
Precipitation (ng/L)
14.47
15.7
15.37
13.63
14.46
14.73
Hg Dry Deposition
(ug/m2/yr)
13.25
13.15
10.49
12.07
11.35
12.06
Using the MDN data, the average annual wet deposition rate was determined to be 14.7
ug/sq. meter and the dry deposition rate was determined to be 12.1 ug/sq. meter/year.
Figure A4-4. Mercury Deposition Network Sampling Locations
A4-3
-------
A4.3 Watershed Hydrologic and Sediment Loading Model
An analysis of watershed loading could be conducted at various levels of complexity,
ranging from a simplistic gross estimate to a dynamic model that captures the detailed runoff
from the watershed to the receiving waterbody. Because of the limited amount of data available
for the Brier Creek watershed to calibrate a detailed dynamic watershed runoff model, a more
simplistic approach is taken to determine the mercury contributions to the Brier Creek from the
surrounding watershed and atmospheric components. Therefore, a scoping-level analysis of the
watershed mercury load, based on an annual mass balance of water and sediment loading from
the watershed is used for the analysis development.
Watershed-scale loading of water and sediment was simulated using the Watershed
Characterization System (WCS). The complexity of this loading function model falls between
that of a detailed simulation model, which attempts a mechanistic, time-dependent representation
of pollutant load generation and transport, and simple export coefficient models, which do not
represent temporal variability. The WCS provides a mechanistic, simplified simulation of
precipitation-driven runoff and sediment delivery yet is intended to be applicable without
calibration. Solids load, and runoff, can then be used to estimate pollutant delivery to the
receiving waterbody from the watershed. This estimate is based on pollutant concentrations in
wet and dry deposition and processed by soils in the watershed and ultimately delivered to the
receiving waterbody by runoff, erosion and direct deposition.
A4-4
-------
A4.4 Water Quality Fate and Transport Model
WASP (Ambrose, et al., 1993) was chosen to simulate mercury fate in Brier Creek.
WASP is a general dynamic mass balance framework for modeling contaminant fate and
transport in surface waters. Based on the flexible compartment modeling approach, WASP can
be applied in one, two, or three dimensions with advective and dispersive transport between
discrete physical compartments, or segments. A body of water is represented in WASP as a
series of discrete computational elements or segments. Environmental properties and chemical
concentrations are modeled as spatially constant within segments. Each variable is advected and
dispersed among water segments, and exchanged with surficial benthic segments by diffusive
mixing. Sorbed or particulate fractions may settle through water column segments and deposit to
or erode from surficial benthic segments. Within the bed, dissolved variables may migrate
downward or upward through percolation and porewater diffusion. Sorbed variables may
migrate downward or upward through net sedimentation or erosion.
Two WASP models are provided with WASP. The toxics WASP model combines a
kinetic structure adapted from EXAMS2 with the WASP transport structure and simple sediment
balance algorithms to predict dissolved and sorbed chemical concentrations in the bed and
overlying waters. WASP simulates the transport and transformation of one to three chemicals
and one to three types of particulate material. The three chemicals may be independent, such as
isomers of PCB, or they may be linked with reaction yields, such as a parent compound-daughter
product sequence. Each chemical exists as a neutral compound and up to four ionic species. The
neutral and ionic species can exist in five phases: dissolved, sorbed to dissolved organic carbon
(DOC), and sorbed to each of the up to three types of solids. Local equilibrium is assumed so
that the distribution of the chemical between each of the species and phases is defined by
distribution or partition coefficients. The model, then, is composed of up to six systems, three
chemical and three solids, for which the general WASP mass balance equation is solved.
The WASP model was parameterized to simulate the fate and transport of mercury for the
development of this analysis. Site specific and literature values were used to predict water
column concentrations as a function of flow.
A4.5 Model Results
A4-5
-------
A4.5.1 Water Quality Model
The WASP toxic chemical program was set up to simulate mercury in the mainstem of
the Brier Creek. The mainstem of the river was divided into 8 reaches. Each reach was further
divided into 2 vertical compartments representing surface water and surficial sediment. The 2
cm deep surficial sediment layer actively exchanges silt and clay-sized solids as well as
chemicals within the water column. In addition, this layer is the site for active microbial
transformation reactions. Sediment-water column diffusion coefficients were set at TO"5 cmVsec.
Two solids classes were simulated: sand and silt. Sand makes up most of the benthic
sediment compartments, which have a dry bulk density of 0.5 g/ml. Given a particle density of
2.7 g/ml, the sediment porosity is about 0.8 and the bulk density is 1.3 g/ml. Silt is found both
suspended in the water column and in the sediment. These simulations assumed that 10 mg/L of
silt enters the mainstem from the subwatersheds, settling out at an assumed velocity of 0.3
m/day. Silt in the surficial sediment compartments is assumed to resuspend at a velocity of
0.006 m/day, giving a concentration of about 0.005 g/ml, or about 1% of the surficial sediment.
The exchanging silt carries sorbed mercury between the water column and surficial sediment.
Mercury was simulated as 3 components - elemental mercury, Hg°; inorganic divalent
mercury, Hg(II); and monomethylmercury, MeHg. Hg(II) and MeHg partition to solids and
dissolved organic carbon (DOC). These are represented as equilibrium reactions governed by
specified partition coefficients. The three mercury components are also subject to several
transformation reactions, including oxidation of Hg° in the water column, reduction and
methylation of Hg(II) in the water column and sediment layer, and demethylation of MeHg in
the water column and sediment layer. These are represented as first-order reactions governed by
specified rate constants. Reduction and demethylation are driven by sunlight, and the specified
surface rate constants are averaged through the water column assuming a constant light
extinction coefficient (here, 0.5 m"1). In addition to these transformations, Hg° is subject to
volatile loss from the water column. This reaction is governed by a transfer rate calculated from
velocity and depth, and by Henry's Law constant, which was set to 7.1 x 10"3 L-atm/mole-K.
Under average flow conditions, velocity ranges from 0.2 to 0.3 m/sec, while depth ranges from
0.37 to 0.69 m. The specified and calculated reaction coefficients used here are summarized in
Table A4-2.
Table A4-2. Specified and Calculated Reaction Rates and Coefficients
Component
Hg°
Hg(H)
Reaction
Volatilization
Oxidation
Reduction
Methylation
Methylation
Partitioning to silt
Partitioning to sand
Partitioning to DOC
Compartment
Water
Water
Water surface
Water column
Water
Sediment
Water, Sediment
Water, Sediment
Water, Sediment
Coefficient Value
0.3 - 3.0 day'1 (calc)
0.001 day'1
0.10 day'1
0.03-0.05 (calc)
0.001 day'1
0.0005 day'1
1 x 10s L/kg
1 x 103 L/kg
1 x 105 L/kg
A4-6
-------
MeHg
Demethylation to Hg(II)
Demethylation to Hg°
Partitioning to silt
Partitioning to sand
Partitioning to DOC
Sediment
Water surface
Water column
Water, Sediment
Water, Sediment
Water, Sediment
0.005 day'1
0.05 day'1
0.015-0.025
1 x 105L/kg
1 x 102L/kg
2 x 105 L/kg
The Brier Creek simulation was conducted using annual average flow and load. The
average flow simulation was run for 30 years, so that steady-state conditions are achieved in the
water and surficial sediment. The flows, depths, length, widths, and volumes used for annual
average conditions are summarized in Table A4-3.
Table A4-3. Flows, Depths, Length and Volumes used in WASP Model
Segment
Upper Brier/Bushy
Upper Middle
Upper Middle 2
Upper Middle 3
Lower 1
Lower 2
Lower 4
Beaverdam
Length (m)
3494.0
13804.7
13804.7
13804.7
13804.7
13804.7
13804.7
18180.0
Width (m)
54.0
71.2
78.3
85.5
92.7
99.8
107.0
107.0
Depth (m)
.2
.3
.4
.6
.7
.8
.9
.9
Flow (cfs)
560.0
139.7
519.3
599.0
578.7
758.3
J38.0
538.0
Volume (cubic
neters)
729139.2
1295173
1553562
1835365
2140582
2469214
2821260
3715447
The Watershed Characterization System calculates mercury loadings to each reach.
These values are specified as constant Hg(II) and MeHg loadings for each surface water
compartment. Loadings for average flow conditions reflect both wet and dry deposition
throughout the watershed, followed by runoff and erosion to the tributary stream network.
Table A4-4 compares the measured sediments characteristics in Brier Creek with the
predicted concentrations and conditions from WASP.
Table A4-4. Measured vs. Predicted for Sediment Components
River Station
BC01 water
BC02 water
BC01 sediment
BC02 sediment
Wasp
Segment
1
8
9
16
Measured
TSS, mg/L
16
4
Calculated TSS,
mg/L
15
8
Measured
VolS fraction
0.09
0.02
Calculated OM
fraction*
0.07
0.04
Table A4-5 provides the predicted water column concentrations under annual average
load and flow for the Brier Creek. The highest predicted water column concentration is used in
the analysis calculation to determine the maximum annual average load that could occur and still
achieve the target.
A4-7
-------
Table A4-5. Predicted and Observed Mercury Concentrations under Annual Average
Load and Flow
Component
HgT, ng/L
MeHg, ng/L
HgT, ng/g
MeHg, ng/g
WASP Reach
1
7.75
0.74
35.5
3.4
2
7.67
0.84
32.9
3.3
3
7.51
0.91
31.3
3.2
4
7.49
1.03
28.3
3.1
5
7.38
1.08
27.0
3.0
6
7.30
1.12
24.0
2.7
7
7.21
1.17
23.0
2.7
8
6.92
1.20
20.0
2.4
A4.6 Brier Creek Watershed Results
Table A4-6 provides measured soil mercury concentrations for both the Brier Creek
watershed and the larger Savannah River watershed which virtually surrounds Brier Creek.
Table A4-6. Soil Mercury Data in Local Region
Basin
Brier Creek
Ogeechee
Canoochee
Savannah
Mean
Standard
Deviation
Station
BC01
BC02
OG1
OG2
OG3
OG4
CAN01
CAN02
Below Horse Cr.
Below Horse Cr.
Clarks Hill
Below Clarks Hill
Below Clarks Hill
Below Butler Creek
Below Upper Three Runs
Creek
Below Lower Three Runs
Creek
Below Brier Creek
Clyo, USGS Gage
Below Ebenezer Creek
Butler Creek
Horse Cr.
Upper Three Runs Creek
Lower Three Runs Creek
Brier Creek
Ebenezer Creek
%VS
27.0
11.0
4.3
1.8
11.0
15.0
6.2
14.0
20.0
16.0
10.0
3.9
2.8
5.2
5.2
6.9
7.1
7.8
16.0
4.4
3.9
%Moisture
42
20
8.2
3.6
14
47
25
23
32.7
32.4
29.8
24.3
24.3
20.9
17.5
17.1
9.2
21.9
5.5
3
17.4
17.5
33.9
16
8.3
20.6
11.4
THg,
ug/kg
130.0
75.0
26.0
13.0
30.0
47.0
32.0
71.0
133.0
41.2
67.2
78.6
80.8
33.1
22.7
56.8
43.6
71.8
33.9
43.8
43.6
56.4
137.7
26.3
28.1
56.9
34.6
MeHg
ug/kg
0.740
0.110
0.028
0.035
0.019
0.940
1.800
0.010
0.054
0.065
2.050
0.042
0.031
0.052
0.003
0.257
0.949
0.011
0.063
0.009
0.013
0.543
0.319
0.109
0.344
0.569
%
MeHg
0.57
0.15
0.11
0.27
0.06
2.00
5.63
0.01
0.04
0.16
3.05
0.05
0.00
0.09
0.23
0.00
0.59
1.32
0.03
0.14
0.02
0.02
0.39
1.21
0.39
0.66
1.27
A4.6.1 Brier Creek Soil Mercury Calibration
A4-8
-------
The WCS Mercury model was applied and run for 100 years to equilibrate the watershed
soils with atmospheric conditions. Figure A4-5 illustrates the buildup of mercury in the soils
over the 100 year period.
Soil Concentration, mg/kg
Brier Creek Soil Mercury Buildup
0 0700
0 0650
0.0600
0 0550
0 0500
0.0450
0 0400 -
0 0350 -
n ntnn
^ ~- — ~~ ~~"
^^-^ ~~^
-------
A4.6.2 Mercury Loading Fluxes
Figure A4-6 illustrates the mercury loadings from the delineated subwatersheds as
predicted by the WCS over the 100 year equilibration period.
30
25
h.
><
I) 20
E
K
-a
ra
10
20
Brier Creek Loading Flux Buildup
40
60
Time, yr
80
100
.Upper Brier
. Brushy Creek
Above Brushy
.Upper Middle Brier
. Lower Brier 1
.Upper Middle Brier 2
.Upper Middle 3
. Beaverdam Cr
Lower Brier 2
.Lower Brier 3
Lower Brier 4
Figure A4-6. Brier Creek Loading Flux Buildup
There is significant variability in loading fluxes from the 11 subwatersheds in the Brier
Creek watershed. Loads at year 100 were used in the water body calibration.
A4-10
-------
A4.6.3 Future Projections
For these projections, the atmospheric loads were cut in half, and the watershed response
was followed for 100 years in the Upper Brier Creek subwatershed (Figure A4-7).
Upper Brier Creek (WS-1) Soil Mercury Attenuation
0.0600
Figure A4-7. Upper Brier Creek Soil Mercury Attenuation
The half life of the soil attenuation response is about 25 years.
The projected loading flux for Upper Brier Creek was projected over 100 years. The
loading fluxes for the other 10 subwatersheds were calculated over 50 years (Figure A4-8).
A4-11
-------
Brier Creek Loading Flux Attenuation
0 10 20 30 40 50 60 70 80 90 100
• Upper Brier
• Brushy Creek
Above Brushy
Upper Middle Brier
-Upper Middle Brier 2
• Upper Middle 3
• Beaverdam Cr
- Lower Brier 1
• Lower Brier 2
Lower Brier 3
Lower Brier 4
Figure A4-8. Brier Creek Loading Flux Attenuation
A4-12
-------
A4.6.4 Sensitivity of Temporal Response
The depth of soil incorporation significantly influences soil response time. The default of
1 cm was varied plus and minus 50% to get a range of response times. This parameter should
not affect concentrations or loadings at steady-state. Figure A4-7 illustrates the response time of
the upper soil layer to decreases in mercury loadings from the atmosphere.
Soil Concentration, mg/kg
0.0600 -,
0.0550
0.0500
0.0450
0.0400
0.0350
0.0300
(
Upper Brier Creek (WS-1) Soil Mercury Attenuation
V
\X
XH^v
X>\^^
^^c^^,^^
"^^rrrrr-
Mean
M + SD
M - SD
) 10 20 30 40 50 60 70 80 90 100
Time, yr
Figure A4-9. Upper Brier Creek Soil Mercury Attenuation
The half life of the soil mercury attenuation response for the base simulation in Upper
Brier Creek was 25 years, and the 90% response was 90 years. Varying the incorporation depth
50% caused the half life to vary between 12 and 38 years. The 90% response time varied even
more, between 45 and about 110 years.
The loading responses for Upper Brier Creek are given in Figure A4-10. An initial rapid
drop-off in loading (due to instantaneous drop in deposition to water surfaces and impervious
runoff) is followed by a slower drop-off in runoff and erosion fluxes, controlled by soil mercury
concentrations. The 50% loading response varied between 8, 10, and 15 years for incorporation
depths of 0.5, 1.0, and 1.5 cm. The 90% loading response times were much longer, varying
between 35, 70, and 100 years, respectively.
A4-13
-------
10 4
Upper Brier Creek Loading Flux Attenuation
.Upper Brier, base case
.Upper Brier, 2=0.5
.Upper Brier, z=1.5
.50% Response
.90% Response
.100% Response
10 20
40 50 60 70
Time, yr
80 90 100
Figure A4-10. Upper Brier Creek Loading Flux Attenuation
Land use changes can also significantly affect future loading response from a watershed.
In particular, changing pervious land use areas to impervious areas will directly affect the
delivery of atmospheric deposition fluxes to the water body. In this model, impervious areas
were assumed to deliver 100% of the deposition load, while pervious areas delivered a much
smaller fraction through runoff and erosion. Although the fraction of the Brier Creek watershed
covered by impervious surfaces is small (about 3% of the upper watershed), even modest growth
over many years could increase the total watershed delivery of deposited mercury, working
against the overall reductions in atmospheric emissions. The loading response of Upper Brier
Creek to 3 land use scenarios is shown in the next figure. All scenarios assume an immediate
50% cut in atmospheric deposition. The base case assumes present land use patterns. The other
two scenarios assume modest impervious surface growth rates of 0.5 % per year and 1 % per
year. For the 0.5% scenario, total watershed loads reach a minimum in 80 years, with watershed
loadings stalling at 40% of present levels. For the 1% scenario, total watershed loads reach a
reduction level of 33% in 50 years, and then increase significantly. After 100 years, the 50% cut
in atmospheric deposition would translate into a 25% drop in ambient watershed loading (Figure
A4-11).
A4-14
-------
Upper Brier Creek Loading Flux Attenuation
28
26
24
w
?• 22
n " "
f 20
t *" •
X
3 18
u. 10 -
O)
1 16
^
o
J 14
12
10
\
\
V
^^^^ _— — ^^"
^^^^^^
^^*~—~^.^
•
Upper Brier, base case
0.5% Impervious Growth
Rate
1.0% Impervious Growth
Rate
50% Response
90% Response
0 10 20 30 40 50 60 70 80 90 100
Time, yr
Figure A4-11. Watershed Loading Flux Attenuation considering Landuse Change
A4-15
-------
A4.7 Brier Creek Water Body Results
A4.7.1 Phase 1: Long Term Buildup
Long-term predicted watershed loadings were applied to the Brier Creek water body
network to simulate the long-term buildup of mercury in the water and sediment. Average flows
were used for this entire simulation. The watershed loadings from year 30 to 100 in the
watershed simulation were used in the first 70 years of this water body simulation. For the last
20 years of the water body simulation, watershed loadings were held constant (Figure A4-12).
Base Case Mercury Buildup
.Upstream Water
.Downstream Water
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85
Time, years
Figure A4-12 Base Case Water Column Mercury Concentration for Brier Creek
Initial low water column concentrations jumped quickly from a background of 1 ng/L to
near 5 to 6 ng/L in response to the loadings. Following this initial response to the industrial era
loadings, the water column mercury slowly increased in proportion to the slowly increasing
watershed loadings.
Mercury concentrations in the upper sediment (2 cm layer) followed the same pattern,
increasing from an initial 5 ng/g to a steady 20 - 35 ng/g. In response to slow internal mixing,
mercury concentrations in the lower sediment (2 - 12 cm layer) slowly increased throughout the
simulation from a background of 5 ng/g to the range of 10-20 ng/g. Lower sediment
concentrations were still increasing at the end of this 85 year simulation.
A4-16
-------
The model dynamics follow the expected pattern of a water column and upper sediment
responding relatively quickly to external loads (a few years), and a lower sediment layer
responding slowly (decades).
Base Case Mercury Buildup
20
30 40 50 60
Time, years
70 80
• Upstream Sediment
Surface
• Downstream Sediment
Surface
Upstream Subsurface
• Downstream Subsurface
Figure A4-13. Base Case Sediment Mercury Concentration for Brier Creek
A4.7.2 Phase 2: Response to 2002 Flows
Mercury concentrations in Brier Creek were measured in a June, 2002 survey. Table A4-
1 illustrates the model predictions for water column and sediment concentrations versus model
predictions. Figure A4-14 - Figure A4-15 provide a graph of model predictions for both total
mercury and methyl mercury over time. A more dynamic simulation was conducted using daily
stream flow data for December 2001 - July 2002. Mercury loadings were assumed to be
proportional to incremental inflows from the subwatersheds. In response to changing flows and
depths, but constant inflow concentrations, simulated water column mercury levels fluctuated
mildly. Sediment levels (not shown) changed very slowly, further buffering the water column
from major fluctuations. The predicted mercury levels compared favorably with the
observations.
A4-17
-------
Table A4-7. June 2003 Survey vs WASP Predictions for Mercury
River Station
BC01 water
BC02 water
BC01 sediment
BC02 sediment
Wasp
Segment
1
8
9+17
16+24
Measured
HgT, ng/L
8.3
6.0
37.0
6.4
Calculated
HgT, ng/*
7.8
6.3
18.6
12.0
Measured
MeHg,
ng/*
0.73
1.40
5.70
0.04
*Lors
Calculated
MeHg,
ng/*
0.80
0.86
1.7
1.3
Measured
MeHg
fraction
0.09
0.23
0.15
0.01
Calculated
MeHg
fraction
0.10
0.14
0.09
0.10
Brier Creek Water Column Mercury, 2002
q
8 5
8
7E
"3>
c 7
£
6.5
g
55
5
A
—- - • ~ _
^N^*- ,-^/\s~ — -— «^
Upstream
— Downstream
A Measured Upstream
e Measured Downstrearr
•30 0 30 60 90 120 150
Time, days
Figure A4-14. Brier Creek Total Mercury Water Column Concentration
A4-18
-------
Brier Creek Water Column Mercury, 2002
1 6
1 4
1 2
1
'a
5«-
1
s 0.6 -
0 4 -
0 2
0 -
A
ZA
^***v VX. ^"" ,-
>MI i^»^ ^»«^^ '• "V
^^^*--^
D
Upstream
— Downstream
a Measured Upstream
A Measured Downstream
-30 0 30 60 90 120 150
Time, days
Figure A4-15. Brier Creek Methyl Mercury Water Column Concentration
A4-19
-------
A4.7.3 Phase 3: Future Attenuation
A 55 year simulation was conducted to explore mercury response to declining watershed
loads due to an immediate decline in atmospheric deposition of 50% (year 5 of the water body
simulation). Figure 16 illustrates the change in mercury concentration as a function of load
reduction in the water column. Figure A4-17 illustrates the change in mercury concentration as a
function of load reduction in the sediments.
Base Case Mercury Attenuation
.Upstream Water
.Downstream Water
5 10 15 20 25 30 35 40 45 50
Time, years
Figure A4-16. Mercury Attenuation over Time in Water Column
A4-20
-------
Base Case Mercury Attenuation
40
35 -
"U 30
c
c 25
.0 ** -
2 20
o>
y 15
0
" 10
o>
1 5-
0 -
\
\^_^
— -_ ' " — -
___JI^^— ^ — • —
— — ' — _
Surface Upstream
Surface Downstream
U ps t ream SubsurfacG
0 5 10 15 20 25 30 35 40 45 50
Time, years
Figure A4-17. Mercury Attenuation over Time in Sediments
Water body and upper sediment concentrations were predicted to decline more rapidly in
the first 10 years, and then more slowly for the next 4 decades. The initial decline is due to the
rapid decline in direct loadings to the water surface and from impervious areas. The later slow
decline is due to the slowly dropping soil concentrations and the internal recycling of sediment
mercury. Ultimately, the predicted water body concentrations should decline proportionally to
the declining loadings. The first half of this response takes about a decade, while 90% of the
response should take on the order of 50 years. The remainder of the water body response should
be even slower, as the lower sediment concentrations become the controlling factor.
Uncertainties in this response include watershed loadings and internal water body
conditions, including infrequent occurrences of very high scouring flows, and sediment mixing
characteristics.
A4-21
-------
A4.7.4 Sensitivity of Time Response
The depth of soil incorporation in the watershed significantly influences soil response
time and loadings to the stream (see previous section), and thus concentrations in the stream.
The default of 1 cm was varied plus and minus 50% to get a range of response times. Within the
water body, the sediment incorporation depth should have a similar influence. The Brier Creek
model was set up with a 2 cm active sediment layer and a 10 cm lower sediment layer. Transport
between these layers was controlled by a bulk dispersion parameter, which mixes pore water,
solids, and mercury. The calibrated base value of 10"8 cmVsec was varied between 10"6 and 10"9
cm2/sec to get a range of response times. These parameters should not affect concentrations or
loadings at steady-state.
These sensitivity runs demonstrate that attenuation of mercury in the Brier Creek water
will be controlled more by watershed loadings than by internal mixing processes. The next
figure shows the response of upper Brier Creek to the three reduction scenarios. The 50%
watershed loading responses of 8, 10, and 15 years for incorporation depths of 0.5, 1.0, and 1.5
cm produced 50% stream concentration responses of 9, 12, and 18 years. The 90% loading
response times were much longer. The fast, medium, and slow stream response scenarios
reached the 90% mark in 40, 74, and 113 years, respectively.
The mercury concentration response in lower Brier Creek follows a more complicated
time trend, influenced not only by the watershed loadings and internal mixing, but also by the
upper Brier Creek dynamics. The 50% response times are 18, 15, and 22 years, for the fast,
base, and slow scenarios. The downstream base and slow scenarios are 3 to 4 years longer than
the upstream response. The downstream fast response scenario, however, takes 9 years longer
than the upstream reach, as the enhanced mixing results in significant short term internal
sediment fluxes. The 90% response times for the fast, medium, and slow response scenarios
were 58, 89, and 132 years, which is 9 - 18 years longer than the upstream response.
A4-22
-------
Sensitivity Range for Upstream Water
0 10 20 30 40 50 60 70 80 90 100
Time, years
.Base Calibration
. Fast Mixing and
Watershed Response
.Slow Mixing and
Watershed Response
.50% Response
.90% Response
.100% Response
Figure A4-18. Sensitivity Range for Upstream Waters
A4-23
-------
Sensitivity Range for Downstream Water
9
8
_i
1* 7 -
o
^
2 6
0)
u
o
O 5 .
O)
X
4
3
^
\
V.
^S^1^-
^^ferr-—
- — =
Base Calibration
Fast Mixing and
Watershed Response
__Slow Mixing and
Watershed Response
50% Response
90% Response
100% Response
0 10 20 30 40 50 60 70 80 90 100
Time, years
Figure A4-19. Sensitivity Range for Downstream Waters
A4-24
-------
APPENDIX A-5
MERCURY LOAD REDUCTION ANALYSIS AND RESPONSE FOR LAKE BARCO
(FLORIDA)
A5.1 Introduction
Lake Barco is a small, seepage lake in northeast Florida with has a water surface area of
0.12 km2 and a negligible catchment area. Lake Barco is located ca. 35 km east of Gainesville,
Florida on the Ordway Preserve that is operated by the University of Florida. The Ordway is
protected from direct human impacts, although some recreational fishing does take place.
Hydrology and geochemistry of Lake Barco have been well characterized by past studies
(Pollman, Lee et al. 1991; EPRI 2003). There are several nearby mercury sources. For example,
Gainesville Regional Utilities (GRU) operates a medium-size coal-fired power plant
approximately 40 km NW of Lake Barco. Emissions of Hg from the Deerhaven Unit No. 2
facility averaged ca. 30 kg/yr between 1998 and 2002 (range 13 to 47 kg/yr).
A5.2 Empirical Data from Lake Barco
Site specific data used to model Lake Barco were obtained from an EPRI report (EPRI
2003), and unpublished data collected by Tetra Tech, Inc. (Schofield 1998) and associates (Pers.
Comm., C. Pollman, Tetra Tech, Inc., 2005).
A5.3 SERAFM Application: Lake Barco
Table A5-1. Lake Barco Parameter Values
Parameter
Watershed Area
Percent Impervious
Percent Forest
Percent Riparian
Percent Upland
Lake Area
Catchment/Lake Ratio
Epilimnion Depth
Hypolimnion Depth
Hypolimnion Anoxia
Hydraulic Residence Time
Inflow/Outflow
Water pH
Epilimnion DOC
Trophic Status
Annual Precipitation
HgH Cone, in Precip
Wet Deposition (Hgll)
Dry Deposition (Hgll)
Wet Deposition (Hgll)
Dry Deposition (Hgll)
Lake Barco
0
n/a
n/a
n/a
n/a
118,000m2
0
3.7m
n/a
n/a
n/a
0
4.5
0.85 mg/L
Oligotrophic
134.8 cm/yr
11.5ng/L
15.5 ug/m2/yr
15.5 ug/m2/yr
0.23 ug/m2/yr
0.23 ug/m2/yr
A5-1
-------
Table A5-2. Measured and Baseline Steady State Values for Lake Barco
Parameter
Water Column MeHg
Unfiltered
Water Column HgH Unfiltered
Sediment MeHg
Sediment Hgll
Largemouth Bass Tissue Hg
Observed BAF: FishHg/MeHg
Measured
0.018ng/L
1.03 ng/L
1.9-6.9ng/grdryl
152-186ng/g[dryl
2 yr old Igmouth bass:
0.55 ± 0.25 ug/g [wet]
Predicted
0.040 ng/L
0.96 ng/L
6.2 ng/g [dry]
177.45 ng/g [dry]
1.2 ug/g
(0.7-1.7)
3.06xl07
Table A5-3. Lake Barco SERAFM Calibrated Rate Constants
Lake Barco Calibrated Rate Constants
Process
Media
Value
Units
Methyl ation
Demethylation
Biotic Reduction
Photo-Degradation
Photo-Reduction (Vis)
Photo-Reduction (UV-B)
Photo-Oxidation (UV-B)
Dark Oxidation
Epilimnion
Sediment
Epilimnion
Sediment
Water Column
Water Column
Water Column
Water Column
Water Column
Water Column
0
0.014
0.0001
0.40
0.03
0.002
0.0003
0.02825
0.05885
1.44
Per day
Per day
Per day
per day
per day
per day per E/m2-day
per day per E/m2-day
per day per E/m2-day
per day per E/m2-day
per day
Table A5-4. Time to Reach 90% Steady State After 50% Reduction in Atmospheric
Deposition
Lake Barco
Epilimnion
Hypolimnion
Sediment
Fish
Fast
13
14
14
Med
27
28
28
Slow
41
45
43
Fast = 1 cm active sediment layer, D (macro-dispersion coefficient) = 10"' cm2/s
Medium =2 cm active sediment layer, D = 70~* cnf/s
Slow=3 cm sediment, D = 5xl&5 cm2/s
A5-2
-------
Table A5-5. SERAFM Model Forecasts with Zero-Out Scenario for Coal-Fired Power
Plants
(Medium Response Time Scenario)
Lake Barco Time
Epilimnion 27
Epilimnion
MeHg
Hypolimnion
Sediment 28
Fish 28
AS.4 References
EPRI (2003). Factors Affecting Predicted Responses of Fish Mercury Concentrations to Changes
in Mercury Loading. Palo Alto, CA, Electric Power Research Institute Report 1005521.
Pollman, C., T. Lee, et al. (1991). "Preliminary analysis of the hydrologic and geochemical
controls on acid-neutralizing capcity in two acidic seepage lakes in Florida." Water
Resources Research 27(9): 2321-2355.
Schofield, C. (1998). Mercury Bioaccumulation in Lake Barco Centrachids. Final Report.
Gainesville, FL, Tetra Tech, Inc.
A5-3
-------
APPENDIX B QUALITATIVE ECOLOGICAL REVIEW OF MERCURY LITERATURE B-l
B.I Introduction B-l
B.2 Potential Exposure Media B-2
B.2.1 Mercury in Air B-2
B.2.2 Mercury in Water B-2
B.2.3 Mercury in Soil B-2
B.3 Bioaccumulation of Mercury B-3
B.4 Exposure and Toxic Effects in Wildlife B-4
B.4.1 Aquatic Plant Species B-4
B.4.2 Aquatic Invertebrate Species B-4
B.4.3 Fish and Amphibian Species B-6
B.4.4 Terrestrial Plant Species B-8
B.4.5 Terrestrial Invertebrate Species B-8
B.4.6 Avian Species B-9
B.4.7 Mammalian Species B-ll
B.5 Ecosystems Potentially Affected B-13
B.6 Conclusions B-13
B.7 References B-14
-------
APPENDIX B
QUALITATIVE ECOLOGICAL REVIEW OF MERCURY LITERATURE
B.I Introduction
Section 3 describes the Agency's approach for identifying ecological benefits that may
result from reductions in mercury emissions. A comprehensive quantitative analysis is not
possible at this time given the current state of the science. However, research on the ecological
effects of mercury exposures provides some qualitative support to the notion that reductions in
mercury emissions could contribute to improvements in overall ecosystem health. This
Appendix expands on the analysis presented in Sections 2, 3 and Appendix A by reviewing some
of the recent scientific findings, focusing on studies not included in and/or published since the
release of the Mercury Study Report to Congress: Volume VI in December 1997. This review is
not a quantitative ecological toxicity assessment and the information presented here does not
identify a "safe" ecological exposure level. The bulk of this research, based on both laboratory
and field studies, suggests that because mercury is persistent in the environment and
biomagnifies up the food chain when methylated, a wide variety of species and ecosystems may
be harmed by excessive levels of mercury in the environment. At the outset, however, it must be
noted that these studies do not distinguish between mercury from emissions of U.S. coal-fired
utilities, other current mercury emissions, and legacy mercury.
Numerous studies have generated field data on the levels of mercury in environmental
media as well as a variety of wild species. Comparison of contemporary measurements of
atmospheric mercury and historical measures from lake sediment indicate that the amount of
global atmospheric mercury has risen by a factor of 2-5 since the start of the industrial period
(Boening, 2000). However, both the sediment and glacier core record have shown declining
trends in mercury deposition since the early 1990s (Schuster et al., 2002). The body of work
examining the effects of these exposures is still incomplete. A large portion of the research
conducted to date has been carried out in the laboratory setting rather than in the wild;
conclusions about effects on natural populations and overall ecosystem health are difficult to
make at this time.
Extensive laboratory-based studies of mercury toxicity using captive-bred animals (mice,
rats, monkeys, etc.) have been conducted. From this research, much has been learned about the
mechanisms of mercury toxicity; unfortunately, many of these captive-bred test species may be
genetically different from their wild relatives since they are bred to ensure consistency in study
results. Also, exposures of animals in laboratory-based studies are not always comparable to
average exposures in wild. Often times, a laboratory-based study will identify a potential
adverse effect in a specific species, but the ability to determine if this effect is present in the wild
is limited by the understanding of mercury fate, transport and exposure dynamics in any given
ecosystem. As a result, laboratory-based experiments using captive-bred animals lacking genetic
variation and environmentally relevant exposures may not accurately reflect the effects of
mercury toxicity in the natural environment. Further study of ecosystem-specific exposures will
be necessary to determine the nature and magnitude of the risks posed to wild species by
mercury.
B-l
-------
The review that follows seeks to summarize some of the adverse effects seen in
organisms exposed to mercury in both laboratory and wild settings. The following review
should help to illustrate that all living organisms and the ecosystems they co-inhabit are
influenced to some degree by mercury pollution. At this time, the magnitude of the power plant
contribution to ecological exposures cannot be quantified nationally, and the corresponding risk
for adverse effect is difficult to determine. While the benefit of further reducing mercury
emissions cannot be quantified for ecosystems at this time, we have described this benefit
qualitatively for context.
B.2 Potential Exposure Media
B.2.1 Mercury in Air
Atmospheric deposition of elemental mercury (Hg°) in the vapor phase is the primary
pathway of global deposition (Boening, 2000). Measurable concentrations of elemental mercury
vapor may be detected in ambient air, especially near sources of mercury emission (U.S.EPA,
1997). The rate and mechanism by which mercury is deposited to the earth's surface depends on
the chemical form present in the air (U.S.EPA, 1997). Atmospheric mercury also exists in the
divalent (Hg2+) state but is often associated with particulate matter when in this form. Mercury
associated with particulate matter typically settles out of the atmosphere at a faster rate than Hg°
in the vapor phase. Wet deposition of mercury with precipitation is also possible as Hg2+ is
soluble in water.
B.2.2 Mercury in Water
Mercury can enter surface water as elemental mercury, divalent mercury, or
methylmercury (MeHg or CH3HgCl). Once in aquatic systems, mercury can exist in dissolved
forms or in complexes sorbed to particles and can undergo a variety of physical or chemical
transformations (such as oxidization, reduction, methylation, and demethylation) depending on
its form and conditions. Dissolved Hg° can volatilize from the water back into the atmosphere,
and particulate mercury forms can become buried in the sediment bed. The percent of total
mercury in surface waters that exists as methylmercury varies, but it has been estimated to
average about eight percent, with some estimates as high as 20 percent (U.S.EPA, 1997).
Within a surface water body, contaminated sediments can act as an important mercury
reservoir for decades, with mercury recycling from the sediment back into the aquatic ecosystem.
Biological processes, such as the methylation of Hg2+ by sulfate-reducing bacteria and
accumulation of mercury in benthic invertebrates, mediate the availability of mercury to other
aquatic animals in the food chain (U.S.EPA, 1997).
B.2.3 Mercury in Soil
Mercury compounds enter the soil through atmospheric deposition and form stable
complexes with soil particles (U.S.EPA, 1997). Both Hg2+ and Hg° are present in soil; however,
these mercury species may be transformed into methylmercury and back by bacteria or other
living organisms (Jereb et al., 2003). Both organic and inorganic forms of mercury undergo
environmental transformation (Boening, 2000). Aquatic organisms are known to be capable of
B-2
-------
transforming mercury species (Gilmour and Henry, 1991). Less is known about mercury
transformation in soil ecosystems. A recent in vivo laboratory study of the terrestrial isopod
crustacean, Porcellio scaber, found that both methylation and demethylation of mercury
compounds are possible in this organism (Jereb et al., 2003).l Thus, many of the mechanisms
affecting biomagnification in the aquatic ecosystem are also possible in terrestrial soil
ecosystems.
B.3 Bioaccumulation of Mercury
Mercury is one of the few metals that has been demonstrated to biomagnify in food
chains (U.S.EPA, 1997), resulting in increasing tissue concentrations of mercury in organisms at
successively higher trophic levels. Three terms are commonly used to describe the mechanism
by which a contaminant accumulates in living tissues - bioconcentration, biomagnification, and
bioaccumulation. The term "bioconcentration" refers to the accumulation of a chemical that
occurs as a result of direct contact of an organism with its surrounding medium (e.g., uptake by a
fish from water through the gills and epithelial tissue or uptake by earthworms from soil pore
water through the skin) and does not include the ingestion of contaminated food. The term
"biomagnification" refers to the increase in chemical concentration in organisms at successively
higher trophic levels as a result of the ingestion of contaminated organisms at lower trophic
levels. The term "bioaccumulation" refers to the net uptake of a contaminant from all possible
pathways and includes the accumulation that may occur by direct exposure to contaminated
media as well as uptake from food. Mercury, in its various forms, can bioconcentrate,
bioaccumulate, and biomagnify. Biomagnification of mercury is apparent in many aquatic
ecosystems, particularly in aquatic systems with food chains that depend on benthic organisms
which live in the sediments, the primary site of methylation of inorganic mercury. Mercury also
magnifies in terrestrial food chains, but to a lesser extent than in aquatic ones owing to lower
levels of mercury methylation, less accumulation of mercury in organisms at the base of the food
chain, and shorter food chains.
All forms of mercury can bioaccumulate to some degree; however, methylmercury
generally accumulates to a greater extent than other forms because of its ability to biomagnify.
Methylmercury is absorbed into tissues quickly, where it becomes sequestered. Inorganic
mercury can also be absorbed, but usually at a slower rate and with lower efficiency than
methylmercury. Elimination of methylmercury from fish is so slow that long-term reductions of
mercury concentrations in fish are often due mainly to growth of the fish ("growth dilution"),
whereas other mercury compounds are eliminated relatively quickly. Therefore, methylmercury
(and thus total mercury) concentrations tend to increase in aquatic organisms as the trophic level
in aquatic food webs increases. In addition, the proportion of total mercury that exists as
methylmercury generally increases with trophic level (U.S.EPA, 1997).
B.4 Exposure and Toxic Effects in Wildlife
1 In vivo studies involve experiments on biochemical reactions inside living organisms or cells. In vitro studies
investigate reactions outside of cells, and therefore may be less representative of responses by organisms in the
natural environment than in vivo studies.
B-3
-------
A large portion of mercury research conducted to date has focused on the effects of
mercury exposure on aquatic ecosystems. In general, organic mercury species are more toxic to
aquatic organisms than inorganic mercury, and toxicity increases with temperature and decreases
with water hardness (Boening, 2000).
B.4.1 Aquatic Plant Species
Accumulation of mercury in the primary producers at the base of the aquatic food web
can have substantial impacts on the amount of mercury available to aquatic animals, not only via
direct ingestion of the plants, but also from ingestion of detritus derived from the decomposition
of plant material (Boening, 2000). Effects of mercury on aquatic plants include senescence,
growth inhibition, decreased chlorophyll content and dry weight, decreased protein and RNA
content, inhibited catalase and protease activities, inhibited and abnormal mitotic activity,
increased free amino acid production, discoloration of floating leaves, and leaf and root necrosis
(U.S.EPA, 1997; Boening, 2000). The level of mercury that results in toxic effects varies greatly
among aquatic plant groups and species (U.S.EPA, 1997). Inorganic mercury concentrations in
water of approximately 1,000 (ig/L can adversely affect aquatic plants. In general, organic
mercury compounds are more toxic to aquatic plants than inorganic forms (Boening, 2000).
Studies of inorganic mercury uptake by aquatic plants indicate that bioconcentration (as
measured by the ratio of mercury concentration in tissue to mercury concentration in water)
decreases with increasing concentrations of mercury in the water column. In laboratory
experiments with concentrations of mercuric chloride (HgCl2) ranging from 50 to 20,000 [ig/L,
maximum bioconcentration in water cabbage (Pistia stratiotes) occurred at water concentrations
of 6,000 [ig/L or less (Boening, 2000). Although mercury concentrations in the plants did
increase with increasing mercury concentrations in the water column, only 20 percent
accumulated at the highest concentration. Other studies of vascular aquatic plants have shown
that mercury uptake may occur in roots rather than stems or shoots, with two to four times the
accumulation of mercury in the roots over the shoots (Boening, 2000).
B.4.2 Aquatic Invertebrate Species
In general, aquatic invertebrate species vary widely in their susceptibility to mercury
toxicity, although most are more sensitive during the larval stage than during other life stages
(Boening, 2000). The concentration and speciation of mercury, developmental stage of the
organism, and the temperature, salinity, and hardness of the water all influence the toxicity to
aquatic invertebrates (Boening, 2000). Acute toxicity values (i.e., the lethal concentration for 50
percent of the population or LC50) of inorganic mercury compounds identified in recent literature
for freshwater invertebrates range from 2.2 |ig Hg2+/L for a cladoceran (Daphniapulex) to 2,000
\ig Hg2+/L for the larval forms of three insects (U.S.EPA, 1997), with organic mercury
compounds from 10 to 100 times more toxic than inorganic mercury (Boening, 2000). In marine
invertebrates, the gastrula stage was found to be the most sensitive period of development in
embryo toxicity tests (Bellas et al., 2001).
Bellas et al. (2001) examined the effects of mercuric chloride on sperm viability,
fertilization, embryogenesis, and larval attachment of the ascidian (Ciona intestinalis), a minute
sedentary marine invertebrate, at concentrations ranging from 8 to 256 \ig /L HgCl2. Larval
B-4
-------
attachment and embryogenesis were the most sensitive endpoints, showing effects at
concentrations of 16 and 64 u.g/L, respectively. No significant differences across concentrations
were observed in fertilization, and the authors noted that the effect of trace metals on sperm
viability is controversial. However, embryonic development was substantially affected and
decreased larval attachment was observed in the newly hatched larvae.
The toxicity of different forms of mercury to aquatic invertebrates depends in part on the
bioavailability of those forms to the animals. Mercury speciation and concentration affect
bioavailability; the organic content of the water and/or sediment also plays a role given the
tendency of inorganic and organic mercury compounds to form complexes with dissolved
organic matter and to sorb to organic particles. Divalent mercury and methylmercury
(CH3HgCl) are both particle reactive and have a strong affinity for organic matter (Bellas et al.,
2001). Because particulate organic matter is a food source for many species of benthic
invertebrates, it can serve as a source of mercury intake by benthic organisms. On the other
hand, organic matter tends to bind with mercury compounds, making them less bioavailable. In
experiments conducted with Leptocheirus plumulosus, Lawrence and Mason (2000) found that
mercury accumulation in the estuarine amphipod was reduced in sediments enriched in organic
matter. They also found that methylmercury was more readily available for uptake. Sjoblom et
al. (2000) showed that in freshwater, most dissolved inorganic mercury is bound to dissolved
organic matter. Dissolved humic substances in freshwater exert a strongly negative influence on
the bioavailability of both inorganic mercury and methylmercury.
In addition, the beneficial role of consuming increased amounts of the algae (Chlorella
vulgaris) was cited in one study. Ramirez-Perez et al. (2004) tested the age-specific responses to
mercuric chloride of a rotifer (Brachionus calyciflorus), an organism found between
phytoplankton and fish larvae in aquatic food chains, using two algal densities. With increasing
mercury concentrations (ranging up to 5 u.g/L HgCl2), the researchers observed an increasingly
negative effect on survivorship, reproduction, and lifespan of the rotifer, although less of an
impact was seen when the food level (i.e., algal density) was higher.
Recent studies in aquatic invertebrates have demonstrated the ability of mercury to
suppress the immune system. In marine bivalves, the internal immune system is based on the
ability of circulating cells known as hemocytes to fend off foreign bacteria and viruses using
phagocytic and microbicidal mechanisms (Cooper and Knowler, 1992; Fournier et al., 2001).
An in vivo study demonstrated that clams (Mya arenarid) accumulate methylmercury to a much
greater extent than HgCl2 and that methylmercury leads to greater immune suppression (Fournier
et al., 2001). Mercury exposure concentrations ranged from 10"9 to 10"5 M. No detrimental
effects were observed at concentrations below 10"6 M, but at higher concentrations, a clear
relationship between increasing mercury accumulation and decreasing phagocytic activity of
hemocytes was evident from 7 to 28 days after exposure. In an in vitro analysis of a small
aquatic worm (Tubifex tubifex), phagocytic responses were tested with methylmercury or
mercuric chloride at concentrations from 10"9 to 10"4 M (Sauve et al., 2002). Although the
authors noted that short-term in vitro assays are often not as realistic as in vivo tests, the results
are similar to Fournier et al. (2001) in that immunotoxicity, expressed as the concentration that
resulted in a 50 percent reduction in phagocytic activity relative to controls, was observed at
mercury levels around 10"6 M.
B-5
-------
B.4.3 Fish and Amphibian Species
Effects of methylmercury on fish include reduced reproduction, impaired growth and
development, behavioral abnormalities, altered blood chemistry, impaired osmoregulation and
immunity, reduced feeding rates and predatory success, effects on oxygen exchange, and death
(U.S.EPA, 1997). Bacterial methylation of inorganic mercury can occur either in sediment or in
bacteria associated with the fish gills or gut. Greater than 90% of the mercury content in
freshwater fish is in the methylmercury form (U.S.EPA, 1997). Symptoms of acute mercury
poisoning in fish include increased secretion of mucous, flaring of gill opercula, increased
respiration rate, loss of equilibrium, and sluggishness. When fish have been exposed to
concentrations of mercury considered to be sublethal (i.e., below 30 p.g/L), abnormalities ranging
from physiological to reproductive to biochemical have been reported (Boening, 2000). Signs of
chronic poisoning include emaciation, brain lesions, cataracts, inability to capture food,
abnormal motor coordination, and various erratic behaviors (e.g., altered feeding behavior)
(U.S.EPA, 1997).
Accumulation of mercury in fish depends on several factors. Although the highest
mercury concentrations in fish generally occur in the blood, spleen, kidney, and liver, and may
exceed those in muscle by 2 to 10 times, most of the mercury contained in a fish at any given
time is associated with muscle tissues due to their larger mass relative to that of other tissues
(U.S.EPA, 1997). Mercury concentrations in fish tissues tend to increase in both marine and
freshwater fish with increasing age and size. For example, Redmayne et al. (2000) found that
methylmercury concentrations in long-finned eels (Anguilla dieffenbachid) in New Zealand
increase with both eel age and length. Also, mercury accumulation in fish is higher in waters
with higher levels of dissolved organic carbon, which may assist mercury in entering the food
chain. And at lower pH levels in water, methylmercury is a higher fraction of the total mercury
in fish tissues than in waters at higher pH levels (Boening, 2000).
The toxicity of mercury to fish varies, depending on the fish's characteristics (e.g.,
species, life stage, age, and size), environmental factors (e.g., temperature, salinity, dissolved
oxygen content, hardness, and the presence of other chemicals), and the form of mercury
available. As with aquatic invertebrates, organomercury compounds, such as methylmercury,
generally are much more acutely toxic than inorganic mercury to fish (U.S.EPA, 1997).
Recent studies provide examples of the toxic effects of mercury in fish at both laboratory
and environmentally-relevant concentrations. For example, Houck and Cech (2004) exposed
juvenile blackfish (Orthodon microlepidotus) to varying levels of methylmercury (0.21, 0.52 ,
22.2, and 55.5 u.g/g) (from 0.21 to 5.5 p.g/g) over a 70-day period and found that the two highest
dose groups exhibited decreased growth rates. In walleye (Stizostedion vitreurri), Latif et al.
(2001) assessed methylmercury effects on embryonic and larval stages. They found that
increases in methylmercury concentrations at environmentally relevant concentrations in the
water (from 0.0001, .002, 0.003, and 0.008 u.g MeHg/L) were generally associated with linear
declines in the hatching success of eggs; however, no statistical analysis was conducted.
Furthermore, hatching success was not correlated with methylmercury concentrations in the
eggs. In the embryonic stage, a decrease in heart rate was observed with higher waterborne
methylmercury concentrations, but larval growth was not subsequently affected. In addition,
information on the toxicity of mercury to fish at environmentally-relevant levels is found in the
B-6
-------
British Columbia Ministry of Environment, Lands and Parks (2001) ambient water quality
criteria and guideline for mercury.
Recent experiments have demonstrated important sublethal effects of mercury toxicity
that might affect fish populations in the field. For example, Fjeld et al. (1998) found that
although morphological disturbances were observed in only the group exposed to the highest
level of methylmercury (i.e., 20 ng/L), there was long-term permanent impairment of feeding
behavior in all exposed grayling (Thymallus thymallus) embryos dosed during the first 10 days
of development to methylmercury at concentrations ranging from 0.8 to 20 \igfL. Berntssen et
al. (2003) demonstrated the relative susceptibility of the Atlantic salmon (Salmo salar) brain to
methylmercury tissue concentrations above a threshold of 10,000 (ig/kg, as compared to kidney
and liver tissues. There was no evidence of reduced growth in the salmon, but substantial
reductions in neural enzyme activity and alterations in feeding behavior were observed.
LeBlond and Hontela (1999) examined the effects of mercuric chloride and methylmercury on
cellular synthesis of the hormone cortisol using an in vitro assay with rainbow trout
(Oncorhynchus mykiss). Although the specific cellular site of action was not determined for the
two mercury species, both disrupted production of cortisol.
Samson et al. (2001) demonstrated effects ranging from delayed mortality syndrome to
physical (e.g., faint heartbeats) and behavioral abnormalities in embryonic zebrafish (Danio
rerio) exposed to methylmercury levels of 5, 10, and 15 ng/L for several different periods of
time. Prey capture ability was impaired in larvae exposed continuously to 10 ng/L, even after 4
days in clean water. Although morphological defects were not always evident, the authors
conclude that functional impairment is a more subtle response to developmental toxicants than
mortality or the production of morphological defects, which may not be sensitive enough
endpoints for determining safe levels of a toxicant in the environment. Sweet and Zelikoff
(2001) cite several in vivo and in vitro studies, with effect-inducing doses of 30 to 350 |ig/L and
0.3 to 10 \iM, respectively, that have demonstrated immunotoxic effects in fish due to mercury
exposure ranging from depressed blood cell production and enzyme activity to enhanced cell
death.
It is generally thought that toxic effects are unlikely to occur in fish in the environment,
where the concentrations of mercury in surface waters are much lower than those in many of the
laboratory experiments conducted. For example, Friedmann et al. (1996) examined 14 northern
pike (Esox luclus) from Lake Champlain for reproductive status and mercury concentrations in
tissues. This species tends to accumulate more mercury than many others because northern pike
are top-level predators. Despite the fact that the average muscle concentration was 325 [jig
Hg/kg wet-weight, higher than the national average of 100 \ig Hg/kg, no correlation was
observed between mercury content, gonadosomatic index, and gonadal sex steroid levels.
However, evidence on more subtle effects, including those discussed in the previous paragraph,
shows that effects on behavior, reproduction, and development can occur at relatively low
mercury concentrations in water (U.S.EPA, 1997). Many of the earlier toxicity studies focused
on endpoints related to growth, reproduction, and mortality. More recent studies are
demonstrating more subtle effects, including effects on neural and immunological endpoints.
Amphibians are similar to fish in sensitivity to mercury. Acute toxicity (i.e., LC50) values
for a variety of embryo-larval stage amphibians exposed to inorganic mercury compounds range
B-7
-------
from 1.3 to 107.5 ng Hg/L, which are similar to values found for fish (Boening, 2000). The
toxicity of mercuric chloride to tadpoles ranges from approximately 50 to 760 |ig Hg/L for the
frog (Rana hexadactyld), clawed toad (Xenopus laevis), and toad (Bufo melanstictus) (Boening,
2000).
B. 4.4 Terrestrial Plant Species
The effect of mercury pollution on terrestrial plants has not been studied as extensively as
aquatic plants. Terrestrial plants typically acquire mercury through contaminated soil. The
accumulation of mercury in terrestrial plants increases with increasing soil mercury
concentration (Boening, 2000). Methylmercury is more toxic to terrestrial plants than Hg2+ (U.S.
EPA, 1997). Wild plant communities located in areas with soil contaminated with a sufficient
amount of mercury atmospheric deposition may be at risk for exhibiting mercury-induced
chronic effects. It is probable that subtle disturbances to a community occur at lower
concentrations (chronic exposures) of mercury than those suggested in the literature (based
largely on acute exposures) (Boening, 2000). Mosses have shown the ability to acquire mercury
directly through atmospheric deposition (Boening, 2000).
Terrestrial plants accumulate mercury primarily in their root structures (Greger et al.,
2005). In a recent laboratory study, the soil of six plant species was dosed with a solution
containing 200 u.g/L HgCl2. The plants included white clover (Trifolium repens), spring wheat
(Triticum aestivum), sugar beet (Beta vulgaris), oil-seed rape (Brassica napus), willow (Salix
viminalis), and garden pea (Pisum sativuni). All of the examined plant species were able to take
up mercury in the root. However, transport of mercury from the root to the shoot was low (0.17
- 2.5%). Mercury that reached the leaves was sequestered and not released into the ambient air
through transpiration. These findings indicate that the studied plant species may serve as a
reservoir of mercury; however, they are not capable of remobilizing terrestrial soil mercury
deposits into the atmosphere. Such studies may eventually lead to development of plants to
assist phytoremediation of mercury contaminated sites.
B.4.5 Terrestrial Invertebrate Species
The available information regarding the toxicity of mercury on terrestrial invertebrate
species is limited, and the topic needs further study. Terrestrial invertebrates contribute to the
diet of numerous species including birds and mammals. Some terrestrial species are known to
transform Hg2+ to methylmercury and back (Jereb et al., 2003). A better understanding of the
toxicity of mercury on terrestrial invertebrates would help to characterize whether, and if so,
how the health of lower trophic level species affects the overall health of the ecosystem.
The phagocytic immune response of terrestrial invertebrate worms was assessed
following in vitro exposure to either mercuric chloride or methylmercury (Sauve et al., 2002).
The authors dosed three earthworm species (Lumbricus terrestris, Eiseniafetida, Aporrectodea
turgidd) with methylmercury or mercuric chloride at concentrations from 10"9 to 10"4 M. Both
HgCl2 and CH3HgCl inhibited the phagocytic immune response to challenge with carboxylate
microsphere beads. The concentration necessary to reduce the phagocytic response by 20%
varied across species from approximately 10"8 to 10"6 M for methylmercury and from 10"6 to 10"7
M for HgCl2. Inhibition of the phagocytic response could potentially lead to increased infection
B-8
-------
by opportunistic pathogens such as viruses and/or bacteria. Species-specific considerations are
clearly a factor when determining mercury immunotoxicity in invertebrate worm species.
Sauve and Fournier (2005) continued the examination of the phagocytic immune
response to methylmercury exposure in the earthworm (Eisenia andrei) and found age-specific
differences. Four age groups were studied in both in vitro and in vivo assays. The youngest
hatchling group exhibited lower in vitro phagocytic activity than that of adults. Neither adults,
nor hatchlings showed a significant decrease in phagocytosis at methylmercury concentrations
up to 10"7 M in invertebrate worms. Thus, hatchlings do not show a higher sensitivity to mercury
exposure, but they have less capacity to respond to immune challenges requiring a phagocytic
response.
A laboratory-based in vivo immune assay was employed in which Eisenia andrei worms
were exposed to varying methylmercury concentrations for 5 days prior to evaluation of
phagocytic response (Sauve" and Fournier, 2005). Adult worms exposed to filter paper dosed
with concentrations of methylmercury (1-2 |j.g/cm2 MeHg) showed an immune response
ranging up to 300% greater than the control. This could potentially mean that adult worms
exposed to sublethal methylmercury may become more effective in dealing with immune
challenges by opportunistic pathogens; however, the long term effects of sublethal exposures to
methylmercury were not examined in this study of invertebrate worms. The same effect was not
observed in hatchling worms, and possible effects of elevated immune function over an extended
period of time are not known.
B.4.6 Avian Species
In exposed birds, the liver and kidney are typically the sites of highest mercury levels
(Boening, 2000). Sublethal effects of mercury on birds include neurobehavioral effects, reduced
food consumption, liver and kidney damage, spinal cord damage, reduced cardiovascular
function, impaired immunity, reduced muscular coordination, impaired growth and development,
altered blood chemistry, and reproductive effects (U.S.EPA, 1997). Bird species occupying top
predator roles in aquatic ecosystems frequently exhibit elevated levels of mercury. For example,
mercury has been detected in adult spectacled (Somateriafischeri) eiders in northern Alaska and
in German Northern Goshawks (Accipiter gentiles) (Kenntner et al., 2003; Wilson et al., 2004).
In non-pisciviorous bird species inhabiting strictly terrestrial ecosystems, the mercury body-
burden are typically lower (Boening, 2000). Nevertheless, terrestrial seed-eating bird species
may be exposed through the application of methylmercury containing fungicides during
agricultural practice (Boening, 2000).
The common loon (Gavia immer) has been used as a potential indicator of
methylmercury contamination in lake ecosystems across the northern United States and Canada
(Meyer et al., 1998; Evers et al., 2003). In a long term field study of the adverse effects of
mercury on the loon, eggs were collected across the northern United States between 1995 and
2001 (Evers et al., 2003). Eggs collected in the same geographic territory showed adult female
blood mercury concentrations that were highly correlated with blood mercury concentrations in
eggs. In the New England region, egg volume significantly decreased with increasing mercury
concentration; however, the authors did not find a significant relationship between egg mercury
concentration and reproductive success. The average mercury concentration found in infertile
B-9
-------
eggs (0.78 + 0.5 |J.g/g, n = 201) was not significantly different from the average mercury
concentration (0.74 + 0.54 \ig/g, n = 205) of the fertile eggs. A previous study found chick
production to be lower at lakes where chick blood mercury concentrations were elevated;
however, decreased chick production was not associated with adult mercury exposures in the
study (Meyer et al., 1998).
A dose-response laboratory study of common loons was conducted to investigate the
adverse effects of mercury exposure on chick development (Kenow et al., 2003). Eggs were
collected in a four county region of northern Wisconsin during the summers of 1999 and 2000
and dosed from hatch to age 105 days with varying levels of methylmercury contaminated
rainbow trout. While the methylmercury administered was found to contain 20% ethyl mercury,
no effect on food consumption, growth in body mass or body length was measured in any of the
exposure groups following dosing. Further, no signs of neurotoxic effects, such as behavioral
abnormalities or loss of muscle coordination, were observed. The authors speculated that rapid
excretion of methylmercury during feather growth offered some protection against adverse
effects during the first 105 days of loon development.
A field study conducted in Hong Kong, China examined the breeding success of two
Ardeid bird species exposed to metals (Connell et al., 2002). Ardeidae are fish-eating birds that
include species such as herons, egrets and bitterns (De Luca-Abbott et al., 2001). The feathers of
the Little Egret (Egretta garzettd) and the Black-crowned Night Heron (Nycticorax nycticorax)
were analyzed for concentrations of copper, iron, manganese, zinc, lead, cadmium, chromium,
and mercury. An examination of the possible adverse effects of metals exposure included a
probabilistic assessment of breeding success. The authors concluded that mercury (0.5 - 7.1
p.g/g dry wt feathers) increased the likelihood of adverse effects on the breeding success of the
Little Egret at one of the six sites monitored. At this site, a maximum Risk Quotient (RQ) of
1.37 was calculated based on a 3.0 yg/g dry wt feathers No Observed Adverse Effect Level
(NOAEL) and an average measured egret feather mercury concentration of 4.1 ng/g. The 3.0
Hg/g NOAEL was derived from the available scientific literature (Burger and Gochfeld, 1997;
Connell et al., 2002). The same analysis did not find any evidence of mercury effects on the
breeding success of the Black-crowned Night Heron.
Burger and Gochfeld (1997) attempted to relate adverse effects of mercury exposures
observed in the laboratory to field biomonitoring observations in birds. The authors identify
laboratory studies that indicate exposures as low as 1.5 ppm in eggs and/or 5 to 40 ppm in
feathers are associated with adverse effects such as impaired reproduction (Burger and Gochfeld,
1997). Egg mercury concentrations as low as 0.5 - 6.0 ppm wet weight are capable of causing
decreased egg weight, embryo malformations, lowered hatchability, decreased chick growth and
lowered chick survival (Burger and Gochfeld, 1997).
An examination of the total mercury (ppm, dry weight) in bird eggs from the New York
City region show levels of mercury which exceed the 0.5 ppm adverse effect level identified by
the authors for a number of species; however, not every studied location found the same result
(Burger and Gochfeld, 1997). At risk species included: Snowy Egret (Egretta thuld), Black
Skimmer (Rynchops niger), Common Tern (Sterna hirundo), Foraster's Tern (Sterna forsteri),
Roseate Tern (Sterna dougallii), and Herring Gull (Lams argentatus). The feathers of these
same species were also analyzed for total mercury and were found to be within the range
B-10
-------
identified in the scientific literature as being associated with reduced hatchability of eggs,
behavioral abnormalities of adults, and infertility (Burger and Gochfeld, 1997).
Work by Way land et al. (2002, 2003) examined the health of the northern common eider
(Somateria millisima borealis). The common eider is a sea duck that inhabits coastal areas in
Canada, Greenland as well as other Arctic territories. In a 1999 field analysis, measured liver
mercury concentrations ranging from 1.3 to 6.5 u.g/g dry weight were negatively correlated to
abdominal fat mass, spleen mass and body mass in males at the time of capture (Wayland et al.,
2002). Further study of this population in 2000 revealed that mercury concentrations were also
negatively correlated with heart mass and body mass at the time of dissection (Wayland et al.,
2003). Immune endpoints were also examined and no statistically significant relationship was
found between mercury concentration and swelling response to an injection of the antigen,
phytohemagglutinin-P.
A laboratory-based egg study in eighty pairs of mallard ducks (Anas platyrhynchos)
exposed to 0, 5, 10 or 20 ng/g methylmercury found neurological effects in exposed offspring
(Heinz and Hoffman, 2003). Females laid 15 eggs while unexposed and another 15 during the
exposure period. Following the exposure period, another 30 eggs were laid and examined for
mercury content. Even-numbered eggs were incubated and allowed to hatch while odd-
numbered eggs were saved for mercury analysis. Mercury in the even-numbered eggs was
estimated by averaging mercury content in the neighboring odd-numbered eggs. Neurological
signs of mercury poisoning included loss of coordination and staggered gait. These altered
behaviors were observed in ducklings hatching from eggs containing 2.3 ng/g estimated mercury
on a wet-weight basis. Developmental deformities were also observed in eggs containing as
little as 1 u.g/g estimated mercury. The authors did not conduct a statistical analysis of the data;
however, they conclude that methylmercury concentrations in excess of 2 u.g/g on a wet-weight
basis will harm the neurological development of sensitive mallard embryos.
B.4.7 Mammalian Species
The effects of mercury on mammalian wildlife are similar to those found in humans, with
the primary target being the central nervous system. Most mammalian studies have been ,
conducted in laboratories. Relatively few studies of mammalian populations in the wild have
been published (Boening, 2000). Mammals drawing all or a portion of their dietary intake from
aquatic ecosystems will likely have greater exposure than those that do not. Thus, aquatic
bioaccumulation is a factor in determining mammalian exposures. Some bioaccumulation may
also occur in the terrestrial setting. Deer mice (Peromyscus maniculatus) sampled in Isle Royale
National Park in the state of Michigan have liver mercury concentrations that may pose an
exposure risk to higher trophic level predators such as the red fox (Vulpes vulpes) (Vucetich et
al., 2001).
Extensive laboratory studies of monkeys (Macaco fascicularis) have been conducted in
an attempt to elucidate the toxicity of mercury in humans. Sensory system impairment was
observed in a cohort of monkeys dosed in utero with methylmercury through age four (Rice and
Gilbert, 1990; Rice and Gilbert, 1995; Rice, 1998; Rice and Hayward, 1999). Exposure
symptoms included impaired hearing, visual function, and ability to detect vibration. Doses of
10 or 25 u.g/kg/day resulted in evidence of delayed neurotoxicity as well as impairment of
B-ll
-------
auditory function (Rice, 1998). These same mechanisms of toxicity described in the laboratory
setting may also be present in the wild. However, the diet of the M. fascicularis monkey in the
wild has not been extensively studied. The wild diet of M. fascicularis is speculated to consist
primarily of fruit but they are believed to be opportunistic omnivores willing to consume
terrestrial invertebrates and bird eggs (Kemp and Burnett, 2003). Regardless, estimates of
mercury exposure in the wild are difficult to make, but mercury exposure is of less concern for
primarily herbivorous mammals.
Concern has recently grown that mercury exposure may be a contributing factor to the
decline in the endangered Florida Panther population (Puma concolor coryt) (Barren et al.,
2004). Barron et al. (2004) performed a probabilistic risk assessment of retrospective and
current mercury exposure using a dietary model that incorporated the variability and uncertainty
in ingestion rate, diet, body weight, and mercury exposure of panthers. Under the worst-case
modeling conditions, the current risk of panthers developing clinical symptoms that may lead to
death was 4.6%. Thus, there was a 4.6% chance that any given panther would receive a lethal
mercury dose under the worst case exposure scenario. The authors concluded that past mercury
exposures likely did adversely affect panthers in the Florida Everglades, but current estimated
risks are significantly lower than past risks because of an estimated 70-90% decline in mercury
exposure over the past decade (Barron et al., 2004).
Mink (Mustela vison) are an example of a species acquiring a portion of their diet
foraging in aquatic ecosystems. Fish compose about 25% of the mink diet (Ferreras and
Macdonald, 1999; Yamaguchi et al., 2003). Yamaguchi et al. (2003) calculated Risk Quotients
for mink at four locations along the Thames River after sampling for mercury contamination in
perch (Percafluviatilis), roach (Rutilus rutilus), dace (Leuciscus leuciscus), eel (Anguilla
anguilld), and pike (Esox lucius). Not all species were available at each sampling site. An RQ
greater than one indicates that the concentration of mercury in fish is likely greater than the No
Observable Adverse Effects Concentration (NOAEC) threshold (Giesy et al., 1994; Henry et al.,
1998; Yamaguchi et al., 2003) and thus poses a risk to the exposed species at the time of
assessment. The calculated RQ values for each species consumed by the mink at each of the
four locations ranged from less than one to near 8. The authors speculate that the RQ of mercury
for mink may be closer to 1 since the fish sampled in the study were close to the upper limits of
the prey size typically selected by mink.
B.5 Ecosystems Potentially Affected
Ecosystems that could be affected by mercury exposure include those that already have
high mercury levels, particularly those with top carnivore populations with high mercury loads;
ecosystems with long aquatic food chains and piscivorous wildlife ; and ecosystems with soils or
sediments low in organic content and high in minerals - promoting more soluble and
bioavailable forms of mercury. Mercury levels in all of these ecosystems are likely declining as
a result of recent regulations, but the quantitative effect on the ecosystems is unclear because the
decline is slow, depending on sediment burial as a primary mechanism.
More pristine aquatic ecosystems tend to have longer food chains than eutrophic systems.
Assuming equal inputs of mercury into the ecosystem from atmospheric deposition,
biomagnification through multiple trophic levels may thus result in exposures to top carnivores
B-12
-------
in pristine systems that are comparable to or higher than corresponding exposures in eutrophic
systems. In mixing zones (e.g., river entering lake, estuaries), the higher levels of suspended
sediments with sorbed mercury compounds correlate with higher rates of bioaccumulation of
mercury in zooplankton, and presumably the rest of the food chain.
B.6 Conclusions
A quantitative analysis of the ecological benefits of reduced mercury emissions is not
possible at this time given the current state of the science. Recent research on the ecological
effects of mercury exposures summarized in this appendix does provide qualitative support to the
notion that reductions in mercury emissions from various sources could lead to improvements in
overall ecosystem health. The bulk of this research, based on both laboratory and field studies,
suggests that because mercury is persistent in the environment and biomagnifies up the food
chain when methylated, a wide variety of species and ecosystems may be harmed by excessive
levels of mercury in the environment.
To some degree, mercury contamination is present in virtually all environmental media,
but aquatic systems appear to experience the greatest exposures due to higher rates of
biomagnification possible in those systems. Elimination of methylmercury from fish is so slow
that long-term reductions of mercury concentrations in fish are often due to growth of the fish
("growth dilution"), whereas other mercury compounds are eliminated relatively quickly.
Piscivorous avian and mammalian wildlife are exposed to mercury mainly through the
consumption of contaminated fish and, as a result, bioaccumulate mercury to levels greater than
those in prey items (U.S.EPA, 1997).
Numerous studies have generated field data on the levels of mercury in a variety of wild
species. The body of work examining the effects of these exposures, particularly in real world
settings, is growing, but our understanding of the consequences is still incomplete. Much of the
research conducted to date has been carried out in laboratory settings rather than in the wild; so
EPA believes reliable conclusions about overall ecosystem health cannot be made at this time.
Nevertheless, numerous adverse effects have been identified at environmentally relevant doses
as well as at doses slightly above environmental concentrations. Although the magnitude of the
power plant contribution to ecological exposures cannot be quantified so the corresponding risk
for adverse effect cannot be determined, reducing the presence of mercury in the environment
should reduce the potential for adverse ecological impacts.
B.7 References
Barren, M. G., S. E. Duvall, et al. (2004). Retrospective and current risks of mercury to panthers
in the Florida Everglades. Ecotoxicology 13(3): 223-9.
Bellas, J., E. Vazquez, et al. (2001). Toxicity of Hg, Cu, Cd, and Cr on early developmental
stages of Ciona intestinalis (Chordata, Ascidiacea) with potential application in marine
water quality assessment. Water Res 35(12): 2905-12.
B-13
-------
Berntssen, M. H., A. Aatland, et al. (2003). Chronic dietary mercury exposure causes oxidative
stress, brain lesions, and altered behaviour in Atlantic salmon (Salmo salar) parr. Aquat
Toxicol 65(1): 55-72.
Boening, D. W. (2000). Ecological effects, transport, and fate of mercury: a general review.
Chemosphere 40(12): 1335-51.
British Columbia Ministry of Environment, L. a. P. (2001). Ambient water quality guidelines for
mercury: overview report - first update.
http://wlapwww.gov.bc.ca/wat/wq/BCguidelines/mercury.html. E. a. R. D. Water
Management Branch.
Burger, J. and M. Gochfeld (1997). Risk, mercury levels, and birds: relating adverse laboratory
effects to field biomonitoring. Environ Res 75(2): 160-72.
Connell, D. W., B. S. Wong, et al. (2002). Risk to breeding success of Ardeids by contaminants
in Hong Kong: evidence from trace metals in feathers. Ecotoxicology 11(1): 49-59.
Cooper, J. E. and C. Knowler (1992). Investigations into causes of death of endangered molluscs
(Partula species). Vet Rec 131(15): 342-4.
De Luca-Abbott, S. B., B. S. Wong, et al. (2001). Review of effects of water pollution on the
breeding success of waterbirds, with particular reference to ardeids in Hong Kong.
Ecotoxicology 10(6): 327-49.
Evers, D. C., K. M. Taylor, et al. (2003). Common loon eggs as indicators of methylmercury
availability in North America. Ecotoxicology 12(1-4): 69-81.
Ferreras, P. and D. W. Macdonald (1999). The impact of American Mink (Mustela vison) on
water birds in the upper Thames. Journal of Applied Ecology 36(5): 701-709.
Fjeld, E., T. O. Haugen, et al. (1998). Permanent impairment in the feeding behavior of grayling
(Thymallus thymallus) exposed to methylmercury during embryogenesis. Sci Total
Environ 213(1-3): 247-54.Fournier, M., J. Pellerin, et al. (2001). Effects of in vivo
exposure of Mya arenaria to organic and inorganic mercury on phagocytic activity of
hemocytes. Toxicology 161(3): 201-11.
Friedmann, A. S., M. C. Watzin, et al. (1996). Effects of environmental mercury on gonadal
function in Lake Champlain northern pike (Esox lucius). Bull Environ Contain Toxicol
56(3): 486-92.
Giesy, J. P., D. A. Verbrugge, et al. (1994). Contaminants in fishes from Great Lakes-influenced
sections and above dams of three Michigan rivers. I: Concentrations of organo chlorine
insecticides, polychlorinated biphenyls, dioxin equivalents, and mercury. Arch Environ
Contam Toxicol 27(2): 202-12.
B-14
-------
Gilmour, C. C. and E. A. Henry (1991). Mercury methylation in aquatic systems affected by acid
deposition. Environ Pollut 71(2-4): 131-69.
Greger, M., Y. Wang, et al. (2005). Absence of Hg transpiration by shoot after Hg uptake by
roots of six terrestrial plant species. Environ Pollut 134(2): 201-8.
Heinz, G. H. and D. J. Hoffman (2003). Embryotoxic thresholds of mercury: estimates from
individual mallard eggs. Arch Environ Contain Toxicol 44(2): 257-64.
Henry, K. S., K. Kannan, et al. (1998). Concentrations and hazard assessment of organochlorine
contaminants and mercury in smallmouth bass from a remote lake in the Upper Peninsula
of Michigan. Arch Environ Contam Toxicol 34(1): 81-6.
Houck, A. and J. J. Cech, Jr. (2004). Effects of dietary methylmercury on juvenile Sacramento
blackfish bioenergetics. Aquat Toxicol 69(2): 107-23.
Jereb, V., M. Horvat, et al. (2003). Transformations of mercury in the terrestrial isopod Porcellio
scaber (Crustacea). Sci Total Environ 304(1-3): 269-84.
Kemp, N. J. and J. B. Burnett (2003). A biodiversity risk assessment and recommendations for
risk management of Long-tailed Macaques (Macaca fascicularis) in New Guinea.
www.indopacific.org/papuamacaques.pdf, Indo-Pacific Conservation Alliance.
Kenntner, N., O. Krone, et al. (2003). Environmental contaminants in liver and kidney of free-
ranging northern goshawks (Accipiter gentilis) from three regions of Germany. Arch
Environ Contam Toxicol 45(1): 128-35.
Kenow, K. P., S. Gutreuter, et al. (2003). Effects of methyl mercury exposure on the growth of
juvenile common loons. Ecotoxicology 12(1-4): 171-82.
Latif, M. A., R. A. Bodaly, et al. (2001). Effects of environmental and maternally derived
methylmercury on the embryonic and larval stages of walleye (Stizostedion vitreum).
Environ Pollut 111(1): 139-48.
Lawrence, A. L. and R. P. Mason (2001). Factors controlling the bioaccumulation of mercury
and methylmercury by the estuarine amphipod Leptocheirus plumulosus. Environ Pollut
111(2): 217-31.
Leblond, V. S. and A. Hontela (1999). Effects of in vitro exposures to cadmium, mercury, zinc,
and l-(2-chlorophenyl)-l-(4-chlorophenyl)-2,2-dichloroethane on steroidogenesis by
dispersed interrenal cells of rainbow trout (Oncorhynchus mykiss). Toxicol Appl
Pharmacol 157(1): 16-22.
Meyer, M. W., D. C. Evers, et al. (1998). Patterns of Common Loon (Gavia immer) Mercury
Exposure, Reproduction, and Survival in Wisconsin, USA. Environ Toxicol Chem 17(2):
184-190.
B-15
-------
Ramirez-Perez, T., S. S. Sarma, et al. (2004). Effects of mercury on the life table demography of
the rotifer Brachionus calyciflorus Pallas (Rotifera). Ecotoxicology 13(6): 535-44.
Redmayne, A. C., J. P. Kim, et al. (2000). Methyl mercury bioaccumulation in long-finned eels,
Anguilla dieffenbachii, from three rivers in Otago, New Zealand. Sci Total Environ
262(1-2): 37-47.
Rice, D. C. (1998). Age-related increase in auditory impairment in monkeys exposed in utero
plus postnatally to methylmercury. Toxicol Sci 44(2): 191-6.
Rice, D. C. and S. G. Gilbert (1990). Effects of developmental exposure to methyl mercury on
spatial and temporal visual function in monkeys. Toxicol Appl Pharmacol 102(1): 151-
63.
Rice, D. C. and S. G. Gilbert (1995). Effects of developmental methylmercury exposure or
lifetime lead exposure on vibration sensitivity function in monkeys. Toxicol Appl
Pharmacol 134(1): 161-9.
Rice, D. C. and S. Hayward (1999). Comparison of visual function at adulthood and during
aging in monkeys exposed to lead or methylmercury. Neurotoxicology 20(5): 767-84.
Samson, J. C., R. Goodridge, et al. (2001). Delayed effects of embryonic exposure of zebrafish
(Danio rerio) to methylmercury (MeHg). Aquat Toxicol 51(4): 369-76.
Sauve, S. and M. Fournier (2005). Age-specific immunocompetence of the earthworm Eisenia
andrei: exposure to methylmercury chloride. EcotoxicolEnviron Saf60(l): 67-72.
Sauve, S., M. Hendawi, et al. (2002). Phagocytic response of terrestrial and aquatic invertebrates
following in vitro exposure to trace elements. Ecotoxicol Environ Saf52(l): 21-9.
Schuster, P. F., D. P. Krabbenhoft, et al. (2002). Atmosphere mercury deposition during the last
270 years: a glacial ice core record of natural and anthropogenic sources. Environ Sci
Technol 36(11): 2303-10.
Sjoblom, A., M. Meili, et al. (2000). The influence of humic substances on the speciation and
bioavailability of dissolved mercury and methylmercury, measured as uptake by
Chaoborus larvae and loss by volatilization. Sci Total Environ 261(1-3): 115-24.
Sweet, L. I. and J. T. Zelikoff (2001). Toxicology and immunotoxicology of mercury: a
comparative review in fish and humans. J Toxicol Environ Health B Crit Rev 4(2): 161-
205.
U.S.EPA (1997). Mercury Study Report to Congress. Vol. 6: An Ecological Assessment for
Anthropogenic Mercury Emissions in the United States. USEPA-452/R-97-008.
B-16
-------
Vucetich, L. M., J. A. Vucetich, et al. (2001). Mercury concentrations in deer mouse
(Peromyscus maniculatus) tissues from Isle Royale National Park. Environ Pollut 114(1):
113-8.
Wayland, M., H. G. Gilchrist, et al. (2002). Immune function, stress response, and body
condition in arctic-breeding common eiders in relation to cadmium, mercury, and
selenium concentrations. Environ Res 90(1): 47-60.
Wayland, M., J. E. Smits, et al. (2003). Biomarker responses in nesting, common eiders in the
Canadian arctic in relation to tissue cadmium, mercury and selenium concentrations.
Ecotoxicology 12(1-4): 225-37.
Wilson, H. M., M. R. Petersen, et al. (2004). Concentrations of metals and trace elements in
blood of spectacled and king eiders in northern Alaska, USA. Environ Toxicol Chem
23(2): 408-14.
Yamaguchi, N., D. Gazzard, et al. (2003). Concentrations and hazard assessment of PCBs,
organochlorine pesticides and mercury in fish species from the Upper Thames: river
pollution and its potential effects on top predators. Chemosphere 50(3): 265-73.
B-17
-------
APPENDIX C CARDIOVASCULAR EFFECTS AND METHYLMERCURY C-l
C.I Introduction C-l
C.2 Acute Myocardial Infarctions and Major Cardiovascular Effects C-l
C.2.1 The Kuopio Ischemic Heart Disease Risk Factor Study (KIHD) Cohort
C-2
C.2.2 The European Multicenter Case Control Study on Antioxidants,
Myocardial Infarction and Cancer of the Breast (EURAMIC) Cohort
C-4
C.2.3 Mechanisms for Cardiovascular Impacts C-4
C.2.4 Other Studies Evaluating CVD and Mercury Levels C-5
C.3 Other Cardiovascular Effects C-l
C.4 Cardiovascular Health Benefits of Fish Consumption C-8
C.5 Conclusions C-9
C.6 References C-10
-------
APPENDIX C
CARDIOVASCULAR EFFECTS AND METHYLMERCURY
C.I Introduction
Some recent epidemiological studies suggest that methylmercury may be a risk factor for
myocardial events, such as acute myocardial infarction (AMI), coronary heart disease (CHD),
cardiovascular disease (CVD) or other adverse cardiovascular effects such as carotid
atherosclerosis, increased blood pressure, or decreased heart rate variability. Other recent
studies did not observe a relationship between methylmercury levels and myocardial events.1
This appendix presents a qualitative discussion of recent studies that have looked at the
relationship between potential for cardiovascular impacts from chronic low-dose methylmercury
exposures. The results of several key peer-reviewed studies are summarized and study
uncertainties and other relevant information are also discussed. This section also includes a
discussion of the beneficial effects offish consumption.
The potential for adverse cardiovascular effects due to consumption offish containing
methylmercury is of particular interest given the evidence for the protective cardiovascular effect
believed to occur from an increased dietary fish intake. Strong evidence indicates that
consumption offish, particularly fatty fish, has a cardio-protective effect (Wang et al. 2004;
2005 Dietary Guidelines Advisory Committee 2004; NRC 2000). The presence of omega-3 («-
3) fatty acids in fish oils is hypothesized to drive the preventive effect on cardiovascular .disease
(Calder 2003). Several mechanisms of action are recognized, including the stabilization of the
atherosclerotic plaque (which, when ruptured, may cause a heart attack) by reducing the
infiltration of inflammatory and immune cells (lymphocytes and macrophages) into the plaque.
However, those studies relevant to the general population show cardiovascular health benefits
were for fish in the diet, not for isolated omega-3 fatty acids, such as in fish oil supplements,
suggesting the potential for synergistic benefits. Thus, consumption offish containing
methylmercury is not necessarily detrimental even though some evidence suggests that the
cardiovascular system may be a target system for methylmercury exposure. The cardio-
protective effect offish consumption has not been observed in all studies (Curb and Reed 1985;
Morris et al. 1992; Folsom and Demissie 2004).
C.2 Acute Myocardial Infarctions and Major Cardiovascular Effects
Salonen et al. (1995), Virtanen et al. (2005), and Guallar et al. (2002) have reported an
association between increased risk of AMI (or other major cardiovascular impacts) and
exposures to methylmercury via fish consumption or mercury levels in the body. The findings of
these studies are summarized in this section.
1 As described in detail by NRC (2000) and elsewhere, consumption fish containing methylmercury is the primary
route of exposure to methylmercury.
C-l
-------
C.2.1 The Kuopio Ischemic Heart Disease Risk Factor Study (KIND) Cohort
Salonen et al. (1995) investigated the association between methylmercury and AMI in a
study of a subset of men from the KIHD study group (n = 1,833 Finnish males age 42 to 60
years). Rissanen et al. (2000) in a follow up to this study extended observation of the same
cohort and Virtanen (2005) also utilized this cohort. In theory, the findings could be specific
only for men in Eastern Finland, who traditionally have a high intake of meat, fish and saturated
animal fat and a low intake of selenium and vitamin C and, most likely, other vegetable-derived
antioxidants. However these studies provide new information about potential mechanisms in
how methylmercury interacts with the cardiovascular system in humans, even though the
consequences of methylmercury intake for the cardiovascular system may vary among
populations due to different confounding factors (Salonen et al. 1995).
In the Salonen et al. (1995) study, all subjects' mercury levels were evaluated by hair
analysis, and subjects were grouped into three exposure groups according to hair mercury levels.
Urinalysis was also conducted for a subset of men who either had an AMI during the follow-up
or acted as the controls to that subset. Fish consumption was measured through an interview-
checked four-day food recording. Daily fish intake ranged from 0 to 619.2 g/day, and hair
concentrations ranged from 0 to 15.67 ppm (u.g/g). Stern (2005) observes that the estimated
mean dietary methylmercury intake was 7.6 ug/day, which is only somewhat larger than the
intake corresponding to the EPA RfD for a 70-kg (average) man of 7.0 ug/day (the
corresponding mean hair mercury concentration was 1.92 p,g/g (ppm) (Stern 2005). For the
subset of men who had an AMI, two control subjects were matched to each patient according to
age, municipality of residence, and date of baseline examination. Occurrence of AMI and deaths
due to CHD and CVD were recorded over the course of a seven-year period. In the analysis of
the results, Salonen et al. found that "the hair mercury (r = .27) and the urinary mercury (r = .47)
correlated with the estimated fish intake." Men with the highest hair mercury content (^ 2.0
Hg/g and ranging up to 15.67 jig/g - Stern (2005) estimates that ^ 2.0 p.g/g is likely equivalent to
the 90th percentile in U.S. men) had twice the risk of AMI when compared to men in the two
lowest exposure groups when adjusting for age, examination year, ischemic exercise,
electrocardiogram (ECG), and maximal oxygen uptake (relative risk [RR] 2.0; 95% confidence
interval [CI] 1.2-3.1). The risk of AMI decreased slightly for men within the highest hair
mercury category (^ 2.0 ng/g) after adjusting for all confounders and risk factors2 but was still
statistically significant (RR = 1.7; 95% CI 1.03-2.8). The relative risk was similar for coronary
deaths but not statistically significant due to the smaller number of events. However, men in this
group also had elevated risks of cardiovascular death (RR = 2.9; 95% CI 1.2-6.6) and death by
all causes (RR 2.3; 95% CI 1.4-3.6) after adjusting for all confounders and risk factors. Urinary
mercury levels were also significantly (and independently) associated with risk of AMI after
adjusting for the strongest risk factors (for each |ig mercury excreted daily, the risk of AMI
increased by 36%; 95% CI 1% to 82%). Based on these results, the study authors conclude that,
"although consumption offish may be healthy in general, some fish may contain agents that are
2 Risk factors included age, examination year, ischemic exercise ECG, maximal oxygen uptake, family history of
CHD, cigarette-years, mean systolic blood pressure, diabetes, socioeconomic status, place of residence (urban vs.
rural), dietary iron intake, and serum apolipoprotein B, HDL2 cholesterol, and ferritin concentrations.
C-2
-------
not healthy for the human cardiovascular system." Further, the authors suggest that mercury is a
risk factor for coronary and fatal CVD.
Two relevant follow-up studies have been conducted on the men in the KIHD population.
Rissanen et al. (2000) examined the interaction of mercury and the serum n-3 end-product fatty
acids docosahexaenoic acid (DHA) and docosapentaenoic acid (DPA). For this analysis, the
men in the KIHD cohort study were divided into quintiles based on their level of serum fatty
acids. Risk of acute coronary events was examined within the study cohort as a function of »-3
fatty acid levels. After adjusting for other risk factors,3 men in the highest fifth (quintile) of n-3
fatty acid level exhibited a 44 percent reduced risk of acute coronary events (95% CI11% to
65%) when compared to the lowest quintile. When the data were stratified by hair mercury
concentration into those with hair mercury content above and below 2 ng/g, individuals with
lower hair mercury who were also in the upper quintile of n-3 fatty acid level had a 67 percent
reduced risk of acute coronary events (95% CI 19% to 87%) compared with men with higher
hair mercury who were also in the upper quintile of n-3 fatty acid level. In each quintile,
subjects with the higher hair mercury concentrations had a higher risk of acute coronary events,
suggesting that the cardio-protective effects of the serum fatty acids was attenuated by the
mercury. Based on these results, Rissanen et al. suggest that their data "provide for the concept
that fish-oil derived fatty acids reduce the risk of acute coronary events. However, a high
mercury concentration in fish could attenuate this protective effect."
In a more recent follow-up study of the KIHD cohort, the association between mercury
and the risk of acute coronary events and mortality from CVD, CHD, and all causes was re-
evaluated in a group of 1,887 men (Virtanen et al. 2005). The study also examined whether
mercury could interfere with the beneficial effects offish oils. The interaction between mercury
and the serum n-3 end-product fatty acids DHA, DPA, and eicosapentaenoic acid was
investigated. Deaths by CVD, CHD, and all-causes were recorded, and fish consumption was
measured through an interview-checked 4-day food recording. Fish intake in men for those in
the highest third (tertile) of hair mercury content was more than double that of the lowest tertile
(65 vs. 30 g/day, respectively). High mercury content in hair was also most strongly associated
with fish intake and serum DHA plus DPA concentrations. Men in the highest tertile of hair
mercury content (>2.03 ng/g) had an increased risk for frank cardiovascular effects, including
AMI (RR 1.60; 95% CI 1.24-2.06), CVD (RR 1.68; 95% CI 1.15-2.44), CHD (RR 1.56; 95% CI
0.99 to 2.46), and any death (RR 1.38; 95% CI 1.15-1.66), compared with men in the combined
lower two thirds.4 For each microgram of mercury in hair, the risk of acute coronary events
increased, on average, by 11 percent (95% CI 6% to 17%). Based on these results, Virtanen et
al. concluded that "high content of mercury in hair may be a risk factor for acute coronary events
3 Risk factors included age, examination years, body mass index, maximal oxygen uptake, hair mercury content,
serum ferritin, serum LDL cholesterol, systolic blood pressure, serum insulin, ADP-induced platelet aggregation,
socioeconomic status, ischemic findings in exercise test, smoking, place of residence, and dietary energy intake.
4 Estimated increased risks were adjusted for age, examination year, high-density lipoprotein (HDL) and low-density
lipoprotein (LDL) cholesterol, body mass index (BMI), family history of ischemic heart disease, systolic blood
pressure, maximal oxygen uptake, urinary excretion of nicotine metabolites, serum selenium, alcohol intake, serum
DHA + DPA as a proportion of all fatty acids in serum, and intake of saturated fatty acids, fiber, and vitamin C and
E.
C-3
-------
and CVD, CHD, and all-cause mortality in middle-aged eastern Finnish men." Furthermore, the
authors concluded that "mercury may also attenuate the protective effects offish on
cardiovascular health."
C.2.2 The European Multicenter Case Control Study on Antioxidants, Myocardial Infarction
and Cancer of the Breast (EURAMIC) Cohort
In a case-control study using subjects from the EURAMIC study population (n = 1,400
men from eight European countries and Israel), Guallar et al. (2002) investigated the relationship
between the risk of a first myocardial infarction in men and mercury levels measured in toenail
clippings5 and DHA levels in adipose tissue. The investigators reported that mercury levels in
patients who had suffered AMI were 15 percent higher (after adjusting for DHA level and
coronary risk factors) than levels in controls. Additionally, men in the highest quintile of
mercury exposures exhibited a risk-adjusted6 2.16-fold increased risk (odds ratio [OR]) of
myocardial infarction (95% CI 1.09-4.29) when compared to the lowest quintile. DHA level also
was inversely associated with the risk of myocardial infarction after adjusting for mercury level
(OR 0.59; 95% CI 0.30-1.19). Consequently, toenail mercury level was directly associated with
the risk of myocardial infarction and adipose tissue DHA level was inversely associated with the
risk. One study location, which had higher mercury than the others, appeared to be influential in
the analysis. Guallar et al. concluded that "high mercury content may diminish the cardio-
protective effect of fish intake."
C. 2.3 Mechanisms for Cardiovascular Impacts
Currently, there is a general lack of mechanistic evidence for the role of methylmercury
in heart disease (Stern 2005). However, Salonen et al. (1995), Virtanen (2005), and Guallar et
al. (2002) summarize several mechanistic bases by which mercury may increase the risk of
adverse cardiovascular impacts. The increased risk may be related to a reduction in the body's
antioxidative capacity and the promotion of free radical stress and lipid peroxidation. A
reduction in antioxidative capacity may be due to the high affinity of mercury for sulfhydryl
groups (thereby inactivating antioxidative thiolic compounds) and mercury's tendency to bind to
selenium and form an insoluble complex (selenium is believed to be a factor in catalyzing the
formation of free-radical scavengers). Mercury is a transitional metal and therefore can promote
the formation of free radicals via Fenton-type reactions. Additionally, Virtanen et al. (2005)
note that mercury inactivates paraoxonase, an extracellular enzyme that may help prevent AMI.
Mercury may also may promote ADP-induced platelet aggregation and blood coagulation,
inhibit endothelial-cell formation and migration, and affect apoptosis (i.e., programmed cell
death) and inflammatory responses.
5 It should be noted that although measuring mercury exposure through toenail clippings appears to quantitatively
reflect dietary intake, this method has not been well characterized in comparison to hair or blood mercury exposure.
Consequently, it is not possible to distinguish elemental mercury exposure from that of methyl mercury (Stern 2005).
Additionally, results from this study cannot be compared to those that measure mercury through hair or blood.
6 Adjusted for age, DHA, BMI, waist:hip ratio, smoking status, alcohol intake, high-density lipoprotein cholesterol,
diabetes, history of hypertension, parental myocardial infarction, a-tocopherol level, B-carotene level, toenail
selenium level, and toenail weight.
C-4
-------
C.2.4 Other Studies Evaluating CVD and Mercury Levels
In contrast with the aforementioned studies that may demonstrate a correlation between
methylmercury and AMI, CHD, and CVD, Yoshizawa et al. (2002) conducted a study
specifically addressing the relationship between total mercury exposure and the risk of coronary
heart disease, and reported no significant association. This study utilized the Health
Professionals Follow-up Study as its study population and examined a subset of 470 patients,
including men who had fatal coronary disease, nonfatal myocardial infarction, coronary-artery
bypass surgery, or percutaneous transluminal coronary angioplasty, as well as controls. Mercury
levels were measured via toenail clippings. After adjusting for age, smoking and other risk
factors, toenail mercury was not associated with the risk of coronary heart disease (CHD).
Adjustment for intake of n-3 fatty acids from fish did not appreciably change these results.
When dentists were excluded from the analysis, an association of toenail mercury with CHD was
observed by comparing the highest and lowest quintiles of mercury exposure (RR 1.27; 95% CI
0.62-2.59; P for trend = 0.43) (Yoshizawa et al. 2002). This relationship, however, was not
statistically significant (possibly due to the 53 percent reduction in total number of cases, to
220), suggesting that further study with a larger sample may be warranted.
In a recent review of the cardiovascular health effects of methylmercury, Stern (2005)
noted that interpretation of the Yoshizawa results may depend on whether it is hypothesized that
total mercury exposure or specifically methylmercury exposure is responsible for the
cardiovascular effects of mercury. If only methylmercury is responsible for cardiovascular
effects then the elevated exposure to elemental mercury of dentists would tend to confound the
underlying association of methylmercury and cardiovascular effects. Stern (2005) also notes that
the strengthening of the association when an adjustment is made for n-3 fatty acids is consistent
with the observations in other studies that overall risk occurs as a balance between the protective
effect of n-3 fatty acids from fish and the adverse effects from methylmercury.
In a study that focused on inorganic mercury exposure from dental amalgams, Ahlqwist
et al. (1999) examined associations between serum mercury and myocardial infarctions (among
other health outcomes), among 1,462 Swedish women (ages 38 to 60 at study initiation) for 25
years. Mercury concentration in serum was measured in all subjects at the beginning of the
study and, 13 years later, for a subsample of 142 women from one age group (all born in 1922).
No statistically significant association between serum mercury and MI was found. However,
Stern (2005) notes that it is difficult to assess the significance of these findings for
methylmercury exposures that occur through fish consumption as serum mercury
disproportionally reflects inorganic mercury exposure.
In a case-control study of both men and women with a first-time MI from a longitudinal
cohort in northern Sweden (n=78), Hallgren et al. (2001) analyzed erythrocyte mercury
concentration in order to study a possible association between increased fish consumption and
reduced coronary heart disease (concentration of mercury in erythrocytes is often used as an
index offish consumption). The researchers summed the n-3 fatty acids, EPA and DHA, to
express the percentage of total plasma polyunsaturated fatty acids (P-PUFA). Mean levels of
mercury in erythrocytes were reported as 4.44 ng/g in cases and 5.42 ng/g in controls. Stern
(2005) notes that these concentrations translate to about 2.8 and 3.4 ng/L, respectively, for
C-5
-------
mercury concentrations in whole blood. Hallgren et al. (2001) concluded that erythrocyte
mercury concentration alone was not significantly associated with the occurrence of MI but
found that higher levels of P-PUFA's and erythocyte mercury (which was used as a marker of
fish intake) are significantly associated with lower risk of myocardial infarction. Various
combinations of high and low P-PUFA and high and low mercury concentration were assessed,
and none of the combinations had a statistically significant odds ratio (OR) greater than 1.0,
including the group of subjects with high mercury and low P-PUFA that might be expected to
have a higher risk of heart disease. The high mercury-low P-PUFA group did have an OR
greater than 1.0, but this group consisted of only four individuals and the result was not
considered statistically significant.
In two studies of death certificates in the region near Minamata City, Japan, Tamashiro et
al. (1984 and 1986) investigated the significance of heart disease as a primary or secondary
cause of death. Tamashiro et al. (1984) studied the causes of death among those with official
diagnoses of Minamata disease in a region including Minamata City in two groups: those who
died from 1954 to 1969 (n=44), and those who died from 1970 to 1980 (n=334). In the first
group, Minamata disease and diseases of the central nervous system were the underlying causes
of death, but nonischemic heart disease accounted for 50 percent of the secondary causes of
death (multiple causes of death could be reported). Analysis of the second group constituted a
case-control study by matching individuals with a control in the same city or town by age, sex,
and year of death. In no disease were the ORs observed to be significantly high or low.
However, when non-Minamata diseases and Minamata diseases were mentioned on the same
death certificate, non-ischemic heart disease (for males and females combined) was the only
statistically significant cause of death for this second group.
Tamashiro et al. (1986) also investigated the causes of death from death certificates in
the Minamata City area, but focused on fishermen and their families residing in a small coastal
area of the city. Certificates from 1970 to 1981 were examined and standard mortality rates
(SMRs) were calculated using age-specific death rates for Minamata City as a standard. The
results showed that SMRs for all categories of heart disease and hypertensive disease were not
significantly elevated in the study population. Stern (2005) note several uncertainties associated
with these results. He notes that deaths from methylmercury-related cardiovascular death could
have possibly peaked prior to 1970, even though incidence of "Minamata disease" was observed
to peak after this date in the study area (thus the selection of the study time period). The
exposure to methylmercury may have varied throughout the study area, and death rates in
Minamata City itself were used as the denominator in calculating the SMRs. Therefore,
although overall deaths were more prevalent in this study area, lower exposures to
methylmercury that resulted in higher rates of heart disease could have occurred. Furthermore,
there were no data reported on the type offish consumed and the relative consumption of n-3
fatty acids which may provide protection against the cardiovascular effects of methylmercury.
Stern (2005) also notes in his review that differences in the results from the two studies by
Tamashiro et al. are probably related to study design, and the weaknesses of the Tamishiro 1986
study likely explain the differences in the studies with respect to the identification of heart
diseases as potentially associated with highly elevated methylmercury exposures resulting in
Minamata disease. Stern also notes that for ischemic heart disease only males appear to be at
risk.
C-6
-------
C.3 Other Cardiovascular Effects
Another possible adverse cardiovascular effect related to methylmercury exposure is
accelerated progression of carotid atherosclerosis (i.e., the progressive narrowing and hardening
of the arteries over time). This effect was observed in men from the KIHD study population, in
which high hair mercury content was the second strongest predictor for the four-year increase in
the mean intima-media thickness (IMT), a marker of early artherosclerosis (Salonen et al. 2000).
Predictors for IMT ranked as follows: systolic blood pressure > hair mercury content>
antidyslipidemic medication> dietary iron intake > cigarette pack years > age. Additionally,
there was an 8 \im incremental increase in the four-year IMT for each \ig/g increase in hair
mercury content.
The case for mercury-induced blood pressure effects is not as strong, but some evidence
suggests a possible link between methylmercury exposure and hypertension and heart rate
variability. Oka et al. (2003) reviewed electrocardiogram data, along with blood pressure and
pulse pressure measurement, in Minamata patients who had been exposed to methylmercury in
utero and were institutionalized for "fetal Minamata disease" as adults. Subjects had a
significantly elevated resting heart rate and a slightly (but not significantly) decreased variability
in heart rate when compared to controls. Blood pressure did not differ between the two groups,
but pulse pressure was significantly decreased among the subjects. Considering the limited data
available, it is difficult to draw any conclusions at this time regarding heart rate variability and
methylmercury exposure; more research is necessary to fully explore this area.
Cardiovascular effects have also been reported for children. A decrease in heart rate
variability, an important measure of the ability of the cardiovascular system to withstand stress,
was reported for children in the Faroe Islands (917 seven-year-old children) exposed in utero to
methylmercury through maternal consumption offish and marine mammals (additional post-
natal exposures, again though consumption offish and marine mammals may have also
occurred). At seven years, children who had been exposed before birth to higher levels of cord
blood mercury (up to 10 ng/L) had increased blood pressure as well as a decrease in heart rate
variability (Sorensen et al. 1999). Blood pressure was based upon a single measure for each
child. At fourteen years, the effect on blood pressure in this cohort was no longer observed but
heart rate variability remained low (Grandjean et al. 2004). It is unclear, however, whether this
effect will persist beyond age 14 and what the significance of this finding is for adverse health
outcomes later in life.
A recent study by Vupputuri et al. (2005) showed no statistically significant association
between total blood mercury and blood pressure in a sample of 1,240 women aged 16 to 49 years
from the National Health and Nutrition Examination Survey 1999-2000. Additionally, no
association was observed between total blood mercury and blood pressure in fish consumers
when data were stratified by dietary fish intake (subjects were separated into those who consume
fish and those who consume no fish).7 More research is required to reduce the uncertainty
7 A weak association was observed, however, between total blood mercury and blood pressure among subjects who
did not consume any fish (i.e., systolic blood pressure increased by 1.83 mm mercury per 1.3 jig/L increase in total
blood mercury; 95% CI 0.36-3.30; P = 0.018; adjusted for age, race, income, body mass index, pregnancy status, and
dietary sodium, potassium, and total calories). Vupputuri et al. (2205) suggest that the oils in fish may counteract
C-7
-------
associated with the possibility of other cardiovascular effects, including blood pressure, from
exposure to methylmercury in both adults and children. A potential confounding effect in
several of the dietary studies could be based on method of preparation offish (Mozaffarian et al.
2004).
C.4 Cardiovascular Health Benefits of Fish Consumption
Current federal dietary recommendations about fish consumption are found in the 2005
Dietary Guidelines for Americans (DHHS and USDA 2005). The Guidelines were based on an
expert scientific report from the 2005 Dietary Guidelines Advisory Committee (2005 Dietary
Guidelines Advisory Committee 2004). The Advisory Committee report in turn references two
major sources: the National Academy of Sciences Institute of Medicine (IOM) report on Dietary
References Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and
Amino Acids (IOM 2002) and the DHHS Agency for Healthcare Research and Quality (AHRQ)
evidence-based AHRQ Report Effects of Omega-3 Fatty Acids on Cardiovascular Disease
(Wang et al. 2004). The main conclusions of the IOM and AHRQ reports are summarized below.
The Advisory Committee also identified the main conclusions of six other national and
international expert groups regarding recommendations for fish consumption.
Available evidence suggests that the cardiovascular health benefits of fish consumption
are related to the long chain omega-3 polyunsaturated fatty acids in fish (LC omega-3 PUFA,
also called LC omega-3 PUFA), which are unique nutrients in fish. The specific omega-3 fatty
acids are docosahexaenoic acid (DHA) and eicosahexaenoic acid (EPA). In the general
population (rather than in heart patients), most of the studies were observational (epidemiologic)
studies offish consumption and cardiovascular disease and not studies of omega-3 fatty acid
supplements (for example, see Wang et al. 2004 and FDA 2004). Therefore, although evidence
indicates that the cardiovascular health benefits offish consumption are related to the omega-3
fatty acids in fish oil, DHA and EPA, the actual studies, relevant to the general population,
showing cardiovascular health benefits were for fish in the diet, not for fish oil supplements.
According to the Dietary Guidelines, the current federal dietary recommendations are that most
fats should come from sources of polyunsaturated and monounsaturated fatty acids, such as fish,
nuts and vegetable oils. Based on the scientific review of the Advisory Committee, the Dietary
Guidelines note that limited evidence suggests an association between consumption of fatty acids
in fish and reduced risks of mortality from cardiovascular disease for the general population.
Other sources of EPA and DHA may provide similar benefit; however, more research is needed.
The Dietary Guidelines further state that "evidence suggests that consuming approximately two
servings offish per week (approximately 8 ounces total) may reduce the risk of mortality from
coronary heart disease and that consuming EPA and DHA may reduce the risk of mortality from
cardiovascular disease in persons who have already experienced a cardiac event." The Dietary
Guidelines explain that the Food and Drug Administration and EPA are advising women of
childbearing age who may become pregnant, pregnant women, nursing mothers, and young
children to avoid some types offish and shellfish and to eat fish and shellfish that are lower in
mercury, and refer readers to the FDA telephone hotline and web site.
the potentially harmful effects of mercury on blood pressure.
C-8
-------
The scientific review by the Dietary Guidelines Advisory Committee on fish
consumption used as its starting point the comprehensive Institute of Medicine report on
macronutrients (IOM 2002). The IOM report found that "a growing body of literature suggests
that diets high in EPA and DHA may afford some degree of protection against CHD." The
Advisory Committee review also relied upon an evidence-based review by AHRQ (Wang et al.
2004). The AHRQ report evaluated 22 prospective cohort studies that were conducted in the US
and other developed countries. AHRQ noted that most of the cohorts had several thousand
subjects; the range was 272 to 223,170 subjects, with most subjects at least age 40. AHRQ
indicated that, despite some limitations, if viewed together, these studies provide evidence that is
highly applicable to the U.S. population. Overall, AHRQ found that the evidence from the
primary and secondary prevention studies supports the hypothesis that the consumption of
omega-3 fatty acids, fish, and fish oil reduces all-cause mortality and various cardiovascular
disease (CVD) outcomes. These outcomes include sudden death and cardiac death (coronary or
myocardial infarct (MI) death). Thus, the AHRQ report concluded that the consumption of
omega-3 fatty acids from fish or from supplements offish oil reduces all-cause mortality and
various CVD outcomes. Additionally, the Advisory Committee based its conclusions on its own
analysis of epidemiologic studies of the cardioprotective effects offish consumption among
healthy populations, such as those by Dolecek (1992), Siscovick et al. (1995), Hu et al. (2002),
and Mozaffarian et al. (2003) (cited in Dietary Guidelines Advisory Committee 2004).
FDA in 2004 (FDA 2004) announced the availability of a qualified health claim for
reduced risk of coronary heart disease (CHD) on conventional foods that contain EPA and DHA
omega-3 fatty acids. The FDA explained that, typically, EPA and DHA omega-3 fatty acids are
contained in oily fish, such as salmon, lake trout, tuna and herring. FDA also stated that these
fatty acids are not essential to the diet; however, scientific evidence indicates that these fatty
acids may be beneficial in reducing CHD. An example of wording of the qualified health claim
as it would appear on a food label would be, "Supportive but not conclusive research shows that
consumption of EPA and DHA omega-3 fatty acids may reduce the risk of coronary heart
disease. One serving of [name of food] provides [x] grams of EPA and DHA omega-3 fatty
acids. [See nutrition information for total fat, saturated fat and cholesterol content.]"
C.5 Conclusions
In summary:
• Studies investigating the relationship between methylmercury and cardiovascular
impacts have reached different conclusions. The findings to date and the
plausible biological mechanisms warrant additional research in this arena (Stern
2005; Chan and Egeland 2004).
• Some recent epidemiological studies of men suggest that methylmercury is
associated with a higher risk of acute myocardial infarction, coronary heart
disease and cardiovascular disease in some populations. Other recent studies
have not observed this association.
• Some studies have suggested that methylmercury attenuates the beneficial effects
offish consumption. A further possible explanation is that the observed effect
C-9
-------
may be present in certain populations but is not generalizable to other
populations.
• There is a significant and well-recognized body of literature documenting the
cardioprotective benefits offish consumption.
• As the science on the impact of methylmercury on the risk of cardiovascular
events remains uncertain, and the weight of the evidence, in fact, supports a
positive association between fish consumption and potential cardiovascular
benefits, the impacts of methylmercury from fish consumption are only discussed
qualitatively.
C.6 References
2005 Dietary Guidelines Advisory Committee, August, 2004. Report of the 2005 Dietary
Guidelines Advisory Committee.
http://www.health.gov/dietaryguidelines/dga2005/default.htm
http://www.health.gov/dietaryguidelines/dga2005/report/
Ahlqwist M, Bengtsson C, Lapidus L, Bergdahl IA, and Schiitz A. 1999. Serum mercury
concentration in relation to survival, symptoms, and diseases: results from the
prospective population study of women in Gothenburg, Sweden. Act Odontol Scand
57:168-174.
Calder PC. 2003. New evidence in support of the cardiovascular benefit of long-chain n-3 fatty
acids. Italian Heart Journal, 4: 427-429.
Chan HM and Egeland GM. 2004. Fish consumption, mercury exposure, and heart diseases.
Nutrition Reviews 62(2):68-72.
Curb JD and DM Reed. 1985. Fish consumption and mortality from cardiovascular disease. N
EnglJMed. 313:821-822.
Dolecek TA. 1992. Epidemiological evidence of relationships between dietary polyunsaturated
fatty acids and mortality in the Multiple Risk Factor Intervention Trial. Proc Soc Exp
Biol Med 200:177-182. (as cited in Dietary Guidelines Advisory Committee 2004)
Folsom, AR and Z Demissie. 2004. Fish intake, marine omega-3 fatty acids and mortality in a
cohort of postmenopausal women. Am. Journal of Epidemiology. 160(10): 1005-1010.
Food and Drug Administration, September 8, 2004. Qualified Health Claim for Omega-3 Fatty
Acids in Foods, http://www.cfsan.fda.gov/~dms/lab-qhc.html
http://www.fda.gov/bbs/topics/news/2004/NEW01115.html.
Grandjean P, Murata K, Budtz-Jorgensen E, Weihe P. 2004. Autonomic activity in methyl
mercury neurotoxicity: 14-Year follow-up of a Faroese Birth Cohort. J Pediatr 144:169-
176.
C-10
-------
Guallar E, Sanz-Gallardo I, Van't Veer P, Bode P, Aro A, Gomez-Aracena J, Kark JD,
Riemersma RA, Martin-Moreno JM, Kok FK. 2002. Mercury, fish oils, and the risk of
myocardial infarction. N Engl J Med 347(22): 1747-1754.
Hallgren CG, Hallmans G, Jansson J-H, Marklund SL, Huhtasaari F, Schiitz A, Stromberg U,
Vessby B, and Skerfving S. 2001. Markers of high fish intake are associated with
decreased risk of a first myocardial infarction. British Journal of Nutrition 86:397-404.
Hu FB, Bronner, W Willett, Stampfer MJ, Rexrode KM, Albert CM, Hunter D, Manson JE.
2002. Fish and omega-3 fatty acid intake and risk of coronary heart diease in women.
JAMA. 287(14):1815-1821. (as cited in Dietary Guidelines Advisory Committee 2004)
Morris, MC, J.E. Manson, B. Rosner et al. 1992. A prospective study offish consumption and
cariovascular disease. Circulation. 86 (Suppl. 1):1-163.
Mozaffarian D, Lemaitre RN, Kuller LH, Burke GL, Tracy RP, Siscovick DS; Cardiovascular
Health Study. 2003. Cardiac benefits offish consumption may depend on type offish
meal consumed. Circulation 107:1372-7. (as cited in Dietary Guidelines Advisory
Committee 2004)
Mozaffarian D, MP Bruce, Rimm EB, Lemaitre RN, Burke GL, Lyles MF, Lefkowitz D,
Siscovick DS. 2004. Fish Intake and Risk of Incident Atrial Fibrillation. Circulation.
110:368-373.
National Academy of Sciences, Institute of Medicine (IOM). 2002. Dietary Reference Intakes:
Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein and Amino Acids.
The National Academy Press, Washington, DC.
National Research Council (NRC). 2000. Toxicological Effects of Methylmercury. Committee
on the Toxicological Effects of Methylmercury, Board on Environmental Studies and
Toxicology, Commission on Life Sciences, National Research Council. National
Academy Press, Washington, DC.
Oka T, Matsukura M, Okamoto M, Harada N, Kitano T, Minke T, Futasuka M. 2003.
Autonomic nervous function in fetal type Minamata disease patients: Assessment of
heart rate variability. Tohoku J. Exp. Med. 198:215-221.
Rissanen T, Vourilaninen S, Nyyssonen K, Lakka, T, Salonen, JT. 2000. Fish oil-derived fatty
acids, docosahexaenoic acid and docosapentaenoic acid, and the risk of acute coronary
events. The Kuopio Ischemic Heart Disease Risk Factor Study. Circulation 102:2677-
2679.
Salonen JT, Seppanen K, Nyssonen K, Korpela H, Kauhanen J, Kantola M, Tuomilehto J,
Esterbauer H, Tatzber F, Salonen R. 1995. Intake of mercury from fish, lipid
peroxidation, and the risk of myocardial infarction and coronary, cardiovascular, and any
death in eastern Finnish men. Circulation 91:645-655.
C-ll
-------
Siscovick DS, Raghunathan TE, King I, Weinmann S, Wicklund KG, Albright J, Bovbjerg V,
Arbogast P, Smith H, Kushi LH, et al. 1995. Dietary intake and cell membrane levels of
long-chain n-3 polyunsaturated fatty acids and the risk of primary cardiac arrest. JAMA
274:1363-7. (as cited in Dietary Guidelines Advisory Committee 2004)
Salonen JT, Seppanen K, Lakka, TA, Salonen R, Kaplan GA. 2000. Mercury accumulation and
accelerated progression of carotid atherosclerosis: A population-based prospective 4-
year follow-up study in men in eastern Finland. Atherosclerosis 148:265-273.
Sorensen N, Murata K, Budtz-Jorgensen E, Weihe P, Grandjean P. 1999. Prenatal methyl
mercury exposure as a cardiovascular risk factor at seven years of age. Epidemiol
10:370-375.
Stern AH. 2005. A review of the studies of the cardiovascular health effects of methylmercury
with consideration of the suitability for risk assessment. Environmental Research
98(1):133-142.
Tamashiro H, Akagi H, Arakaki M, Futatsuka M, and Roht LH. 1984. Causes of death in
Minamata disease: analysis of death certificates. Int Arch Occup Environ Health 54:135-
146.
Tamashiro H, Arakaki M, Futatsuka M, and Lee ES. 1986. Methylmercury exposure and
mortality in southern Japan: a close look at causes of death. Journal of Epidemiology and
Community Health 40:181-185.
U.S. Department of Health and Human Services and U.S. Department of Agriculture (DHHS and
USDA), January, 2005. 2005 Dietary Guidelines for Americans.
http://www.healthierus.gov/dietaryguidelines/
http://www.health.gov/dietaryguidelines/dga2005/document/
Virtanen J, Voutilaninen S, Rissanen TH, Mursu J, Tuomainen T-P, Korhonen MJ, Valkonen V-
P, Seppanen K, Laukkanen JA, Salonen JT. 2005. Mercury, fish oils, and risk of acute
coronary events and cardiovascular disease, coronary heart disease, and all-cause
mortality in mean in eastern Finland. Arterioscler Thromb Vase Biol 25:228-233.
Vupputuri S, Longnecker MP, Daniels JL, Guo X, Sandier DP. 2005. Blood mercury level and
blood pressure among US women: Results from the National Health and nutrition
Examination Survey 1999-2000. Environmental Research 97:195-200.
Wang C, Chung M, Lichtenstein A, Balk E, Kupelnick B, DeVine D, Lawrence A, Lau J. 2004.
Effects of Omega-3 Fatty Acids on Cardiovascular Disease. Summary, Evidence
Report/Technology Assessment No. 94. (Prepared by the Tufts-New England Medical
Center Evidence-based Practice Center, Boston, MA.) AHRQ Publication No. 04-E009-
1. Rockville,MD: Agency for Healthcare Research and Quality. March 2004.Agency for
Healthcare Research and Quality (AHRQ), DHHS March, 2004. Omega-3 Fatty Acids
Effects on Cardiovascular Disease, http://www.ahrq.gov/clinic/epcindex.htmftdietsup
C-12
-------
Yoshizawa K, Rimm EB, Morris S, Spate VL, Hsieh C-C, Spiegelman D, Stampfer MJ, Willett
WC. 2002. Mercury and the risk of coronary heart disease in men. N Engl J Med
347:1755-1760.
C-13
-------
APPENDIX D D-l
NORMALIZATION OF MERCURY IN FISH TISSUE SAMPLES D-l
D.I Methods D-l
D.I.I National Descriptive Model of Mercury in Fish (NDMMF) D-l
D.2 General Examination of Model Performance D-2
D.2.1 NDMMF Estimated Values D-2
D.2.2 Accuracy of NDMMF Estimated Values D-3
D.2.3 Spatial Examination of Model Performance D-5
D.2.4 Predictive Examination of Model Performance (Withheld Data Set) . D-6
D.3 References D-9
Tables
Table D-l. Statistical Distribution of Residuals D-3
Table D-2. Differences between the Performance of Lake and River Samples Used as Inputs
into the NDMMF D-5
Table D-3. Statistical Distribution of Residuals from Withheld Data Set D-7
Figures
Figure D-l. Box and Whisker Plots of the NLFWA Observed, NDMMF Estimated, and
Residuals Measurements in ppm D-3
Figure D-2. Scatterplot of Predicted vs. Observed Measurements D-4
Figure D-3. Scatterplot of Residual vs. Observed Measurements D-4
Figure D-4. Locations of Withheld Observations D-7
Figure D-5. Box and Whisker Plots of Observed, Predicted, and Residual (Error) Distributions
for the Withheld Data Set D-8
Figure D-6. Scatterplot of Predicted vs. Observed Measurements D-8
Figure D-7. Scatterplot of Residual vs. Observed Measurements D-9
-------
APPENDIX D
NORMALIZATION OF MERCURY IN FISH TISSUE SAMPLES
This section of the appendix provides a detailed description of the procedures used to
normalize the fish tissue data discussed in Section 5 using the National Descriptive Model of
Mercury in Fish (NDMMF). We also provide a general examination of model performance. It
should be noted that the examination of model performance is based on a normalization of the
NLFA data for the years 1990 to 2002. The validation demonstrates that the NDMMF performs
well with this set of data. Additional data recently became available for the year 2003 for the
NLFA and for the first two years of the NLFTS after the validation was conducted. Therefore,
EPA updated the normalization with the new data for application in the final benefit analysis
presented in Section 10.
D.I Methods
D.I.I National Descriptive Model of Mercury in Fish (NDMMF)
The United States Geological Survey developed a procedure called the National
Descriptive Model of Mercury and Fish Tissue (NDMMF) (Wente 2004). The NDMMF model
provides a translation factor to convert a mercury concentration taken from one
species/size/sample method to an estimated concentration for any other user pre-defined
species/size/sample method. The model provides species/size/sample estimates for each
sampling site and year. New concentrations can only be estimated by the NDMMF if at least
one sample has already been taken at a particular location. In other words, predictions can only
be made for locations and dates where at least one sample has already been recorded.
The NDMMF is a statistical model related to covariance. The model is designed to allow
the prediction of different species, cuts, and lengths offish for sampling events, even when those
species/lengths/cuts offish were not sampled during those sampling events.
The NDMMF models mercury concentration as a power function offish length, i.e., y =
axb, where y = mercury concentration, x = length, and a,b are parameters. Potentially, that could
be done for each species at each site, but the data are far too sparse for that, and, further, he
wanted to devise some way to predict mercury concentrations for a species that was not even
collected at some given site. So, the assumptions are made that: (i) a universal, national value
of "b" for each tissue type from each fish species, and (ii) the value of "a" at a given site at a
given time is the same for all. species. Thus a prediction of mercury concentration for a given
tissue from a fish of an arbitrary species of length x could then be predicted by "looking up" the
right value of "b" for that species and tissue and the value of "a" for that site and time. The
following is the SAS code used to implement the NDMMF.
D-l
-------
data Hg.nlfwaStepTwo (keep = rec dl hg leng spc event upper lower llength);
set Hg.nlfwaStepOne;
rec = rec;
dl = dl;
hg = hg;
leng = leng;
spc = spc;
event = event;
upper = log(hg*1000+l);
if DL = 1 then lower = .; else lower = upper;
if Leng > 0 then llength = log(leng+l); else delete;
run;
proc lifereg noprint data=Hg.nlfwaStepTwo outest=Hg.b6;
Class SPC Event;
Model (lower, upper) = SPC*Llength Event/ d=normal noint;
output out=Hg.Mod6Pred p=pred6;
run;
quit;
Code filename and variable Definitions:
Hg.nlfwaStepOne is the data filename
rec = unique ID for each input record;
dl = if dl = 1 then the sampled MeHg was below detection limits;
hg = Sampled MeHg;
leng = length of the sampled fish;
spc = unique code for every unique species/sample method combination starting with 1
and ending at n where n = number of unique combinations; and
event = unique code for every unique location/date combination starting with 1 and
ending at n where n = number of unique combinations.
D.2 General Examination of Model Performance
D.2.1 NDMMF Estimated Values
The 2002 NLFA was cleaned and subset according to the methods described in chapter 5
and used to calibrate the NDMMF. The output slopes and intercepts were then used to estimate
every observation that was an input to the model calibration. Thus, for this section, an input of a
whole 3 in. chub sampled from waterbody A on January 1, 1992 is re-predicted using the
NDMMF as a whole 3 in. chub sampled from waterbody A on January 1, 1992.
The input observed data characteristics and estimated NDMMF data characteristics are
very similar. The mean fish tissue concentration of the observed input data set was a 0.39, and
the mean of the NDMMF estimated fish tissue concentration was 0.36. The range changed from
0 to 9 ppm, to an NDMMF estimated 0 to 5 ppm. Hg fish tissue concentration. The standard
deviation of the observed data was 0.4. The standard deviation of the estimated data was 0.36.
Figure D-l graphically depicts the characteristics of the observed and NDMMF estimated data
through the use of box and whisker plots. These results include censored values (where input
concentrations were below detection limits).
D-2
-------
D.2.2 Accuracy of NDMMFEstimated Values
The NDMMF was used to estimate every observation that was entered into the model.
The estimated concentrations are then compared with the observed data inputs to obtain residuals
(estimated - observed = residual). Table D-l details the statistical distribution of these residuals
and compares them to the observed data. Figure D-2 graphically displays the distribution of the
observed, NDMMF estimated, and residual values in box and whisker plots. The red x indicates
the mean of the distributions. Boxes are drawn around the 75th and 25th percentiles. The ends
of the whiskers are drawn at the minimum and maximum values. Figures D-3 and D-4 are
scatter plots of observed, predicted, and residual values.
Table D-l. Statistical Distribution of Residuals
Mean
Min
Max
Std. Dcv.
NLFA Observed Data
0.39
0
8.9
0.42
NDMMF Estimated
0.36
0
5.3
0.36
Mean of Residual (Error)
-0.02
^t.80
4.8
0.2
ppm.
5.0
2.5 -
-2.5
0 -
Observed Estimated
-5.0
I
Residual
Figure D-l. Box and Whisker Plots of the NLFWA Observed, NDMMF Estimated, and
Residuals Measurements in ppm
D-3
-------
Predicted (ppm.)
5-f
4 -
3-
2 -
1 -.
0-
-1 -
I 1 I
-1012
Observed (ppm!)
T I
3 4
Figure D-2. Scatterplot of Predicted vs. Observed Measurements
Residual (ppm.)
6 -I
Observed (ppm.)
Figure D-3. Scatterplot of Residual vs. Observed Measurements
D-4
-------
The box and whisker plots indicate that the distribution of the predicted data is a little
narrower than the observed data. In both cases there are extreme high-end outliers. The
distribution of the residuals from the NDMMF is centered around zero. The whiskers do
indicate that there are some outliers within the residual data.
The average residual value of-0.02 is relatively small and we believe this indicates a
fairly even mix between over and under predictions. The residual standard deviation of 0.2
indicates that most of the predictions are within 0.2 ppm. of the observed value. 0.2 ppm. is 51
percent of the mean fish tissue concentration. Figure D-4 indicates a systematic underestimation
of higher values.
D.2.3 Spatial Examination of Model Performance
D. 2.3.1 Accuracy Differences Between Lake and River Environments
The samples found within the NLFA, and used as inputs into the NDMMF, were taken
from over 4,000 different locations across the U.S. These sample locations range in ecological
character from Florida lakes to Wisconsin rivers.
The type of waterbody sampled was not recorded in the NLFA. To estimate the type of
waterbody samples originated from, the distance to the nearest lake and nearest flowing water
source was determined within a Geographic Information System (GIS). The shortest distance
source waterbody type was assigned to each geocoded sample point. Approximately 1/3 of the
observations in the NLFA were assigned a "lake" source type, and 2/3 were assigned "river"
source type.
To examine if there is a difference in the performance or accuracy of estimates between
observations originating from these two different types of sources, the observed concentrations,
estimated concentrations, and residuals are reported by source type in Table D-2 below.
Table D-2. Differences between the Performance of Lake and River Samples Used as
Inputs into the NDMMF
Observed
NDMMF Estimated
Residual
Lake
Mean
Min
Max
Std. Dev.
0.36
0
5.7
0.36
0.33
0
2.8
0.3
-0.02
-3.64
1.16
0.18
River
Mean
Min
Max
Std. Dev.
0.4
0
8.9
0.45
0.38
0
5.3
0.39
-0.02
-4.76
4.79
0.22
D-5
-------
An examination of the residuals shows that the NDMMF performs very similarly for lake
and river source type predictions of Hg fish tissue concentrations. Any variability in model
performance is not attributable to differences between rivers and lakes.
D. 2.4 Predictive Examination of Model Performance (Withheld Data Set)
The examination of residuals where the prediction fish type is equal to the observed
sample fish type is useful for evaluative purposes, but does not exactly parallel the planned
future application of the NDMMF for data preparation for benefits analysis. To calculate
benefits from Hg fish tissue samples, where the method of exposure is consumption, we must
have either observations, or predictions, of Hg fish tissue concentrations found in consumable
fish. For this reason, the NDMMF will be used to predict concentrations for fish types (species,
length, sample method) that may not be equal to the observed sample fish type.
For example, if a 5 in. sunfish were sampled using a whole fish sample method, the
NDMMF may be used to predict what the Hg fish tissue concentration would have been had the
sample been taken from the fillet of an 11 in. bass. In this case, and in the application of the
NDMMF for benefits analysis, the prediction fish type is not the same as the observed sample
fish type.
To examine the NDMMF's performance within the context of its application for benefits
analysis, where the observed and estimated fish types are likely to be different, a subset of the
data (approximately 10 percent) was withheld to use as a validation dataset. The NDMMF was
then implemented utilizing the remaining observations as a training dataset. The output slopes
and parameter estimates were then used to predict the Hg fish tissue concentrations of the
withheld data set. Figure D-4 shows the spatial distribution of withheld data observations.1
'Lack of observations throughout Pennsylvania, Kentucky, West Virginia, Tennessee, Ohio, Virginia, Kansas, and
Missouri reflect to a lack of input observations in those states into the NDMMF. Filters (described in detail in
chapter 5) for the most part removed these observations because no fish length or weight were recorded with the
Hg concentration.
D-6
-------
• "Withheld Observations Analyzed
CD States
Figure D-4. Locations of Withheld Observations
The mean fish tissue concentration of the observed withheld data was 0.38 ppm. The
mean predicted fish tissue concentration for the same withheld observations was 0.35 ppm. The
range changed from 0 to 3.9 to 0 to 4.25. The standard deviation of the fish tissue concentrations
of the withheld observed data concentrations was 0.41. The predicted concentration standard
deviation was a 0.37. Table D-3 details the statistical distribution of the observed and predicted
data, and the residuals. Figure D-5 graphically depicts the characteristics of the withheld
observed and predicted concentrations and the residuals. Figures D-6 and D-7 are scatterplots of
predicted and residual measurements vs. observed measurements for the withheld data.
Table D-3. Statistical Distribution of Residuals from Withheld Data Set
NLFA Observed Data
NDMMF Estimated
Residual
Mean
Min
Max
Std. Dev.
0.39
0
3.9
0.41
0.35
0
4.25
0.37
-0.03
-2.14
3.05
0.24
D-7
-------
ppm
5 -
4 -
3 -
2 -
1
0
-1
I I
Observed Predicted
ppm
4 1
3
2 -I
1
0 •
-1 •
-2
-3 -
1
Residual
Figure D-5. Box and Whisker Plots of Observed, Predicted, and Residual (Error)
Distributions for the Withheld Data Set
Predicted (pprn.)
5-1
A -
3 -
2 -
-1 -
o
o
-1
I
0
1
I r
3 A
Observed (ppm.)
Figure D-6. Scatterplot of Predicted vs. Observed Measurements
D-8
-------
Observed
Figure D-7. Scatterplot of Residual vs. Observed Measurements
An analysis of the residuals from the withheld predicted data set shows that the average
residual is -0.03 ppm. fish tissue concentration. EPA believes the closeness of the average
residual to the mean, the visual analysis of the box and whisker plot, and the scatterplot of
residual vs. observed measures, indicate that the model is neither systematically under or
over-predicting the majority of the data. There is a slight under-prediction of higher values;
however EPA believes it is small enough, and the observations are few enough, to be acceptable
for benefits analysis. The range of individual observation prediction error is from -2.14 (an
under-prediction) to 3.05 (an over-prediction). A residual standard deviation of 0.24 (62 percent
of mean observed concentration) indicates that a large portion of the observations were predicted
to within 0.24 ppm. fish tissue concentration.
It is evident that there are many issues surrounding the NLFA and sample selection, and
how that influences the outputs of the NDMMF. Even given these less than ideal input data
circumstances, EPA believes the NDMMF adds significantly to the value of a benefits analysis
by generating a reasonably accurate estimate of Hg fish tissue concentrations from consumable
fish too small for consumption or of a species not typically targeted by anglers. This enables the
more complete use of the NLFA data set and allows EPA to generate benefit estimates for a
larger portion of the population.
D.3 References
Wente, S.P. 2004. A Statistical Model and National Data Set for Partitioning Fish-Tissue
Mercury Concentration Variation between Spatiotemporal and Sample Characteristic
Effects: U.S. Geological Survey Scientific Investigations Report 2004-5199
D-9
-------
APPENDIX E-1 ANALYSIS OF TRIP TRAVEL DISTANCE FOR FRESHWATER
ANGLERS El-1
El.l Data El-1
E1.2 Analysis of Travel Distance Data El-2
El.3 Summary Results Applied in the Population Centre id Approach El-4
APPENDIX E-2 METHODOLOGY FOR ESTIMATING FRESHWATER FISHING
DAYS BY WATERSHED E2-1
E-2.1 Data E2-1
Tables
Table El-1. Reported Trip Travel Distance for Freshwater Anglers (miles) El-2
Table El-2. Demographic Characteristics of Freshwater Anglers3 El-3
Table El-3. Demographic Characteristics of Freshwater Anglers El-3
Table El-4. OLS Regression Results for Determinants of Reported Trip Travel
Distance (miles) El-5
Table El-5. Travel Distance Frequencies by Demographic Group (Percentage in each Distance
Category) El-6
Table E2-1. Frequency Distributions for HUC Level-of Use Indicators E2-4
Table E2-2. Variable Definitions and Descriptive Statistics E2-5
Table E2-3. Estimated Determinants of HUC Level-of-Use Indicators for Lake Trips: Negative
Binomial Regressions E2-6
Table E2-4. Estimated Determinants of HUC Level-of-Use Indicators for River Trips: Negative
Binomial Regressions E2-7
Table E2-5. Predicted Level-of-Use Indicators for HUCs in Study Area: Negative Binomial
Regression Model Predictions E2-8
Figures
Figure E2-1. U.S. Hydrologic Regions E2-2
-------
APPENDIX E-l
ANALYSIS OF TRIP TRAVEL DISTANCE FOR FRESHWATER ANGLERS
As described in Section 10.1.2, using the population centroid approach to estimate
exposures to mercury in freshwater fish requires information about how far individuals typically
travel for freshwater fishing. This appendix describes the data and methods used to analyze
travel distance patterns by freshwater anglers, and it reports the results that were used in
applying the population centroid approach.
El.l Data
To conduct an analysis of trip travel distance for freshwater anglers, we used data from
the NSRE 1994. As described previously, this 16,000-person survey elicited information on
water-based recreation activities—specifically boating, fishing, swimming, and wildlife
viewing—during the previous year. Respondents were asked about their most recent trip taken
in each of the four categories. Of particular interest to this analysis is data concerning fishing
trip characteristics for all respondents who fished in freshwater bodies during the previous year.
Of the 3,220 respondents who had reported fishing, 2,482 visited either a lake, pond, river, or
stream on their most recent trip.
The fishing module elicited location information about most recent fishing trip taken
during the preceding 12 months. This trip was recorded as either a single- or multiday trip to a
specific water body ("site") identified by the respondent. Subsequently, a series of questions
were asked to gather location data on the specific site visited, including the site name, the state in
which the site was located, and the name of the city or town nearest the site. To identify
potential determinants of travel distance for a freshwater fishing trip, we analyzed the 2,384
available responses to the following survey question: "What was the one way travel distance, in
miles from your home, to your destination on *site*?" Table El-1 presents summary statistics
for travel distance, which are reported separately for single-day, multiday, and aggregated trips.
As would be expected, median travel distance varied according to trip type, from 20 miles for a
single-day trip to almost 140 miles for a multiday trip. Across both trip types, the average travel
distance was slightly under 100 miles.
El-1
-------
Table El-1. Reported Trip Travel Distance for Freshwater Anglers (miles)
N Mm" P5 P25 P50 Mean P75 P95 Max
All trip types
Single-day trips only
Multiday trips only
2384"
1791
586
0
0
3
2
2
18
10
10
70
20
20
138
91.9
41
248.2
45
45
300
125
125
850
3000
1100
3000
a Seven respondents reported traveling 0 miles for their most recent trip; all were described as single-day trips.
b Seven respondents did not report whether trip was single-day or multiday.
Note: Ninety-eight respondents who visited freshwater bodies on their most recent fishing trip did not report the
travel distance.
E1.2 Analysis of Travel Distance Data
The influence of multiple demographic characteristics on travel distance was tested using
multivariate regression analysis. Table El-2 reports descriptive statistics for the anglers
included in this analysis. As indicated by the table, over 90 percent of the sample is white; males
comprise a higher percentage of the sample (62 percent) than females. More than half the
sample had completed at least some college and three-fourths of the sample reported being
employed. The survey asked respondents to classify their place of residence as either rural,
suburban, or urban. Approximately 40 percent described their area as rural, 37 percent as
suburban, and 23 percent as urban. Respondents were assigned to a U.S. Census geographic
region by matching their zip code to a corresponding state. The states were then aggregated to
the appropriate Census region (http://www.census.gov/geo/wwvv/us_regdiv.pdf). The majority
of respondents resided in the South and Midwest, followed by the West and Northeast.
Table El-3 presents additional characteristics on the demographic distribution of the
sample. The average age of respondents was 38 years, while household size averaged
approximately three members, with fewer than one person under the age of six. Respondents'
average weekly leisure time was 28 hours. However, this varied significantly across the sample,
from zero to 168 hours. In the survey, family income is reported as a categorical variable, with
respondents selecting the income range that reflected family income in the previous year. The
midpoint of this range was taken to produce a continuous income variable. Subsequently, this
value was converted to (2000$) using the consumer price index. Median (mean) income was
estimated to be $57,325 ($66,496) annually.
El-2
-------
Table El-2. Demographic Characteristics of Freshwater Anglers8
N
Frequency
Gender
2267
62% Male
Race
2250
91% White
4% Black
2% Hispanic
2% Other
Education
2262
11 % Less than high school degree
34% High school degree/equivalent
55% Some college or more
Work status
2263
75% Employed
Geography
2237
23% Urban
37% Suburban
41% Rural
Region
2205
13% Northeast
33% South
31% Midwest
23% West
a In total, 2,384 respondents reported information on trip travel distance to a freshwater destination.
Note: Values may not add to 100 percent due to rounding.
Table El-3. Demographic Characteristics of Freshwater Anglers
Age
Household size
Persons <6 yrs
Persons >1 6 yrs
Weekly leisure time (hrs)
Family income (2000$)
N
2245
2255
2270
2254
2025
1851
Mean
38.4
3.1
0.3
2.2
27.7
66496
SD
14.5
1.5
0.7
0.9
23.9
57324
Min
16
1
1
0
0
8938
Max
92
10
5
7
168
208547
El-3
-------
Multivariate regression analysis was used to identify determinants of travel distance to
freshwater fishing sites. The dependent variable in this analysis was the miles traveled to the
most recent freshwater fishing site. The explanatory variables included several demographic and
geographic characteristics of the respondents.
Separate regressions were conducted for the full sample (1), single-day trips only (2), and
multiday trips only (3). The results are reported in Table El-4. Family income was estimated to
have a positive and highly significant effect in all three models. Dummy variables for urban and
suburban location were also found to have positive and highly significant effects in all models.
These results suggest that wealthier anglers and those living in or near metropolitan areas tend to
travel further to fishing sites, relative to less-wealthy anglers and those living in rural areas. In
models (1) and (2) dummy variables for the Midwest and West regions also had positive and
highly significant effects on trip travel distance, relative to the South region. The Northeast
region did not have a statistically significant effect on distance traveled. Education was
estimated to be positively and significantly related to distance traveled in the first and second
models. (Note that the respondent's level of education, recorded in the survey as a categorical
variable, was receded as a continuous variable for the regression analysis.) Neither age, race,
nor gender had significant effects (at a 5 percent level) on travel distance in any of the models.
£1.3 Summary Results Applied in the Population Centroid Approach
Given the high significance of geographic area and family income across the regressions,
nonparametric results (frequency distributions) were generated for four mutually exclusive
subgroups of respondents and five travel distance categories. The result are reported in
Table El-5. Respondents were categorized into the four following groups:
• Gl: family income >$50,000 (in 2000 dollars) and urban or suburban resident
• (N = 452 for single-day trips)
• (N = 649 for single- and multiday trips)
• G2: family income <$50,000 and urban or suburban resident
(N = 329 for single-day trips)
• (N = 417 for single- and multiday trips)
• G3: family income >$50,000 and rural resident
• (N = 295 for single-day trips
• (N = 376 for single- and multiday trips)
• G4: family income <$50,000 and rural resident
• (N = 309 for single-day trips)
• (N = 386 for single- and multiday trips)
El-4
-------
Table El-4. OLS Regression Results for Determinants of Reported Trip Travel Distance
(miles)
Variable Description
CONSTANT
AGE
GENDER
EDUC
MINORITY
FAMILY INCOME (log)
URBAN
SUBURBAN
NEAST
MIDWEST
WEST
(1)
Full Sample (both
single- and multiday
trips)
Coefficient t-stat
0.6966 1.54
0.0044 1.83*
0.0572 0.83
0.1729 2.48**
-0.0437 -0.36
0.187 4.41**
0.3491 3.95**
0.3422 4.48**
-0.0387 -0.36
0.3856 4.65**
0.6103 6.73**
R2 = 0.077
N= 1,798
(2)
Single-Day
Trips Only
Coefficient t-stat
1.7954 3.89**
0.0011 0.44
0.0173 0.25
0.1552 2.21**
0.0228 0.19
0.0827 1.92*
0.2799 3.12**
0.193 2.50**
-0.2549 -2.42**
0.1 1.21
0.3374 3.59**
R2 = 0.041
N= 1,360
(3)
Multiday
Trips Only
Coefficient t-stat
2.2493 3.26**
0.001 0.28
0.1446 1.39
0.128 1.22
-0.1391 -0.76
0.1759 2.78**
0.2121 1.62*
0.4298 3.67**
0.1525 0.89
0.4923 3.63**
0.3239 2.32**
R2 = 0.112
N = 434
** = significant at 5 percent level.
* = significant at 10 percent level.
El-5
-------
Table El-5. Travel Distance Frequencies by Demographic Group (Percentage in each
Distance Category)
(Gl) (G2)
High-Income and Low-Income and (G3) (G4)
Urban/Suburban Urban/Suburban High-Income and Low-Income and
Travel Distance (mi) Resident Resident Rural Resident Rural Resident
Single-day trips only (N =
N
Distance <10 mi
>10mito20mi
>20 mi to 50 mi
>50mito 100 mi
Distance >100 mi
Full sample (both single-
N
Distance <10 mi
>10mito20mi
>20mito50mi
>50mito 100 mi
Distance > 100 mi
= 1,385)
(N = 452)
23%
18%
31%
17%
11%
and multiday trips) (N
(N = 649)
16%
13%
24%
19%
27%
(N = 329)
32%
23%
20%
19%
6%
= 1,828)
(N = 417)
26%
18%
18%
19%
18%
(N = 295)
31%
22%
28%
14%
5%
(N = 376)
24%
18%
25%
16%
17%
(N = 309)
34%
24%
26%
11%
5%
(N = 386)
29%
21%
25%
14%
11%
These categories were selected because they match categories that can be easily
identified in Census data and because they split the sample into roughly similar group sizes.
Travel distance was categorized into ranges reported in the first column of Table El-5. The
results are consistent with those generated from the regression analysis. Among respondents on
single-day trips, the number that traveled longer distances (greater than 100 miles) increased
from the low-income rural cohort (5 percent) to the higher-income urban/suburban cohort (11
percent). The same pattern holds for those taking either a single- or multiday trip. The number
traveling longer distances more than doubled, from 11 percent among low-income rural
respondents to 27 percent among high-income urban/suburban respondents. These results
indicate higher-income urban/suburban anglers travel greater distances to freshwater destinations
than lower-income urban/suburban anglers and rural anglers.
As described in Section 10.1.2, the trip frequency estimates reported in Table El-5 for
the full sample were used in the population centroid approach to weight exposures to mercury in
fish according to distance from the Census Block Group centroid, income levels in the Block
Group, and whether the Block Group is predominantly rural or urban/suburban.
El-6
-------
APPENDIX E-2
METHODOLOGY FOR ESTIMATING FRESHWATER FISHING DAYS BY
WATERSHED
This appendix describes how data from the NSRE and NSFHWR were used in our
analysis to develop HUC-level approximations of fishing activity. These methods were
developed specifically to support the angler destination approach described in Section 10.1.2 for
estimating the size of exposed populations to mercury in freshwater fish.
As reported in Table 10-4, the NSFHWR provides estimates of the total number of lake-
and river-fishing days by state (in 2001 for both resident and nonresident anglers). These state-
level estimates are informative for estimating exposures through fish consumption, but they are
limited by a low degree of spatial resolution. Particularly within larger states, there is likely to
be considerable geographic variation in angler activity. Accounting for this variation should
improve estimates of mercury exposures by anglers. Unfortunately, data comparable to the
NSFHWR are not available for smaller or more homogeneous geographic units.
This appendix describes an empirically based method for distributing state-level
estimates of fishing activity from the NSFHWR to the eight-digit HUCs within their boundaries.
This distribution method takes into account the varying attributes across watersheds, which make
them more or less likely to attract anglers. Using this approach we estimated lake- and river-
fishing days for the 1,362 HUCs located in our 37-state study area. These HUCs range in size
from 0.02 to 7,939 square miles, with an average (median) of 1,353 (1,186) square miles.
E-2.1 Data
To supplement the NSFHWR, the primary source of data for this analysis is the NSRE
1994. Anglers responding to NSRE 1994 provided detailed information regarding their last
fishing trip, including where the fishing site was located and the number of times they visited the
site over the previous year. The fishing site locations have been geocoded to identify the HUC
corresponding to each trip. There are 2,111 identified HUCs in the continental United States,
which are grouped into 18 regional "hydrologic units." Figure E2-1 shows the boundaries for
hydrologic units in the United States.
E2-1
-------
Figure E2-1. U.S. Hydrologic Regions
Source: U.S. Geological Survey. Hydrologic Unit Maps. Last update February 17, 1999, accessed March 31, 1999.
.
In addition to detailed information about last fishing trip destinations, NSRE 1994
contains information on the type of waterbody visited on the last trip. Waterbodies are broken
down into four categories: lakes, rivers and streams, wetlands, and coastal areas. Of the 3,247
respondents who had fished in the continental United States in the previous year, 1,812 indicated
that their last trip was to a lake, 694 to a river or stream, 5 to a wetland, and 598 to a coastal area.
(Type of waterbody visited was not available for 141 responses.) Specific reach numbers, which
include the HUC identifier, were able to be identified for 1,758 of the 2,506 noncoastal (i.e.,
freshwater) destinations. For example, none of the 123 trips to hydroregion 17 (Pacific
Northwest) were able to be assigned a reach index, because the reach indexing system does not
include this region.
After selecting only the freshwater fishing destinations from the NSRE, we grouped them
by HUC. Using the freshwater last-trip destination information, we then constructed four "level-
of-use" indicators. For each HUC, the first and second indicators (RESP_COUNT_LAKE and
RESP_COUNT_RIV) are equal to the number of respondents in the survey who reported the
HUC as their last-trip destination for lake and river fishing, respectively. The third and fourth
indicators (TRIP_COUNT_LAKE and TRIP_COUNT_RIV) are equal to the total number of
trips taken by respondents in the previous year to their last visited site, when this site was in the
HUC. In all four cases, the observed trip frequency from the NSRE sample is taken as an
indicator of aggregate visitation to each HUC.
E2-2
-------
Table E2-1 reports frequency distributions for these four indicators across 1,892 HUCs
(excluding hydroregion 17). In all four cases, the number of HUCs in each level-of-use category
gradually declines as the level-of-use indicator increases, which is broadly consistent with a
Poisson/negative binomial distribution for this indicator.
To examine whether these level-of-use indicators are systematically related to other
HUC-level characteristic, we use GIS along with geographic, Census, and EPA Reach File 3
data to create several HUC-level variables. Summary statistics and descriptions of these
variables are provided in Table E2-2. These variables include continuous measures of the size of
each HUC, the number of reach miles in each HUC (lake and river), and populations within 25
and 50 miles of the HUC centroid. All of these measures are expected to have a positive effect
on level of use. We also included categorical variables to indicate whether a HUC is located
along the coast, whether it is located along one of the Great Lakes, and which hydroregion it
falls in.
To analyze the relationship between the HUC level-of-use indicators and these HUC-
level characteristics, we used negative binomial regression analysis. The negative binomial is a
variant of the Poisson count model. Using a simple Poisson regression model, we would
assume that the probability of observing a given count (nonnegative integer value th) at HUC h
for one of the lake or river level-of-use indicators (Tlh or T^) can be expressed as
fori=r, 1 (E2.1)
E2-3
-------
Table E2-1. Frequency Distributions for HUC Level-of Use Indicators
Lake Trip
RESP COUNT LAKE
0
1
2
3
4
5
6
7
8
9
10+
Total
TRIP COUNT LAKE
0
1
2
3
4
5
6
7
8
9
10
11-20
21-50
51-100
100+
Total
Counts
Number of HUCs
1310
317
112
64
39
20
12
4
4
2
8
1892
Number of HUCs
1312
93
66
57
37
43
34
17
20
20
15
80
76
15
7
1892
River Trip
RESP COUNT RIV
0
1
2
3
4
5
6
7
8
9
10+
Total
TRIP COUNT RIV
0
1
2
3
4
5
6
7
8
9
10
11-20
21-50
51-100
100+
Total
Counts
Number of HUCs
1554
234
61
32
6
4
0
1
0
0
0
1892
Number of HUCs
1555
52
47
33
26
21
27
10
7
2
11
48
33
13
7
1892
E2-4
-------
Table E2-2. Variable Definitions and Descriptive Statistics
Variables
LAKEMILES
RIVERMILES
POPO_25
POP25_50
HUCAREA
COASTHUC
GLAKEHUC
H1-H18
Definitions
Number of lake shore miles (RF3)
Number of river/stream miles (RF3)
Population within 25 miles of HUC centroid (2000
Census)
Population between 25 and 50 miles of HUC centroid
(2000 Census)
Surface area of the HUC (thousand square miles)
=1 if HUC is located adjacent to the coast
=1 if HUC is located adjacent to a Great Lake
Hydrologic region dummy variables
Mean
90.87
499.82
197350
566266
1.429
0.1317
0.0375
—
Standard
Deviation
153.17
542.92
593910
1121439
0.892
0.3382
0.1899
—
where Xh represents a vector of HUC-level characteristics and p( represents a corresponding
vector of coefficients. The Poisson specification assumes that the mean and variance of the
distribution are the same. In contrast, the negative binomial allows the variance to differ and
therefore includes an additional parameter (a), which is referred to as the overdispersion
parameter. The coefficient vectors are estimated using maximum likelihood methods and are
trips.
The regression results are reported in Table E2-3 for lake trips and Table E2-4 for river
1 For lake use (both respondent- and trip-level indicators), as expected, the number of lake
miles, size of proximate populations, geographic size, and Great Lakes proximity all have a
positive and statistically significant (at a 10 percent level) effect.2 Proximity to the coast has a
negative and significant effect. For both river-use indicators, the number of river miles, size of
population within 25 miles, and geographic size had positive and significant effects. For the
'The variable listed in the tables but not included in the regression results were found to be jointly not significantly
different from zero (at a 5 percent level) in previous regressions.
2The lake mile and population variables were included in logarithmic form because their distributions are heavily
skewed to the right.
E2-5
-------
Table £2-3. Estimated Determinants of HUC Level-of-Use Indicators for Lake Trips:
Negative Binomial Regressions
Dependent Variable:
Explanatory Variables
Log(LAKEMILES)
Log(POPO_25)
Log(POP25_50)
HUCAREA
COASTHUC
GLAKEHUC
HI
H2
H3
H4
H5
H6
H7
H8
H10
Hll
H12
H13
H14
H15
H16
H18
CONSTANT
alpha
N
Pseudo R2
RESP COUNT
LAKE
Standard
Coefficient Error
0.6339**
0.1517**
0.2317**
0.0979*
-0.6371**
1.5940**
—
—
-0.5832**
—
—
—
—
2.0139**
—
—
—
—
0.8110**
1.0561**
—
1.0318**
-8.1065**
0.7761
1884
0.1735
0.0426
0.0429
0.0535
0.0528
0.1291
0.1626
—
—
0.1699
—
—
—
—
0.2983
—
—
—
—
0.2377
0.308
—
0.205
0.4529
0.1039
TRIP COUNT
LAKE
Coefficient Standard Error
0.7506**
0.1730*
0.3290**
0.2584*
-1.2226**
2.3050**
1.1766*
1.2429*
0.9615*
0.9022
1.3917**
0.9678*
0.8245*
4.3948**
1.1988*
1.2947*
0.8991
1.6459*
1.1379*
2.0487*
1.0613
2.8459**
-9.8803
1.7149
1884
0.0737
0.0688
0.0735
0.0839
0.094
0.2531
0.4281
0.5855
0.5495
0.5495
0.5717
0.5349
0.5556
0.4998
0.6842
0.5169
0.5098
0.5606
0.5556
0.4991
0.6292
0.736
0.5889
1.0189
0.064
** Significant at 1 percent level.
* Significant at 10 percent level.
E2-6
-------
Table E2-4. Estimated Determinants of HUC Level-of-Use Indicators for River Trips:
Negative Binomial Regressions
Dependent Variable:
RESP COUNT RIV
Explanatory Variables
Log(RIVERMILES)
Log(POPO_25)
Log(POP25_50)
HUCAREA
COASTHUC
GLAKEHUC
HI
H2
H3
H4
H5
H6
H7
H8
H10
Hll
H12
H13
H14
HIS
H16
H18
CONSTANT
alpha
N
Pseudo R2
Coefficient
0.6265**
0.1371*
0.1519*
0.2977**
0.2354*
0.6069*
—
0.6038**
—
—
—
0.6076*
—
3.8641**
—
—
—
0.8902*
0.6823*
—
—
0.4614
-9.4317**
0.8416
1884
0.1335
Standard
Error
0.1123
0.6511
0.0802
0.073
0.1426
0.2665
—
0.1841
—
—
0.2854
—
0.7789
—
—
—
0.4855
0.3015
—
—
0.3222
0.8285
0.1947
TRIP COUNT RIV
Coefficient
0.5258**
0.4723**
—
0.4323*
—
—
—
0.7030*
0.5763*
—
0.6736*
1.0691*
—
3.3431**
—
—
-1.1334**
1.2390*
1.3690*
—
1.7103*
—
-8.9700**
2.5765
Standard Error
0.1483
0.0749
—
0.146
—
—
—
0.3049
0.33
—
0.3231
0.4612
—
1.0372
—
—
0.3553
0.762
0.6687
—
0.9127
—
1.0066
0.0762
1884
0.05
** Significant at 1 percent level.
* Significant at 10]
E2-7
-------
respondent-level indicator, the size of the population beyond 25 miles, as well as the Great Lake
and coastal dummy variables, were found to also have a positive and significant effect. In all
cases, the overdispersion parameter (a) was found to be significantly different from zero, which
indicates that the negative binomial model is preferable to a simple Poisson.
Using the results of the regressions reported in Tables E2-3 and E2-4, level-of-use
indicators were then predicted for each HUC (h) and waterbody type, based on differences in
HUC-level characteristics. For example, based on the regression results, the models predict
relatively higher angler activity in HUCs that have more river and lake miles, larger populations
within 25 and 50 miles, and larger surface area. For lakes (1) and rivers (r) respectively, we
A A.
refer to these predicted values as Tlh,Trh. From Eq. (E2.1) it follows that
Tih = E(Tlh\Xh) = exp(A^) for i= r, 1. (E2.2)
Table E2-5 provides a summary of the predicted level-of-use values.
Table E2-5. Predicted Level-of-Use Indicators for HUCs in Study Area: Negative
Binomial Regression Model Predictions
Predicted Indicator
RESP_COUNT_LAKE f,
TRIP_COUNT_LAKE f,
RESP_COUNT_RIV fr
TRIP_COUNT_RIV fr
min
0.001
0
0
0
pi
0.01
0
0
0.01
p50
0.2788
1.2264
0.1507
0.8551
mean
0.6192
4.3204
0.241
2.2631
p99
4.08
38.1314
1.5347
17.8171
max
28.3667
578.688
4.4308
85.3717
The predicted-use indicators were then used as measures of the relative level of angler
activities across FfUCs within each state (s). Specifically, they were used to approximate the
percentage of state-level lake- and river-fishing days that occur in each FIUC in the state as
follows:
yv
T
J-
T
Prhs = V ^f • (E2'4)
hs rh
E2-8
-------
The denominators in each of these two expressions is the sum of predicted-use indicators across
HUCs in the state.3
For the analysis summarized in Section 1 1, the predicted level-of-use indicators from the
trip-based models (TRIP_COUNT_LAKE and TRIP_COUNT_RIV) were applied in Eq. (E2.3)
and (E2.4) to estimate HUC-level indicators. The two models produce similar results regarding
the spatial distribution of angler activity, but the trip-based model was selected because it
includes more information about the frequency of specific trip destinations. In the analysis,
these predicted values were truncated at the 99th percentile of their distribution (see Table E2-5)
to prevent a small number of outliers from skewing the results.
To estimate absolute levels of angler activity in each HUC, these estimated percentages
of fishing activity were combined with total state-level fishing estimates from the NSFHWR. In
effect, the state-level estimates of lake- and river- fishing days by residents and nonresidents (Dls
and Dn; see Table 10-4) were distributed to HUCs based on the estimated percentages. The
number of lake- and river-fishing days in each HUC (in 2001) were estimated as:
Alh=plhs*Dls (E2.5)
(E2.6)
For example, HUCs with a predicted level-of-use indicator of 3 were assumed to have
three times as many angler days as HUCs with a predicted indicator of 1 . The results of these
calculations are summarized in Section 10.1.2.
3For HUCs located in more than one state, the predicted-use indicators were distributed across states in proportion to
the percentage of HUC surface area located in each state.
E2-9
-------
|