842R04002
Benefits of Reducing Mercury in Saltwater Ecosystems
A Case Study
Final Report
for
Debora Martin and John Wilson
Work Assignment Managers
Office of Wetlands, Oceans, and Watersheds
Oceans and Coastal Protection Division
U S. Environmental Protection Agency
1200 Pennsylvania Avenue NW
Washington, D C 20460
by
Douglas Rae and Laura Graham
January 2004
-------
Table of Contents
Executive Summary ES-1
1 0 Introduction 1
2 0 Identification of Study Area 6
3 0 Mercury Deposition Scenarios 15
40 Human HeaMi Benefits 22
5 0 Benefits to Recreational Anglers 64
6 0 Benefits to Commercial Fishing 79
7 0 Research Needs 102
80 References 105
-------
List of Figures
Figure 1-1 - Benefits Linkages 2
Figure 2-1 - Geographical Area of Mercury "H otspot" 6
Figure 4-1 - Mix of Fish and Shellfish Species Consumed by Males in
Flonda. 28
Figure 4-2 - Mix of Fish and Shellfish Species Consumed by Females in
Flonda 28
Figure 4-3 - Adjustment in NHANES Distribution of Fish Meals per Month
for Flonda Consumers 31
Figure 5-1 - Consumers' and Producers' Surplus 69
Figure 6-1 - Marine Recreational Trips by Mode in 2002 83
-------
Executive Summary
Mercury alters aquatic ocean environments pnmanly as direct deposition or runoff Combustion
facilities are the primary sources of mercury emissions in the United States to aquatic
environments microorganisms transform mercury into its more toxic form, meftiylmercury,
which bioaccumulates and biomagnifies in marine fish and shellfish Mercury can affect human
health directly as a result of consuming fish, and it may also affect human welfare through the
effect offish consumption advisones on commercial and recreational fishing
This report is a case study that quantifies the benefits of reducing mercury concentrations in
marine fish and shellfish in the Southeast United States It evaluates die impacts on human
health and commercial and recreational fishing for a single coastal area The rationale for a case
study approach is to demonstrate the feasibility of quantifying these benefits by focusing
resources on a defined area Quantification of benefits relies on a damage function approach that
links changes in environmental quality to impacts on human health and welfare and then values
the impacts
In this case study, benefits of reducing U S mercury emissions between 30 and 100 % result in a
range of annual benefits in the Southeastern U S between about $600 million to over $2 billion
Reduced health nsks, especially highly uncertain cardiovasular nsks account for a large
percentage of these benefits Additional research is needed to reduce the uncertainty in these
estimates, refine the analysts, and apply the methodologies to different regions of the country
Study Area
This study area comprises the South Atlantic coast from North Carolina to northern Flonda
EPA's Mercury m Marine Life database contains over 3,000 samples of marine fish and shellfish
caught in South Atlantic inshore and offshore waters, and these samples provide a reasonable
characterization of mercury concentrations in over 100 species offish and shellfish harvested by
commercial and recreational fishermen Included in the study area are the states of North
Carolina, South Carolina, Georgia, and northeastern Flonda. The population of tins area is about
22 5 million with about 8 3 million living within 100 miles of the coast
Mercury Reduction Scenarios
Current mercury emissions, deposition, and fish concentrations represent a baseline
environmental quality From that baseline the study considered three mercury reduction
scenarios
Scenario 1 reduce U S mercury emissions by 30 percent,
Scenano 2 eliminate all U S mercury emissions, and
Scenano 3 reduce mercury concentrations in fish to EPA's 0 30 parts per million (ppm)
criterion value as an average for all species categones, with Ihe exception of the "top
predator" category
A fundamental assumption of this analysis is that changes in emissions result in proportional
changes in mercury concentratioas in fish This lineanty assumption facilitates the analysis at
ES-1
-------
the nsk of oversimplifying the complexities of mercury fate and transport modeling in a
saltwater environment Regional air quality models of mercury emissions and deposition
indicate that U S sources account for about 40 percent of total deposition to coastal locations in
the South Atlantic study area but for less than 2 percent of emissions to the global pool of
airborne mercury Table ES-1 summarizes the impact of U S mercury reduction scenarios on
overall mercury deposition to the study area and on deposition to distant areas attributed to the
global pool
Table ES-1
Reduction in Mercury Deposition Resulting from Reductions in U.S. Emissions
Mercury Reduction
Scenario 1
Scenario 2
Scenario 3
U.S. Emissions
Reduction
30%
100%
51%
Reduction in Deposition
to Study Area
12%
40%
20%
Reduction in
Global Deposition
05%
1 6%
08%
Note The U S share of total deposition to South Atlantic waters is about 40 percent
A second major assumption of this analysts is that changes in emissions do not require decades
to translate into reductions in marine fish concentrations Oceans and soils act as reservoirs of
mercury that may buffer the effect of changes in anthropogenic emissions, but the rate of flux in
and out of these reservoirs is poorly understood The benefit estimates summarized below reflect
the full impact of changes in emissions and ignore the time penod necessary to achieve the full
impact If subsequent research finds that changes in emissions require lengthy periods to yield
the full impact, it would be appropriate to discount the benefits accordingly
Impact of Mercury on Human Health
Valuing the benefits of avoiding health endpoints starts with estimation of baseline exposures
from consuming fish The National Health and Nutrition Examination Survey (NHANES) of
nearly 10,000 respondents in 1999-2000 provides national estimates offish consumption for 26
species offish and shellfish over a 30-day recall penod In addition, a Florida survey on
consumption offish determined that coastal populations in the Southeast consume about 85
percent more fish and shellfish than the national average NHANES was not designed to provide
a representative sample of coastal fish consumption, and the Flonda study provides a basis for
estimating higher levels of consumption for study area populations in the coastal zone
Combining data on the number of meals offish consumed by species, estimates of portion size,
and data on mercury concentrations by species yields estimates of daily mercury intake
Baseline estimates of median daily mercury intake amount to 717 micrograms per day (ug/day)
for pregnant women and 917 (ug/day) for adult men, age 40-59 NHANES data are available at
the individual level, and estimating daily mercury intake for each respondent yields the
distnbution of mercury intake by percentile With dail> mercury intake as an input a one-
compartment pharmacokinetic model generates estimates of mercury concentrations in blood and
hair
ES-2
-------
Applying the lineanty assumption, the analysis calculates the reduction in mercury
concentrations for each species offish and, based on the mix of species consumed, the
corresponding changes in mercury intake and exposure The impact of mercury reduction
scenanos is highly sensitive to the proportion of consumption attributed to locally caught fish
versus fish harvested in distant waters Since prevailing winds blow U S mercury emissions
eastward, the analysis credited fish harvested from the South Atlantic and other eastern locations
with the full impact of reductions in U S emissions For fish harvested elsewhere the analysis
credits only the small, indirect impact of the reduction in U S emissions on global deposition
Chronic mercury exposure from consuming fish is associated with neurological symptoms of
mercury poisoning in adults and neurodevelopmental deficiencies in developing fetuses and
hypertension and cardiovascular disease in children and adults Dose-response functions exist
for mercury poisoning (paraesthesia), neurodevelopmental effects on fetuses, hypertension in
children age 7, and cardiovascular and coronary heart disease in adult men age 42-60
Neurodevelopmental effects include a number of neuromotor and neuropsychological endpoints,
including loss of IQ Mercury poisoning episodes and neurodevelopmental effects on fetuses
have been the subjects of extensive review by medical and epidemiological researchers, and
there is less uncertainty in applying dose-response functions for these health endpoints
However, the hypertension and cardiovascular dose-response functions denve from single
studies of foreign populations that have not been replicated in the United States Nor have they
been subject to the same level of peer review as the neurological endpoints Thus, considerable
uncertainty attaches to results of applying the dose-response functions for hypertension, non- fatal
acute myocardial infarction (heart attack), and all cause mortality Table ES-2 summarizes the
annual impacts on the vanous health endpoints for each of the three mercury reduction scenanos
Table ES-2
Annual Impact of Mercury Exposure Reduction on Health Endpoints
Mercury
Reduction
Scenano 1
Scenano 2
Scenano 3
Paraesthesia
(cases
avoided)
0
0
0
IQ
(points
saved)
7,999
26,617
] 3,798
IQ<70
(cases
avoided)
8
27
14
Hypertension
(cases
avoided)
0
2,277
947
Non-fatal
heart attacks
(cases
avoided)
73
247
121
All cause
mortality
(cases
avoided)
80
272
133
Note The dose-response function for 10 points saved is applied without a threshold
To value the health effects avoided that result from reducing mercury exposure the analysis
relied primarily on estimates of cost of illness, foregone earnings, and willingness to pay to avoid
pre-mature death However, values based on costs of illness and foregone earnings
underestimate willingness to pay, since these metrics do not include costs of pain and suffering,
side effects of medication, changes in lifestyle, and other effects
The benefits of avoiding neurodevelopmental impacts on the fetus rely on quantifying the value
of IQ points saved The value of IQ is measured in terms of foregone earnings Additional costs
of special education accrue to children whose IQ falls below 70, which is the threshold that
ES-3
-------
defines mentally handicapped These values range from nearly $80 million ($2003) per year in
scenario 1 to over $260 million ($2003) per year in scenario 2
The cardiovascular endpoints include hypertension in children and non- fatal acute myocardial
infarction (heart attacks) and all cause mortality in adult men To quantify the benefits of
avoiding hypertension in children the analysis relates reductions in children's blood pressure to
decreased cases of hypertension and values those cases based on the annualized present value of
costs of treatment and foregone earnings through age 30 The benefits of avoiding heart attacks
reflect the costs of medical treatment and follow-up over apenod of five years plus lost earnings
The value of avoiding hypertension and non-fatal heart attacks ranges from about $6 million
($2003) per year in scenario 1 to over $20 million ($2003) per year in scenario 2
For all cause mortality, however, the analysis applies estimates based on a willingness-to-pay
(WTP) measure (rather than ihe cost of illness and lost income measure) Current EPA
guidelines recommend a value of $6 7 million ($2003) per statistical life saved EPA has
employed this WTP value in previous Regulatory Impact Analyses, but there is considerable
uncertainty and debate about the appropriate value (or range of values) for use in the context of
environmental damages Consequently, there is additional uncertainty in applying this WTP
estimate to the results of an unreplicated dose-response function The value of avoiding pre-
mature death associated with exposure to mercury ranges from about $0 5 billion ($2003) per
year in scenario 1 to about $ 1 8 billion ($2003) per year in scenario 2 These highly uncertain
benefit estimates are useful in providing guidance in pnontang research to resolve these
uncertainties
Table ES-3 summarizes the annual benefits of avoiding mercury health effects for all endpoints
Table ES-3
Annual Benefits of Reducing Mercury Exposure
(Smillions $2003/year)
Reduction
Scenario 1
Reduction
Scenario 2
Reduction
Scenario 3
Benefits of Avoiding Mercury Health Endpoints
IQ Points
Saved
$780
$2595
$1345
IQ<70
$05
$16
$08
Non-fatal
Heart Attacks
$6 1
$207
$102
All Cause
Mortality
$5346
$1,8200
$8909
Child
Hypertension
$0
$07
$03
Total
$6192
$2,1025
$1,0367
Impact of Mercury on Commercial Fishing
Increased awareness of the health impacts of mercury may cause consumers to adjust their
purchases of fish and fish products U S Food and Drug Administration issued a fish
consumption advisory in January 2001 for four species offish king mackerel, swordfish, sharks,
and dlefish All four species are found in the study area. In addition, state heallh departments in
ES-4
-------
each of the four states in die study area have also issued advisories for some of these species If
consumers respond to these advisories by adjusting their purchases of some or all species of fish,
the shift in demand will result in changes in economic surplus, which measure economic
benefits Removing or reducing the stnngency of a fish consumption advisory could result in an
increase in demand with a consequent increase in benefits
To value such changes it is necessary to measure the shift in demand attnbutable to mercury, but
a review of Ihe economic literatute did not uncover any studies that measure the effect offish
consumption advisones on consumer demand In the absence of guidance from the literature, the
analysis attempted to quantify the impact of mercury fish consumption advisones on demand for
two species king mackerel and swordfish Using South Atlantic landings data from 1975
through 2002 and a time-series, econometric approach, the analysis modeled per capita
consumption as a function of speues pnce, apnce index of meat substitutes, per capita income,
and a dummy to capture the shift in demand attnbutable to mercury fish consumption advisones
In both the king mackerel and swordfish regressions the coefficient on the vanable that measured
the impact of the fish consumption advisory dummy was not significantly different from zero
with reasonable confidence Consequently, it was not possible to estimate any benefits to
commercial fishenes from reducing mercury concentrations in fish.
However, m 2004 the Centers for Disease Control (CDC) plans to release data from the 2001 and
2002 National Health and Nutrition Examination Survey A comparison of fish consumption
before and after issuance of FDA s mercury fish consumption advisones may provide evidence
of an impact on consumer demand from which to estimate benefits
Impact of Mercury on Recreational Fishing
Fish consumption advisones also .affect the behavior of recreational anglers Studies have shown
that a high percentage of saltwater anglers are aware of advisones, and many respond to
advisones by adjusting their choices of species, fishing mode, and site The recreational fishing
literature provides some estimates of losses in angler willingness to pay that result from fish
consumption advisones However, mercury fish consumption advisones are species-specific
over broad areas, whereas the on!) available studies that measure the impact of fish consumption
advisones are area-specific over many species
hi the absence of readily applicable studies the analysis employs a benefits transfer approach
Data collected by National Marine Fishenes Service indicate there are about 18 million saltwater
angler-tnps per year with detail bj shore, boat, and charter The data include information on
travel and time costs, which permil the estimation of random utility models that estimate the
value of access to species or species groups Several such studies provide estimates of per trip
values for coastal/migratory pelagic species in the Southeast This group includes king
mackerel, swordfish, sharks, and other species with elevated mercury levels A study using data
from the most recent 1997 Marine Recreational Fish Statistical Survey (MRFSS) yielded an
estimate of about $4 64 ($2003) per boat-tnp for access to coastal/migratory pelagic species in
the study area. Values of access to these species for shore and charter tnps derive from the
estimate for boat-tnps
ES-5
-------
To answer the question of how anglers respond to mercury fish consumption advisories die
analysis relies on a contingent valuation study of angler responses to hypothetical changes in fish
consumption advisones in the polluted area of Green Bay, an arm of Lake Michigan (Breffle et
al, 1999) This study provides estimates of utility both for catching fish and for reductions in the
stringency of various levels offish consumption advisones for yellow perch, trout-salmon,
walleye, and smallmouth bass Anglers were willing to pay $21 71 ($1998) to avoid 'do not eat'
for three of the four species groups in favor of 'unlimited' consumption The consumption value
amounted to about 46 percent of the combined value of catching and consuming fish Assuming
this ratio is applicable to saltwater anglers, it is possible to assess the impact of changes in fish
consumption advisones on the value of saltwater recreational fishing tops
To apply the Green Bay ratio requires translating the reductions in mercury levels for
coastal/migratory pelagic species into changes in fish consumption advisones U S Food and
Drug Administration uses a cntenon value of 1 0 ppm, but EPA and olher state health agencies
use different values Thus, there is considerable uncertainty in how regulators would adjust their
advisones as mercury concentrations in fish declined The analysis made Ihe following
assumptions
Scenano 1 would represent a change from 'do not eat' to' 1 meal per month',
Scenano 2 would represent a change from 'do not eat' to' 1 meal per week', and
Scenano 3 would represent a change from 'do not eat' to '2 meals per month'
Applying these changes to Ihe per tnp value of access to migratory/pelagic species in the South
Atlantic and the number of fishing hips per year yields benefit estimates that range from about
$3 6 million ($2003) per year for scenario 1 to $5 9 million ($2003) per year for scenano 2
ES-6
-------
1.0 Introduction
Mercury is a toxic pollutant that is widely dispersed in the ambient environment In high
concentrations it is poisonous, and EPA has regulated some point source emissions and
discharges to protect human health and the environment It persists in the environment
and bioaccumulates and biomagmfies in methylated form in freshwater and marine
aquatic environments Mounting evidence from low dose epiderrnological studies shows
that consumption offish and shellfish in large quantities can pose health nsks from
mercury ingest on As a result, U S Food and Drug Administration (FDA) and many
states have issued health-based fish consumption advisories (FCAs) for both freshwater
and saltwater species
Concern over the health and environmental nsks of mercury prompted Congress to
commission a comprehensive study of mercury in the mid-1990s (LJ S EPA, 1997) In
2000 the National Research Council (2000) under the auspices of the National Academy
of Sciences undertook a comprehensive review of the health effects research on mercury
and concluded that fetal nemodevelopmental impairment may be occurring at low doses
of methylmercury In response to the growing consensus that chronic, low-dose mercury
exposure may pose health nsks EPA is working with FDA to review and expand the
current fish consumption advisones
Most non-occupational exposure to mercury results from consuming fish, and marine fish
and shellfish compnse a high percentage of fish consumption. EPA's Office of
Wetlands, Oceans, and Watersheds (OWOW) together with the National Oceanographic
and Atmosphenc Administration (NOAA) is responsible for protecting marine life In
the past year OWOW completed a research effort that compiled a dataset of mercury
concentrations in marine fish and shellfish (Cunningham et al, 2003) The Office also
sponsored an economic study that identified the gaps in data, information, and knowledge
that need to be filled in order to quantify the benefits of reducing mercury deposition to
saltwater environments (Rae, 2002)
This study builds on the pncr studies to quantify the health, recreational, and commercial
fishing benefits of reducing rnercuy concentrations in a case study Since air deposition
modeling identified a mercuiy deposition "hotspot" in the South Atlantic, that area was
chosen for this case study Tins economic benefits case stud> employs a damage
function approach to assess the impacts and quantify the benefits of reducing mercury
concentrations in the case study area.
As shown in Figure 1-1, measuring benefits or damages requires the ability to quantify
the following links
mercury emissions to changes in environmental quality
changes in environmental quality to changes in social and environmental impacts,
changes in social and environmental impacts to changes in well-being or welfare,
aggregation of changes in welfare across individuals and over time
-------
The link from emissions to changes in environmental quality requires Ihe ability to model
mercury emissions, dispersion, deposition, methylation, and aquatic uptake The link
from changes in fish concentrations to environmental and social impacts requires the
identification and quantification of human responses and the resulting impacts on human
health, recreational fishing, and commercial fishing The link from changes in impacts
on humans to changes in welfare requires quantification of benefits or damages using
methodologies that measure willingness to pay Estimation of total benefits or damages
requires the aggregation of benefits or damages across individuals or households and over
ume
Figure 1-1 - Benefits Linkages
Mercwy Emissions
Changes to Welt-being: Basefifs or Damages
Aoross lodividimls and Over Time
-------
This report focuses on collecting and evaluating information to quantify and value the
health, recreational, and commercial impacts that result from changes m mercury
concentrations in saltwater species The case study approach limits the complexity of the
analysis by reducing the geographic extent of the impacts and b> limiting the number of
species affected The mercury reduction scenanos also employ simplifying lineanty
assumptions to bndge significant gaps m scientific knowledge of the fate and transport of
mercury in marine environments (boxes 1 and 2 in Figure 1-1) These simplifications
allow the report to focus on the impacts, changes in economic well-being, and
aggregation of benefits/damages (boxes 3,4 and 5 in Figure 1-1)
Definitions
Mercury in its various forms LS widespread throughout the environment, and humans
receive exposure from air and water as well as food Mercury in ambient air and water is
in elemental and ionic forms, but exposures from these sources are negligible Food,
primarily fish and shellfish, is the primary source of mercury exposure, and the form of
mercury in fish is predominantly methylinercury Consequently, when the report refers
to mercury intake or exposure from fish, it is referring to methylmercury exposure
Mercury concentrations are usually expressed in micrograms of mercury per gram offish
(u,g/g) Since a rmcrogram is one millionth of a gram, this unit is equivalent to parts per
million (ppm)
Key Assumptions and Uncertainties
This quantification of benefits relies on two key simplifying assumptions
reductions in mercury concentrations in fish are proportional to reductions in
deposition, and
the length of time between changes in deposition and steady state changes in
mercury concentrations in fish is not so long that discounting would significantly
reduce the values
The first assumption allows the analysis to bndge the gap in knowledge of mercury fate
and transport in the manne environment The assumption of lineanty between deposition
and fish concentrations means that a 25 percent reduction in mercury deposition
eventually results in a 25 percent reduction in mercury concentration in fish EPA's
Mercury Maps project (Cocca, 2002) concluded that lineanty is a reasonable assumption
The second assumption involves the time penod to achieve the full impact of any
deposition changes, and this is highly uncertain If a large reservoir of mercury exists in
terrestrial and ocean sediments, it may require years, or even decades, for any reductions
in mercury deposition to achieve steady state conditions as measured by concentrations in
marine fish A long delay until the full realization of the benefits of reductions in
mercury deposition would require discounting the benefits stream. Preliminary reports
from the Metalhcus project, which uses mercury isotopes to detect the time until mercury
-------
appears in fish tissues, suggest a response to mercury deposition measured in years rather
than decades, but this result applies to a freshwater lake environment (U S EPA, 2003)
Results of monitoring in the Flonda Everglades confirmed that reducing mercuiy inputs
into the aquatic environment has resulted in 60 percent reductions in mercury
concentrations in fish and 70 percent m birds over 10 years (Flonda DEP, 2003) This
case study sidesteps the issue over the time stream of benefits and the impact of
discounting by reporting annual benefits that reflect Ihe achievement of steady state
conditions However, readers should keep in mind that it may require years to achieve
steady state conditions, and a lengthy penod to achieve steady state conditions could
reduce the present value of that benefit stream significantly
Quantifying benefits requires combining estimates of population, exposure, dose-
response, and valuation All have uncertainties attached to them. The discussion
references the underlying uncertainties and attempts to provide some indication of the
direction of bias However, there is no effort to quantify the range of the uncertainty in
the final benefit estimates A more rigorous analysis would include high, low, and
midpoint estimates as an indication of the range of uncertainly Since this case study is
the first attempt to value the benefits of reducing mercury in all species of marine fish
and shellfish, its limited resources are better applied to bridging the gaps in data and
information Consequently, the analysis discusses uncertainties underlying the vanous
components of each estimate, but presents only one, midpoint estimate of benefits for
each impact
Caveats
This report is a case study It quantifies Ihe impact of reducing mercury concentrations in
marine fish on populations in the study area and values the changes in health,
recreational, and commercial fisheries It is often tempting for readers to extrapolate
these results to other areas, but readers should resist this temptation for the following
reasons
the mix of fish species consumed in the South Atlantic study area is unlikely to be
the same as in other areas,
mercury concentrations for fish of the same name (e g tuna) may represent
multiple species (albacore, blackfin, yellowfin, little tunny, etc) whose mercury
concentrations may vary greatly by species and from area to area,
the amount of fish consumed vanes between coastal and non- coastal areas and
may also vary from coast to coast and north to south
Use of this Report
This report is one step along the path (hat leads to Ihe capability of valuing the benefits of
reduced mercury emissions It is not the first step on this path nor is it the last It relies
on the progress made by scientists and epidemiologists in understanding the fate and
transport of mercury and its effects on fetal development and cardiovascular function As
an economic benefit study it builds on pioneering analysis of Ihe health nsks of
-------
consuming seafood (Camngton and Bolger, 2002) and pioneering analysis of economic
benefits of consumption advisones for striped bass (Jakusetal 2002) This report
extends those studies to include additional health risks and additional fish and shellfish
species, but applies them to a limited area of the Southeast As noted above, there
analysis makes use of a number of assumptions to badge gaps in data and knowledge,
and considerable uncertainty underlies some of the benefit estimates presented in this
report
Next steps along the path to ^ aluing the benefits of reducing mercury emissions include
filling in the gaps in knowledge and understanding This report provides information that
will allow EPA and other organizations to identify the gaps and uncertainties in the
benefits calculus and to prioritize research to fill those gaps An important next step is to
replicate this case study in another area to determine how the benefits compare
Report Organization
There are six sections to this report Section 2 0 identifies a geographic area for study
and summarizes the population affected and the baseline mercury concentrations
Section 3 0 explains the mercury reduction scenanos and compares the reductions in
mercury concentrations to the baseline Section 4 0 estimates the economic benefits to
human health from reduced exposure to mercury Section 5 0 estimates demand and
supply functions for commercial catch and attempts to quantify the changes in consumer
surplus that result from reducing mercury concentrations in commercial species Section
6 0 uses a benefits transfer approach to quantify the benefits to recreational anglers from
reduced mercury concentrations in fish and possible elimination offish consumption
advisones Section 7 0 discusses the research needed to eliminate key gaps in our
knowledge and reduce the uncertainties for future benefits estimation
-------
2.0 Geographical Area for the Case Study
Mercury modeling using the Regional Modeling System for Aerosols and Deposition
(REMS AD) determined that deposition is high m an offshore area of die South Atlantic
This mercuiy deposition "hotspot" occurs due to significant rainfall in the offshore area
that washes out large amounts of mercuiy emitted by powerplants and other sources The
presence of this mercury "hotspot" provides a geographical focal point for assessing the
benefits of reducing mercury concentrations in marine fish and shellfish The "hotspot"
compnses an area of about 130,000 square kilometers (50,000 square miles) of offshore
waters along the continental shelf from about 30 to 33 degrees north latitude and about 74
to 78 degrees west longitudel This area extends roughly from North Carolina to
northern Flonda
The Atlantic coast from North Carolina to Flonda is the focus of this study Importantly,
this area includes a sufficient number of samples in the Mercury in Marine Life (MML)
database (Cunningham et al, 2003) to provide a reliable characterization of the mercury
concentrations for most species of commercial or recreational significance Figure 2-1
shows the South Atlantic coast, including the "hotspof area of elevated mercury
deposition
Figure 2-1
Geographical Area of Mercury "Hotspot"
! I degree ot Latitude is equivalent to about 111 km (69 miles) 1 degree of Longitude at 30 N Latitude is
equivalent to about 97 km (60 miles)
-------
2.1 Study Area Population
A wide variety of fish species inhabit the inshore and offshore waters from North
Carolina to northern Florida Some of these fish are species of commercial importance
that appear in restaurants and supermarkets across the study area from coastal locations to
hundreds of miles inland Other fish species that inhabit or migrate through the area are
the target of recreational anglers, whose trips may originate hundreds of miles away
Consequently, the mercury deposition that impacts these marine waters affects both
coastal and inland inhabitants of four states North Carolina, South Carolina, Georgia,
and Florida
Florida poses a problem because it borders both the South Atlantic and the Gulf of
Mexico, but the case study only considers fish caught in Atlantic waters In evaluating
die impacts on human health die analysis limits the impacts to populations in northeastern
Florida from the Georgia border to Daytona Beach (Volusia County) and omits counties
on the Gulf side of the peninsula However, in considenng commercial and recreational
fishing the study area includes all of Atlantic Florida, since National Marine Fisheries
Service harvest totals for the South Atlantic include all of Atlantic Flonda
Table 2-1 summarizes the populations of these states from the 2000 Census for each of
four zones coastal counties, non-coastal counties within 100 miles of the coast, counties
101-200 miles from the coast and counties 201-300 miles from the coast Measurements
from the county centroid to the coast determined the assignment of zones within the study
area (see Appendix 2-A) Hie population in counties that border the coast in the study
area amounts to about 3 4 million An additional 4 9 million inhabitants live within 100
miles of the coast About 8 2 million live from 100-200 miles, and about 6 0 million live
from 200-300 miles
Table 2-1
Study Area Population by Zone
Zone Honda Georgia North Carolina South Carolina Total
ZoneO Coastal 1,593,666 439,154 656,508 742,274 3,431,602
Zonel XMOO 678,923 1,167,330 1,751,878 1,302,543 4,900,674
Zone II >100-200 0 1,806,955 4,491,970 1,900,980 8,199,905
Zone III >200-300 0 4,773,014 1,148,957 66,215 5,988,186
Total 2,2^2,589 8,186,453 8,049,313 4,012,012 22,520,367
Source: Bureau of the Census, 2000 Census of Population data.
2.2 Baseline Mercury Concentrations in Fish
In 2002 EPA compiled a master database of available federal, regional, and state
databases of mercury concentrations in fish and shellfish This Mercury in Manne Life
(MML) database (Cunningham et al, 2003) includes over 14,600 samples of marine
species harvested from U S estuanne and marine waters since 1990 The MML database
overlays sampling data with GIS information, which allows identification of the location
-------
of each species sampled Since these data comprise multiple sources, each with different
objectives, coverage by area and species is uneven For example, the database includes
about 7,200 samples for the Gulf of Mexico, about 4,850 samples for the Atlantic, but
only about 2,600 samples for the Pacific Coverage for many recreational species is quite
extensive, since most are caught close to shore where sampling stations are most
prevalent For some commercial species, however, coverage is quite limited For
example, there are no data on mercury concentrations for swordfish
The most extensive data on mercury concentrations in fish exist for species that inhabit
the Gulf of Mexico and the South Atlantic This is because the state of Flonda has a very
comprehensive sampling program. The study area extends from Atlantic Flonda to North
Carolina, and for this area the MML provides sufficient samples for most commercially-
and recreationally important species to provide reliable estimates of mean mercury
concentrations
2.2.1 Mercury Concentrations by Species and Species Grouping
Table 2-2 summarizes the mercury concentration data by species from the Mercury in
Marine Life Database The table classifies the various species into five categories
according to their position in the food web and their preferred diet The categones
include pnmary, secondary, tertiary, and quaternary consumers and top predators
Higher category species feed on lower category species with the largest sharks at the top
of the manne food chain This classification loosely follows the organization of marine
food web models that balance energy or biomass flows in the food web (see Okey, 2001)
The database contains nearly 3,500 samples of mercury concentrations for 109 species,
but the table combines some individual species within the groups into larger species
aggregations For example, the grouping, sharks, combines a number of shark species
within a category The mean mercury concentration for all samples is 0 37 ppm, but the
range is from under 0 1 ppm for most pnmary and secondary consumers to over 6 0 ppm
for the largest sharks in the top predator category Means for pnmary, secondary, and
tertiary consumers are under 0 3 ppm The tertiary consumer category averages 0 22
ppm, but bonefish, small sharks, and grunts all exceed 0 3 ppm The quaternary
consumer category mean is 0 38 ppm with about half the species in excess of 0 3 ppm,
including leatherjacket and little tunny which exceed 1 0 ppm. In the top predator
category the mean is 0 74 ppm. As expected, sharks are at the top of this group with a
group mean in excess of 1 0 ppm Readers should note that a few species of small sharks
are in the tertiary and quaternary consumer categones and that the mean mercury
concentration for all sharks is less than 1 ppm
-------
Table 2-2
Mercury Concentiations in the Southeast Atlantic by Group and Species
Spec las Grouo
Nanrw
Count?
Maant
Min
Max
ForaarY 18P
Q.18Q,
~&4&P,
0 189
IPiaflsh
^iwnassanj L~
11 P.14O< 9*14Q*.
30 - 0 135, Q-O.2Q,
nnn \-Q*iw v»sw>vs^VvA-jfrM^tvwvwt-j^wOMvJ.ftsM. ft'**
P.14Q.
-0410
P,J,25l
Jfff^vwvwwwM^fJffvM
Qray tnaaertish
.a^sa
Sana dfum
Groun Summnrv
rfrA V J I Ai^ V MHtTS Qi. 1 wQ
21 A.a.oat, """Q.QIT.
Z4, »OQa4, ,,,0,933. P.I go.
5] 0 0564 P 043^. _ P.073
124 JL22B ojuiL 1.400.
Q«aleraacy consumers \
:; Lemon sharfc.
2* 1 45P*
44 1.141,,..
_ , . -1Q}. 9^75^
1 2PP!
P.79S1.
1-7DQ
Qrowpers
,Twoas
,,
Banded. rwdderTish
ijaalatx
Grouo Summarv
1^00
1^092,
0 741
jojoaBi cuiai
O 02O^ 68OO1
All Groups
Cunmnatiam >
3499! O 371 O OO1- 8 9OO
2OO2 Mercurv in Marine Life Database
-------
2.2.2 Mercury Concentrations in Commercially Important Species
The National Marine Fishenes Service (NMFS) collects data on commercial fisheries
landings, both by weight and value, for the South Atlantic NMFS' South Atlantic area
compnses Atlantic landings for the states of North Carolina, South Carolina, Georgia,
and Florida. NMFS' commercial landings data include a few freshwater species, such as
catfishes/bullheads and ulapias, but most of the landings are marine species
Table 2-3 combines tfie quantity and value of catch, by species, for the most important
commercial species or species groups with data on mercury concentrations from the
Mercury in Marine Life database Where species are combined into species groups,
sharks for example, the table reports a weighted mercury concentration for all species in
that group with weights based on number of samples The cntenon for inclusion as a
commercially important species is annual landings in excess of 100 metnc tons The
table highlights those species with mean mercury concentrations over 0 3 ppm using light
blue to highlight species between 0 3 and 0 499 ppm, pink to highlight species between
0 5 and 0 999 ppm, and red to highlight species in excess of 1 0 ppm.
By annual value, the most important commercial species in the South Atlantic are shnmp,
blue crab, and clams, which together amount to about $107 million and comprise 61
percent of the total dollar value Shnmp and clams are quite low in mercury with
concentrations below 012 ppm, but blue crabs are surprisingly high with concentrations
estimated at 0 536 ppm.
The highest annual value marine finfish are flatfish/flounders, which account for over
$12 million, but are low in mercury concentration Next most important are king/cero
mackerel, snappers, and swordfish, which are all worth more than $3 million annually
King mackerel are high in mercury with average concentrations of 0 95 ppm Snappers
as a group average below 0 3 ppm, but mercury concentrations for red snapper exceed 0 8
ppm. The database has no mercury samples for swordfish, but mercury sampling done by
FDA suggests values of about 1 0 ppm Other marine species with annual landings
greater than $ 1 million and mercury concentrations greater than 0 3 ppm include tunas,
sharks, Spanish mackerel, tilefish, weakfish, and groupers Species of lesser value with
mercury concentrations in excess of 0 3 ppm include amberjacks, other jacks, bluefish,
seatrout and porgies
10
-------
Table 2-3
Contmei cial Landings in the South Atlantic for 2003
Species
MARINE FINFISH
AMBERMCK
BALLYHOO
BASS STRIPED
QtUEFISH
CROAKER, ATLANTIC
DOLPHIN
DRUMS
FINFISHES.UNC GENERAL
FLATFISH/FLOUNDERS
GOOSEFISH
GROUPERS
JACKS '
KING WHITING
MACKEREL, KH«5 AND C£«0
MACK£R£L> SPANISH
MENHADEN ATLANTIC
MOJARRAS
MULLETS
PERCH
PORGIES
SEA BASS, BLACK
SEAJROUT
SHADS
SNAPPERS
SPOT
Metric Tons
2404
1266
3007
4,0949
4,557 7
1972
2569
1,7497
3,471 5
1262
Sf?$
' ';' *uur
4390
, t^K&if
U2D8
31,4469
1195
1 ,848 8
1630
1730
2936
1S&6
1,7342
*7045
1,0081
Pounds
829,779
279017
662931
2,39^769
10,047,952
434,673
566,312
3,857,422
7,653,186
278,226
1t141,$*Q
967*912
2.441:892
2,691,341
69,327867
263,414
4075922
359321
381,3640
647,334
, 221665
3,823,151
1,553,384
2,222,485
Value ($)
405,048
203,409
836,817
660,363
2,604,527
604,782
240,323
1,715,717
12,366,102
314,238
2J%&$&Si&
maso
959,456
4,095,299
1,691,412
2 476 761
319,955
2110,447
263,062
328,672 0
1,092,594
3K5J8&
1,060,437
3^626,973
935,761
Mercury Concentration(ppm)
MML FDA
0442
0168
Q.m
0118
0071
0286
0152
&j59f
,6^
6^^
03&4
0070
0117
0068
0268
0187
0122
050?
0015
0251
0111
WEAKFiSH
MARINE SHELLFISH
CLAMS/BIVALVES
CRAB, OTHER
LOBSTER, CARIBBEAN SPINY
OYSTER, EASTERN
SCALLOPS
SHELLFISH
SHRIMPS
FRESHWATER FISH
CATFISHES & BULLHEADS
FINFISHES, FW, OTHER
TILAPIAS
TOTAL
2352
1860
2493
1860
7978
11,9355
2,239 5
1934
1,7698
97.774 2
518396
409,972
549,513
410,023
1,758,756
26,313,025
4937,193
426,340
3,901,645
215.552.963
887,733
1,919,191
2,148,006
1,377,108
454,730
56,654,860
3,743,305
397,421
1,227,823
174.904.137
0056
0072
0130
0050
0070
ND
Note Criteria for species inclusion landings over 100 metric tons
Source National Marine Fishene s Service, website httpJ/www st nmfs gov/pls/vvebpls/MF_ANNllAL_LANDINGS RESULTS
U S Food and Drug Administration, website http //www cfsan fda goW-frf/sea-meha html
11
-------
2.2.3 Mercury Concentrations in Recreationally Important Species
NMFS also collects data on the recreational harvest through its Marine Recreational
Fisheries Statistics Survey (MRFSS) The MRFSS includes both intercept and telephone
surveys of marine recreational anglers, and NMFS compiles data on the species catch by
weight and numbers for shore, boat and charter modes MRFSS data for the South
Atlantic area includes Atlantic landings for the states of North Carolina, South Carolina,
Georgia, and Florida
Table 2-4 summarizes the data on recreational catch in the South Atlantic area for major
species or species groups over all recreational fishing modes The data include only fish
landed and exclude fish that are caught and released (MRFSS type B2) The cntenon for
inclusion of species in the table is annual landings in excess of 100 metric tons As
above, the data on mercury concentrations are from the MML database
The highest tonnage landings for recreational species in the South Atlantic area are
dolphin fish (mahi mahi), tunas, king mackerel, drums, Spanish mackerel, and bluefish,
all exceed 750 metric tons On a per species basis, recreational fishermen account for
more than 50 percent of the total commercial and recreational landings for amberjacks,
barracudas, dolphin fish, drums, other jacks, king mackerel, porgies (especially
sheepshead), sea trout, snappers, stnped bass, tunas, and weakfish Major recreational
species with mercury concentrations above 0 5 ppm include king mackerel, tunas, sharks,
other jacks, groupers and sea trout Also above 0 3 ppm are amberjacks, bluefish,
Spanish mackerel, weakfish, and barracuda
12
-------
Table 2-4
Recreational Landings in the South Atlantic for 2002
Species Metric Tons Mercury Concentratton(ppm)
Study Area
MARINE FINFISH
AMSERvJACK 39&6 8.4
BARRACUDAS 2743 05
BASS, STRIPLD 330 1 0 1
BLUEFISH 831 8 CM
CROAKER 1815 01
DOLPHIN 4,8159 00
DRUMS 1,6501 02
FLATFISH/FLOUNDERS 433 7 0 1
2540 83
KINGFISH 5836 0091
MACKEREL, SPANISH 90&5
MULLETS 2676
POROIES 7422
SEA BASS, BLACK 1 50 2
SBCIHOUT - *'« ^
SNAPPERS 704 5
SPOT 5152
WEAKFISH ' 874 0305
TOTAL 19,4381
Note Criteria for species inclusion landings over 100 metric tons
Source National Marine Fisheries Service, website
http //www st nmfs gov/pls/webpls/MR_CATCH_SNAPSHOT RESULTS
13
-------
Appendix 2~A
Map Counties by Distance to the Coast
This mapping exercise assigned counties in Florida, Georgia, North Carolina and South
Carolina into four zones according to Iheir distance from the coast The process relied on
Arc View 3 2 GIS The program defines county boundaries using the U S Census 2000
County shapefile It also defines the coast using the U S EPA's Reach File 1 (Rfl) GIS
coverage to select those reaches of type "C" coastal or continental shoreline reach
segments
The object of the mapping was to assign counties to one of four zones
zone 0 bordenng on the coast,
zone 1 not bordenng the coast but within 100 miles,
zone 2 101-200 miles, and
zone3 201-300 miles
First, those counties that bordered the coastal reaches were identified Then, ArcView's
Buffer tool drew zonal boundanes at 100, 200, and 300 miles from the coastal reaches
The cntena for assigning counties within a buffer zone was whether their centers fell
within the zone, excepting those already identified as coastal counties
14
-------
3.0 Mercury Reduction Scenarios
Mercury loadings to ocean environments and the resultant uptake by marine fish and
shellfish originate pnmanly as airborne emissions Some emissions are directly
deposited to marine environments through precipitation In addition, runoff from soils
into streams and nvers contributes significant amounts of mercury to bays and estuaries
where smaller organisms begin the process of bioaccumulation EPA and states have
taken initial steps to control mercury at vanous combustion facilities, and EPA is
currently considering additional regulatory actions to reduce mercury on a national basis
EPA and states also set standards for mercury in aquatic environments Where toxic
pollution of water results in a failure to meet water quality standards, EPA requires states
to perform a Total Maximum Daily Load (TMDL) analysis under the Clean Water Act
Where this failure to meet standards is due to pollution from non-point sources, including
airborne deposition, determining how to achieve compliance requires an understanding of
the complex linkages between air emissions, deposition, and fate and transport in aquatic
environments In the case of mercury TMDLs one objective of TMDL analysis is to
combine atmosphenc modeling of deposition with fate and transport modeling so as to
understand how changes in air emissions translate into concentrations in aquatic
environments and uptake by 1he fish, birds, and mammals that inhabit those
environments This analysis of air sources is then combined with other sources to allocate
future emissions to levels thai will achieve water quality standards Flonda Department
of Environment (2003) recently conducted a TMDL analysis of mercury in the Flonda
Everglades to quantify the linkages between emissions, deposition and mercury
concentrations in fish and birds in the Everglades
EPA has issued a water quality criterion for mercury lhat is expressed as a concentration
in fish tissue EPA's mercury cntenon value is 0 3 ppm, and the Agency has asked states
to use this cntenon as a guide in developing standards for mercury
Scientists have made considerable progress in understanding the complex linkages
between air emissions of mercury, patterns of deposition, runoff, methylation, fish
uptake, and biomagmficabon in the food chain However, much of our current
knowledge comes from studying freshwater environments The marine environment
differs from freshwater lake systems in two fundamental ways First, the physical and
chemical characteristics of estuanes, bays, and oceans differ from freshwater systems
Second, ocean environments are more complex than closed freshwater lake systems
Saltwater has much higher pH than freshwater, especially in lakes subject to acidification,
and mercury methylation rates vary with pH and other chemical characteristics Tidal
washing, storms, and the diverse water conditions at different depths also affect the rates
of methylation Oceans also act as sinks for mercury and the reactions that remobihze
mercury from waters and sediments are poorly understood Saltwater estuanes, bays, and
deeper waters provide habitat for a nch variety of fish and other organisms that compnse
a complex food web Many of these fish migrate or travel great distances As a
consequence, the uptake of mercury for these species depends on their feeding behavior
and the concentrations of mercury found in prey fish over great distances
15
-------
Lacking validated models of mercury fate and transport in tfie marine environment, this
study relies on relationships demonstrated for freshwater environments EPA's Mercury
Maps project determined lhat the relationship between mercury deposition in lake
environments and the resulting mercury concentrations in fish is approximately linear
(Cocca, 2002) This study proceeds on ihe assumption that changes in mercury
deposition causes proportional changes in fish concentrations
3.1 Proposed Mercury Reduction Scenanos
This case study evaluates three different mercury reduction scenarios Scenano 1
represents a 30 percent reduction in mercury emissions, which is within the range of
reductions considered by EPA in vanous mercury regulatory strategies Scenano 2
considers the impact of eliminating all U S mercury emissions And Scenano 3
evaluates the impact of achieving EPA's cntenon level of mercury for most categories of
fish
Scenano 1 reduce U S mercury emissions by 30 percent,
Scenano 2 eliminate all U S mercury emissions, and
Scenano 3 reduce mercury concentrations in fish to 0 30 ppm as an average for
all species categories, with the exception of the "top predator" category
3.2 Reductions in U.S. Mercury Emissions
The three approaches result in very different percent reductions in mercury deposition
Calculation of the overall reduction in mercury deposition to the manne environment
requires calculating the amount of deposition from U S andnon-US (including natural)
sources and evaluating the impact of a change in U S deposition on the total
EPA's REMSAD model calculates the contnbution to mercury deposition at a particular
spatial location from vanous sources Modeling results for the gnd squares of maximum
deposition m each of the four states in the study area shows that U S sources account for
37 to 68 percent of total deposition, as shown in Table 3-1
16
-------
Table 3-1
U.&. Contribution to Mercury Deposition
(grams/square kilometer)
Mercury Deposition
State Total U.S. Sources Non-U.S. Sources U.S. Percent
North Carolina 781 45 3 32 8 58 00%
South Carolina 859 533 326 6200%
Georgia 516 193 325 3700%
Florida 1256 854 402 6800%
Total 3412 2033 1381 5951%
Note: Mercury deposition is m grams/km2 for the gnd square with
maximum deposition
Source: Atkinson, D, 2003, EPA, REMSAD modeling output
These gnd squares of maximum deposition occur near the coast However, an "average"
deposition for the coastal areas including terrestrial watersheds and shallower waters in
bays and estuaries is about 40 percent (Atkinson, personal communication, 10/27/03)
3.2.1 Scenario 1
Reducing mercury deposition from U S sources by 30 percent will result in a net
reduction in deposition to the coastal zone of about 12 percent (=30%* 4)
3.2.2 Scenano 2
Eliminating all mercury deposition from U S sources will result in a net reduction in
deposition to the coastal zone of about 40 percent (=100%* 4)
3.23 Scenano 3
Determining the reduction HI mercury deposition that is required to achieve a criterion
level of mercury in fish of 0 30 ppm depends on how one defines compliance with that
level The regulatory problem is that it is not realistic to achieve a 0 3 ppm mercury
concentration for all species of marine fish Many marine predators, sharks for example,
live for many decades and bioaccumulate mercury to very high concentrations In most
large sharks mercury concentrations exceed 1 0 ppm, and concentrations are significantly
higher in the largest sharks Even eliminating all U S mercury emissions would not be
sufficient to reduce mercury (oncentrations to 0 3 ppm for any species whose current
concentrations exceed 0 5 ppm
As defined above, Scenano 3 requires that species categones for primary, secondary,
tertiary, and quaternary consumers, as measured by average concentrations, attain the 0 3
ppm criterion level Table 3-2 shows the mean and species maximum concentrations in
17
-------
each species group and the percent reduction required to attain the 0 3 ppm criterion
level
Table 3-2
Mean and Maximum Mercury Concentrations in Each Species Category
(ppm)
Category
Primary consumers
Secondary consumers
Tertiary consumers
Quaternary consumers
Top Predators
Grand Total
Source:
Species
Count
Category Species
Mean
Max
Percent Reduction
to 0.3 ppm
3
8
31
31
36
109
0057
0065
0229
0377
0741
0371
0211
0152
0570
1450
4578
4578
from Mean
00%
00%
00%
20 4%
59 5%
19 1%
from Max
00%
00%
47 4%
79 3%
93 4%
93 4%
Cunningham et al 2003, Mercury in Marine Life Database
Table 3-2 shows that all species in the primary and secondary consumer categories are
below the 0 3 ppm level For tertiary consumers the category mean is below 0 3,
although some species categonzed as tertiary consumers do exceed 0 3 ppm The
category mean for quaternary consumers is 0 377 ppm, and to meet the 0 3 ppm cntenon
will require an overall reduction in total deposition of about 20 4 percent For top
predators (he category mean is 0 741 ppm, and it would require a reduction of nearly 60
percent in overall deposition to reduce this category's mean concentration to 0 3 ppm.
However, to achieve the 0 3 ppm cntenon at the species (rather than the category mean)
level would require a reduction in total deposition of about 79 percent for the quaternary
consumer category and 93 percent for the top predator category Deposition reductions
of this magnitude would require significant cutbacks in mercury emissions beyond U S
borders This is because the maximum deposition reduction achievable from elimination
of all U S mercury emissions is about 40 percent, which assuming linearity will only
achieve a 40 percent reduction in mercury concentrations in fish
Given that the definition of compliance is achievement of the 0 3 ppm cntenon
concentration as a mean for all but the top predator species categones, then compliance is
possible for species categones one tfirough four Pnmary, secondary, and tertiary
consumer groups are already in compliance, but it will require a reduction in the U S
emissions of about 51 0 percent (= 2047 4) to achieve compliance for quaternary
consumers A 51 0 percent reduction in U S emissions that reduces overall deposition by
20 4 percent will cause the mean mercury concertration for top predators to decline from
0 74 ppm to 0 59 ppm (=(1-0 204)*0 741)
3.2.4 Summary
The three mercury reduction scenanos result in three different percent reductions in
mercury concentrations in fish As summarized below in Table 3-3, Scenano 1 reduces
18
-------
U S emissions by 30 percenl and achieves a 12 percent reduction in total deposition
Scenario 2 eliminates all U S deposition, which achieves a reduction of about 40 percent
in total deposition, and Scenario 3 requires a 51 percent reduction in U S emissions,
which results in a 20 4 percent reduction in total deposition Table 3-3 summarizes the
impact of the three mercury r sduction scenanos for fish harvested from the South
Atlantic
Table 3-3
Summary of Deposition Reduction Scenarios for the South Atlantic
Scenario
1
2
3
Reduction in U.S. Deposition
30 0 percent
1000 percent
51 0 percent
Reduction in Total Deposition
120 percent
40 0 percent
20 4 percent
Note Total deposition includes mercury deposition from both U S and global sources
3.3 Impact on Global Deposition
U S mercury emissions also contnbute to the global pool of airborne mercury and affect
the rate of global deposition Gobal emissions include both anthropogenic and natural
sources Mercury emitted in elemental form behaves as a gas and may remain in the
atmosphere for periods measured in months or years The natural mercury cycle is
imperfectly understood, but oceans, sediments, soils and vegetation may all act as
mercury sinks that release mercury back into the atmosphere under vanous conditions In
addition, natural events such as volcanic activity and wildfires, also contnbute to
mercury emissions
There is considerable uncertainty about the global emissions inventoiy of mercury Best
estimates from the Global Emissions Inventory Activity project are that current
anthropogenic sources account for about 2,000 metnc tons each year (GEIA, 2003)
Pacyna et al (2003) estimated total current emissions from anthropogenic sources in
1995 at 2,427 metnc tons About 75 percent are from combustion of fossil fuels,
especially coal Using a box model of mercury cycling between atmosphere and ocean
on a global scale, Lamborg el al (2002) estimated that anthropogenic emissions,
including current emissions and recycled emissions from mercury in soils and oceans,
account for about 60 percent of the total and natural emissions account for the remaining
40 percent GEIA (2003) reported estimates of mercury emissions from terrestrial and
oceanic sources that included both natural and recycled anthropogenic components, but
emphasized that all estimates of natural emissions are highly uncertain
Oceans 770-2,300 metnc tons,
Volcanoes 20-447 metnc tons,
Soils 500-3,200
Vegetation 850-2,000 metnc tons,
Fires up to 100 metnc tons
19
-------
Total annual emissions from soils and oceans, including recycled anthropogenic
emissions, probably account for about 4,000 metric tons Best guesses are that current
anthropogenic emissions, recycled anthropogenic emissions from soils and oceans, and
natural emissions each account for roughly one-ted of total annual emissions, but there
is considerable uncertainty about estimates of global emissions (Bullock, personal
communication, 10/27/03)
Annual U S emissions of mercury amount to about 125 metnc tons wilh about 30 percent
being deposited within the U S and about 70 percent escaping to the global pool
Assuming other anthropogenic sources behave similarly, the roughly 2,000 metnc tons of
anthropogenic emissions cortnbutes about 1,400 metnc tons to the global pool Table 3-
4 estimates that the total global pool of airborne mercury amounts to about 5,400 metnc
tons and that the U S contnbutes about 87 5 metnc tons, which is less than 1 percent of
that total
Table 3-5 analyzes the impact of reducing U S mercury emissions on the global pool and
resulting global deposition Scenano 1, a 30 percent reduction in U S emissions, results
in a reduction of 26 3 metnc tons (=87 5 metnc tons*0 30), which amounts to a reduction
of about 0 5 percent (=26 3/5400) in global deposition Scenarios 2 and 3, which
correspond to 100 percent and 51 percent reductions in U S emissions, result in
reductions of about 87 5 (=87 5 metnc tons*l 0) and 44 6 (=87 5 metnc tons*0 510)
metnc tons, respectively These changes correspond to reductions of 1 6 (=87 5/5400)
and 0 83 percent (=44 6/5400) in global deposition
Table 3-t
Global Pool of Airborne Mercury
Emissions Source
Anthropogenic
US
Deposition
Global
Natural
Oceans
Soils
Vegetation
Volcanoes
Fires
Total
Notes
Emissions Global Pool
(metric tons) (metric tons)
2000
125
375
875
21408047
770 2 300
500-3,200
850 2 000
20-447
0-100
Percent of Total
(%)
1 400 25
4 000 74
5400 100
93%
07%
00%
1) Midpoint of natural emissions range based on estimate provided by R Bullock EPA
2) Based on U S models, total anthopogenic emissions are assumed to behave
the same as U S emissions with 70 percent entering the global pool
Source
Seigneur C Global Emissions Inventory Activity Review Mercury
website http //www geiacenter org/reviews/mercury html
20
-------
Table 3-5
Impact of U.S. Mercury Reduction Scenarios on Global Deposition
Emissions Source
Total
U S Contnbution
Baseline
Scenano 1
Scenario 2
Scenano 3
Source
Mercury
Global Pool
(metric tons)
5,400
875
61 3
00
429
Reduction
Atributable to U S
(metric tons)
263
875
446
Reduction as a Pet
of Global Total
(%)
0 49%
1 62%
0 83%
Calculated based on Tables 3-3 and 3-4
21
-------
4.0 Quantify the Benefits to Human Health of Reducing Mercury Exposure from
Consuming Saltwater Fish in me Study Area
Mercury exposure in low doses from consumption of fish may potentially affect the
heallh of newboms exposed in utero, young children, and adults Calculation of the
benefits of reducing mercury exposure follows the standard damage function approach
laid out in Section 1 0 Section 4 1 applies the mercury concentrations in fish from the
study area and elsewhere to the distribution offish consumption to estimate a baseline
mercury exposure Section 4 2 applies the mercury reduction scenarios to the mercury
concentrations in fish to estimate the change in mercury exposure Section 4 3 applies
dose-response functions for specific health endpoints to estimate numbers of cases of
specific impairments or illnesses that would be avoided as a result of the reductions in
mercury exposure Section 4 4 applies economic estimates of the values of avoiding
these adverse health effects to estimate the corresponding benefits Section 4 5 discusses
the assumptions and uncertainties that underlie 1he analysis
The approach to estimating mercury exposure in Sections 4 1 and 4 2 follows pioneering
work done by Camngton and Bolger (2002) in their analysis of the exposure to
methylmercury from seafood consumption However, this analysis makes use of richer
datasets for fish consumption from the National Health and Nutrition Examination
Survey (NHANES) and mercury concentrations in fish from the Mercury in Marine Life
(MML) database
White the damage function approach is straightforward, estimation of each of the
components is not Each section requires evaluating and applying data and information
from multiple sources Because of gaps in data and knowledge some of the calculations
require simplifications and assumptions The exposition of the methodology attempts to
clarify exactly how each calculation was performed and to list all assumptions embedded
in the analysis In addition, each section discusses the uncertainties and biases that
underlie the estimates of exposure, nsks of impairment or illness, and economic benefits
4.1 Baseline Distribution of Mercury Intake and Exposure
Mercury occurs in air, water, and food, and humans receive some exposure to mercury
from all three sources Excepting occupational exposures, food, primarily fish, is the
largest source of exposure to mercury Mercury found in fish is mostly methylmercury,
which is more toxic than the ionic (Hg++) or elemental (Hg°) forms Moreover, the ionic,
elemental, and methylmercury forms of mercury affect different organs in the body
Hence, there is some question as to whether exposures to different forms of mercury are
additive Since the exposure from air and water sources is typically very small, the
analysis does not include them, and mercury exposures in this analysis reflect only
methylmercury intake from fish and shellfish
The approach to estimating mercury exposure from consumption offish is to
1 estimate the consumption of fish, by species, in the study area,
2 estimate the concentration of mercury in each species offish, and
22
-------
3 thereby denve a baseline distribution of mercuiy consumpton among study area
residents
The analysis uses estimates offish consumption from two sources the National Health
and Nutrition Examination Survey (NHANES) and a study offish and shellfish
consumption in Florida (Degner et al, 1994) released by 1he Florida Department of
Environmental Protection
4.1.1 National Health and Nutrition Examination Survey 1999-2000
The National Center for Health Statistics (NCHS), Division of Health Examination
Statistics (DHES), part of the Centers for Disease Control and Prevention (CDC), has
conducted a series of national health and nutntion examination surveys (NHANES) since
tiie early 1960's In 1999 NHANES began collecting data on fish consumption by key
species The survey includes both a home interview and a medical examination, which
requires respondents to go to mobile examination centers (MECs)
One of the survey objectives is to establish and maintain a national probability sample of
baseline information on health and nutritional status Part of the nutntional information
in the 1999 NHANES includes questions on the number offish meals consumed in the
last 30 days The 30-day recall question provides a more accurate estimate of quantities
offish consumed than 24-hour, 3-day, or 7-day recall periods used in other surveys, since
many people who eat fish may go weeks between meals offish or shellfish However,
asking respondents to recall details on all fish meals eaten over a 30-day penod is a
demanding task and is likely 10 result in some inaccuracies
The NHANES target population is the civilian, non-institutionalized U S population, and
the survey design is a stratified, multistage probability sample The primary sampling
units (PSUs) are counties or small groups of contiguous counties The 1999 NHANES
sampling design selected a sample of 5,000 respondents each year from 15 PSUs In
2003 NCHS released results for the first two years of surveying, 1999 and 2000 These
data included interviews for 9,965 respondents Due to a delay in implementing the
survey in 1999 sampling covered only 12 PSUs in that year For the combined 1999-
2000 survey there were 27 PSUs, but one large PSU entered the survey in both years, and
the 1999-2000 survey has 26 distinct PSUs As a result of the limited number of PSUs
sampled, NCHS has not included geographic information, and the results of the survey
are applicable only to a national sample Persons contacted at NCHS indicated that one
of the PSUs sampled was in the study area, and several of the PSUs sampled were in
coastal locations (Schober, personal communication, 9/29/03) This is significant
because marine fish consumption is known to be greater in coastal and near coastal
locations The NHANES sampling design did not specifically address the issue of
consumption differences between coastal and non-coastal populations Consequently,
extrapolating national NHANES results to the coastal Southeast is likely to result in some
downward bias in seafood consumption This study attempts to avoid this bias by
applying estimates from the Florida survey to populations near the coast
23
-------
Hie nutritional interview disaggregates the estimates offish and shellfish meals into
further detail for marine and freshwater finfish and shellfish species The survey asks
two questions did you eat [species] in the last thirty days and [if yes] how many times
did you eat it Altogether, the 1999 NHANES provides detailed information on the
consumption of 8 species of shellfish and 19 species/products of finfish plus an
'unknown' category for both shellfish and finfish Shellfish species include clams, crabs,
crayfish (assumed to be freshwater), lobsters, mussels, oysters, scallops, shnmp, and
other shellfish Finfish include 14 manne categories breaded fish products, tuna, cod,
flatfish haddock, mackerel, perch, pollock, porgy, salmon, sardines, sea bass, shark,
swordfish, five freshwater species bass, catfish, trout, walleye, and pike, and an other
category This level of detail makes it possible to generate estimates of mercury intake
by applying species-specific concentrations of mercury to species-specific estimates of
consumption
NHANES data include age and gender so that it is straightforward to estimate the
distribution offish and shellfish meals consumed for vanous population groups The
NHANES data include sampling weights for each respondent, and it is necessary to apply
the interview weights to the data on fish meals consumed to yield accurate estimates
Table 4-1 summanzes the mean, median, and 90th, 95th, 99th, and maximum consumption
levels of marine shellfish, freshwater shellfish, marine finfish, and freshwater finfish for
women of childbeanng age (15-44), adult males (15 and over), adult females (45 and
over), and children (5-14) Analysis of the data on fish consumption revealed that the
mean consumption of shellfish amounted to 4 06 meals per month for men, 3 53 meals
per month for women of childbeanng age 4 08 meals per month for older women, and
1 67 meals per month for children However, half the population eats few, if any, meals
of fish, and the median number of meals per month is only 2 for adults and 0 for children
At the tail of the distnbution 95th percentile consumption amounts to over 13 meals per
month for adults, but only 7 meals per month for children As expected, seafood,
including both marine fish and shellfish, accounts for over 75 percent of total meals of
fish in all population groups
24
-------
Table 4-1
Consumption of Fish and Shellfish Meals by Selected Population Groups
Gender/Age Marine Shellfish Marine Fish Freshwater Fish Unknown Total
Wbmen 1S44
N
100% Max
99%
95%
90%
75% Q3
50% Median
Mean
Men 15 and over
N
100% Max
99%
95%
90%
75% Q3
50% Median
Mean
\Abmen 45 and over
N
100% Max
99%
95%
90%
75% Q3
50% Median
Mean
1,885
50
16
5
4
2
0
139
3,009
42
15
7
4
2
0
154
1,433
38
10
6
4
2
0
128
1,885
35
16
8
5
2
0
1 68
3,009
52
13
7
5
2
0
175
1,433
48
14
7
5
3
1
203
1,885
28
3
2
1
0
0
025
3,009
28
6
2
1
0
0
039
1,433
30
5
2
1
0
0
043
1,885
30
4
1
0
0
0
021
3,009
90
6
2
1
0
0
038
1,433
28
5
2
1
0
0
033
1,885
80
23
13
9
4
2
353
3,009
92
25
14
11
6
2
406
1,433
72
26
13
9
6
2
408
Children under 15
N
100% Max
99%
95%
90%
75% 03
50% Median
Mean
Source Calculated based on 1999-2000 data from National Health and Nutrition Exammaton Survey
3,921
29
6
3
2
0
0
049
3,921
20
9
4
3
1
0
089
3,921
18
2
1
0
0
0
012
3,921
21
4
1
0
0
0
016
3,921
36
13
7
5
2
0
167
25
-------
4.1.2 Florida Fish Consumption Survey
In 1994 the Flonda Department of Environmental Protection released a study offish and
shellfish consumption in Flonda (Degner et al, 1994) This study included three
different samples
1) a random statewide telephone survey of 8,000 Flonda households from all 67
counties,
2) a random telephone survey of 1,000 households of five counties where
mercury effluents from a paper mill may have increased mercury
concentrations in fish,
3) a random in-person survey of 500 food-stamp recipients
The statewide survey provided estimates offish and shellfish consumed by Flonda
adults, age 18 and over These results are of particular interest, since this survey provides
a basis for comparing fish and shellfish consumption in a coastal state within the study
area with the national estimates from NHANES
The statewide Flonda survey employed a telephone survey to elicit estimates offish and
shellfish consumption based on respondent recall for the previous 7 days The
researchers confirmed the accuracy of recall after 7 days in companson with diary
studies, but employed aided recall questions to enhance survey accuracy The
questionnaire mentioned six commonly eaten species offish and five species of shellfish
as memory cues to enhance recall of consumption
The questionnaire elicited information on how much was eaten at home and away from
home Interviewers first elicited information on consumption away from home in the last
7 days from a randomly selected adult (the adult with the next birthday) Then, they
asked to speak with the pnrnary meal preparer to elicit information on fish meals eaten in
the home The survey included detail on the portions consumed Instead of relying on
standard portions, this survey asked respondents to estimate fish portions by reference to
a "slice of sandwich bread" and shellfish portions (e g shnmp or scallops) by number
and size
Telephone interviewers made calls on a constant quota basis, about 168 per week, over
52 weeks from March 15, 1993 through March 13,1994 Since consumption of certain
types offish is seasonal, the one-year interview penod minimizes any bias that could
result from shorter survey penods
The statewide sample of 8,000 was 41 6 percent male and 58 4 percent female By age
the sample included age 18-24,107 percent, age 25-34,22 0 percent, age 35-49, 28 5
percent, and age 50 and over, 38 8 percent The researchers also collected and reported
results for ethnic groups and by income level
The authors reported seafood consumption on an annual basis Table 4-2 summanzes the
mean consumption of fish and shellfish, both at home and out of home, for Flonda males
and females, age 18 and over Flonda males consumed, on average, 41 6 g/day offish
26
-------
and 10 7 g/day of shellfish Florida females consumed slightly less, about 32 4 g/day of
fish and about 9 29 g/day of shellfish,
Table 4-2
Mean Consumption of Fish and Shellfish for Flonda Adults
Gender Category At Home Away from Home Total Total
(kg/year) (kg/year) (kg/year) (g/day)
Male
Finfish 742 775 1517 4156
Shellfish 118 274 392 1074
Total 860 1049 1909 5230
Female
Finfish 642 542 1184 3244
Shellfish 1 09 2 30 3 39 9 29
Total 750 772 1522 4170
Source Degner et al, 1994, Per Capita Fish and Shellfish Consumption in Flonda, Table 4
Additional analysis of the Flonda data (Porner et al, 1995) provided additional detail on
the distribution of fish and shellfish consumed by seafood groupings The researchers
combined species consumed into 9 groups freshwater bottom feeders, freshwater
crustaceans, freshwater predators, panfish, manne crustaceans, marine mollusks, manne
finfish, sharks, and processed fish, and reported means and percenules of consumption
for each species group This information provides insight into Ihe mix of species of
finfish and shellfish consumed in the Southeast
Figures 4-1 and 4-2 show the percentage of fish and shellfish consumed by species group
for males and females in Flonda. For both males and females, the largest consumption
category is manne finfish, thi> category compnses about 41 percent offish consumption
for males and 37 percent for females For Flonda adults this group compnses a wide
variety of species, including swordfish, tuna, cod, flounder, salmon, and haddock along
with more regional species, such as pompano, red drum, king mackerel, and sea trout
The second largest consumption category is processed finfish, which accounts for about
29 percent of consumption by males and about 32 percent for females Shellfish,
including both crustaceans (crab, lobster, shnmp) and mollusks (clams, mussels, oysters,
scallops), account for about 21 percent of consumption for males and about 22 percent
for females For Flonda consumers freshwater fish account for a very small percentage
offish consumption, about 8 percent for both males and females Freshwater bottom
feeders, primarily catfish, account for most of this consumption
27
-------
Figure 4-1
Mix of Fish and Shellfish Species Consumed by Males in Florida
Freshwater bottom feeders
6%
Freshwater crustaceans
0%
Freshwater predators
1%
Freshwater panflsh
Processed (Irtish
29%
Marine crustaceans
14%
Marine mollusks
7%
Marine fmfish
41%
Figure 4-2
Mix of Fish and Shellfish Species Consumed by Females in Flonda
Processed finfish
32%
Freshwater bottom feeders
5%
Freshwater crustaceans
0%
Freshwater predators
1%
Freshwater panfish
2%
Marine crustaceans
16%
Marine mollusks
6%
Marine finfish
37%
28
-------
4.1.3 Baseline Mercury Exposure Levels for Key Populations
Combining estimates offish consumption from toe NHANES and Flonda surveys with
toe mercury concentration data allows estimation of toe mercury exposure for each
individual Grouping individuals by sex and age allows estimation of means and
distributions for key population groups
Daily mercury intake is simply toe product of meals consumed of each species, portions
consumed per meal, and mercury concentrations for each species The following
equation calculates toe daily mercury intake from consumption of species s
^* H&,
where E^ = MeHg exposure to individual i from consuming species s in
micrograms per day (ug/day),
MS = Meals of species s in a 30 day period,
P, = Meal portions in grams for individual i, and
Hg, = Mercui} concentrations for species s m ug/g
Summing mercury intake over all species consumed yields a total mercury exposure
While the mercury intake calculation is relatively straightforward, apportioning the meals
reported for broad species categories reported in NHANES to species and subspecies
requires considerable judgment
This case study focuses on mercury exposure to South Atlantic area populations from
consuming marine fish Consumers in these four states have access to a wide vanety of
seafood harvested by U S fishermen and additional products imported from abroad The
calculation of mercury exposure from fish and shellfish consumption applies mercury
concentrations from the MML database for species harvested from the South Atlantic and
other areas Where MML data are unavailable the analysis relies on sampling done by
the US Food and Daig Administration (U S FDA, 2001)
A number of assumptions underlie toe assignment of mercury concentrations and the
resulting estimates of mercury exposure from fish and shellfish consumption.
Where the commeraal catch for a species harv ested from the South Atlantic area is
significant (>$100 million), the analysis assigns mercury concentrations based on
samples collected in toe South Atlantic study area on toe assumption that most of the
consumption of that species is from local sources
Where a species is not harvested locally, such as salmon, the analysis relies on
mercury concentration data for areas where toe species is harvested or from FDA
sampling data
* Where two or more subspecies are subsumed in a single consumption category, crab
for example, the analysis assumes toe local consumers eat mostly toe local species
29
-------
For example, blue crabs are the predominant crab species found tn the South Atlantic,
and the analysis assumes all crab consumption in the study area is blue crab
Where a single consumption category compnses two or more subspecies and the non-
local species is more common, as is the case with lobster, the analysis apportions
consumption between them based on relative amounts available for consumption
(landings plus imports minus exports) and applies mercury concentrations for each
subspecies
Where NHANES records consumption in an 'other' category, the analysis assumes
that in the study area this 'other' category would consist wholly of marine species and
computes a mercury concentration based on an average of 'other' species weighted by
landings
Where NHANES records consumption of freshwater fish, such as walleye and pike,
which are generally unavailable in the study area, the analysis adds the consumption
of these species to the 'other' fish category
Where NHANES estimates of tuna include both fresh/frozen and canned products and
estimates of mercury concentrations indicate a significant difference between these
products, the analysis uses data on landings, imports, exports, and canned tuna
production to estimate relative shares of each product and weights the mercury
concentrations by the relative amounts consumed
Where NHANES indicates consumption of breaded fish products that include
products sold in supermarkets and fast food restaurants, interviews with processors
revealed that nearly all of the breaded fish products sold in the U S consists of
Alaskan pollock with a small amount of imported fish, primarily cod from Norway or
hoki from New Zealand (Sasiela, personal communication, 10/3/03, Feener, personal
communication, 10/3/03)
NHANES did not collect data on portion sizes, which is an important determinant of fish
consumption, and this lack of information results in some uncertainty in estimating fish
consumption In general, the common wisdom is that men eat larger portions than
women, and adults eat more than children However, pregnant women may eat portions
equal in size to men In translating NHANES meals into mercury exposure this analysis
applies the following portion sizes
Adult men 8 ounces (0 5 Ib)
Pregnant women 8 ounces (0 5 Ib)
Adult women 6 4 ounces (0 4 Ib)
Children 5 3 ounces (0 33 Ib )
As noted above, the Florida survey estimates that Florida residents consume about 1 85
times the amounts of fish and shellfish recorded by NHANES in its national sample of
households Tnis analysis assumes the NHANES data reflect fish consumption
representative of study area populations who reside more than 100 miles from the coast
For stud> area populations within 100 miles of the coast the analysis adjusts NHANES
consumption to reflect the amounts reported by Flonda residents but retains the relative
mix of species reported in NHANES and adapted for the study area
30
-------
Adjusting the consumption mean requires adjusting the entire distribution of meals per 30
day period so that the mean of the distribution equals the desired value In the absence of
parameters to charactenze frilly the distribution of Flonda fish and shellfish consumption,
the analysis employs best professional judgment to redistribute the NHANES percentile
values to yield a higher meai This entails reducing the number of consumers who ate no
fish meals in a 30 day period and increasing the frequency of respondents who ate greater
numbers of fish meals in that period, as shown in Figure 4-3 The adjustments do not
shift the distribution any further to the nght, since the maximum number of meals
consumed in 30 days is about 90 for most adults and children (teenage boys excluded)
Figure 4-3
Adjustment in NHANES Distribution of Fish Meals per Month for Florida
Consumers
2000
1800
1600
1400
1200
1000
800
600
400
200
0
Ft
Mean = 36 meals/month
6 8 tO 12 M 16 IS 20 22 24 26 28 30 32 35 38 4t 44 46 51 53 55 59 72 92
Maals/month
Table 4-3 presents the distribution of baseline mercury exposure in micrograms per day
for four key population groups, pregnant women, men age 40-59, adult women, and adult
men In the coastal zone median methylmercury exposure amounts to 7 17 ug/day for
pregnant women, 551 u,g/day for adult women (15 and over), 9 17 u,g/day for men age
40 to 59, and 6 86 fig/day for adult men (15 and over) More than 100 miles from the
coast meihylmercury exposure is significantly less Median exposures amount to 2 69
fig/day for pregnant women, 1 69 fig/day for adult women, 3 28 ug/day for men age 40 to
59, and 1 94 u.g/day for adult men
31
-------
Table 4-3
Distribution of Baseline Mercury Exposure for Key Study Area Groups
Zones
0,1
II, HI
Source:
Midpoint
Median
Mean
Median
Mean
Pregnant Women
7 17
11 66
269
608
Adult Women
551
933
169
483
Men: 40-59
888
1278
328
682
Adult Men
682
11 37
1 94
605
Calculated by applying mercury concentrations in fish, by species, to
consumption, by species
The are a number of uncertainties associated with translating estimates offish
consumption into mercury exposure The most important concern the accuracy and
reliability of the mercury concentration data and the accuracy of the fish consumption
survey data Estimates of mercury concentrations depend on the number of samples,
which vary greatly by species To conform to NHANES' species categones it was
necessary to aggregate across species Due to limited samples for some species this
aggregation yielded estimates weighted by sample size, whereas weightings based on
relative amounts consumed or landed would better reflect mercury exposure
Accurate estimates of baseline mercury exposures also depend on accurate respondent
recall offish consumption data NHANES employed a 30-day period which provides a
more accurate estimate of fish consumption over extended penods, especially for
consumers who eat fish less regulariy, but the one-month recall penod makes heavy
demands on respondents' memories Additional uncertainty anses from the lack of
information in NHANES on portion size Also, the NHANES finfish categones create
some uncertainty in matching species consumed with mercury concentrations because
some species, such as bass, perch, trout, and catfish, apply to both freshwater and marine
species
The analysis relies on the Flonda fish consumption survey as an estimate offish
consumption in coastal populations This survey relies on 7-day recall, which is likely to
be more accurate than 30-day recall However, the 7-day consumption penod generates
less precise estimates of daily fish consumption for a similar sample size because of the
large percentage of people who eat no fish in a 7-day penod There is also some
uncertainty in the ad hoc procedure to adjust the distribution of fish consumption from
NHANES to approximate the distribution of meals offish per 30-day penod for a coastal
population with a significantly higher mean, but the magnitude and direction of any bias
are unknown
4.2 Changes in Mercury Exposure Attributable to the Mercury Reduction Scenarios
Measuring changes in mercury exposure that result from reductions in mercury emissions
and deposition requires two key steps The first is to apply the change in deposition to
32
-------
the baseline mercury concentrations in fish from the South Atlantic and other areas The
second is to determine what percentage of consumption is attnbutable to species
harvested in various locatiors Since the prevailing winds mostly blow pollution
eastward, U S reductions in mercury emissions will pnmanly affect species harvested in
the Atlantic and Gulf areas This also includes any locally-caught freshwater species
Fish and shellfish harvested in Pacific or foreign waters are onl> indirectly affected by
the U S emissions reductions through changes in the U S contnbution to the global
mercury pool
This second step requires estimating the amount of each species available for
consumption from domestic landings minus exports plus imports and calculating the
percentage attnbutable to different harvest areas The analysis groups Atlantic and Gulf
landings and estimates their relative share of consumption of each species
4.2.1 Landings, Imports, and Exports
The NHANES and Flonda surveys provide information on the quantity and mix of
species consumed, but the area of harvest is equally important This is because any
reduction in U S mercury emissions has direct impact on uptake by fish in Atiarfcc and
Gulf locations To address this reality the analysis computes the total amount of each
species available to consumers and the share of that total from Atlantic/Gulf and
Pacific/Import locations NMFS data on landings, exports, and imports are available by
species and provide the data inputs for this calculation The total amount available is
given by the following equation
Cs = L\QS + Lps - X + Is,
where Ca = Consumption for species s,
LAGS = Landings for species s from Atlantic and Gulf locations,
Lp, = Landings for species s from Pacific locations,
Xg = Exports of species s, and
lg = Imports of species s
The share of consumption of species s attnbutable to Atlantic and Gulf landings, SAGS, is
LAG» adjusted for exports divided by C,
Table 4-4 calculates the share of consumption by both tonnage and value applicable to
species originating m the Atlantic and Gulf Consumption shares can differ significantly
between tonnage and value because many imports are already processed into hgher value
products, such as fillets and blocks In general the analysis relies shares calculated based
on value
33
-------
Table 4-4
Share of Consumption Attributable to Atlantic and Gulf Harvests
Species
Shellfish
Crab
Lobster
American
Caribbean Spiny
Shrimp
Clam
Mussels
Oysters
Scalops
Items Fnfish
Cod
Haddock
Pol ex*
Dolphin
Flatfish/founder
Mackerel
Mullet
Ocean Perch
Safrnon
Canned salmon
FreshAFrozen
Sardine
Sea Bass
Stiark
Snapper
Swwdfish
Tuna
Canned tuna
Fresh JiFrozen
Freshwater fish
Crayfish
Tlapia
Trout
Note
Source
US Imports
(metric tori) (1 millions!
6966600 65310'
1003517 9624!
506371 93045
384306 58503
122066 34542
4293027 34220!
141301 498}
207273 521'
86566 3687
22,2074 14669
7753047 2715!
664743 343 5(
241488 10467
790365 14714
9.205 5 28 K
459538 20407
194915 2933
2601 062
91030 3031
2136739 92007
38820 1267
209 791 9 907 4£
247941 54 5C
1777 122
21650 460
164376 5421
157120 8827
2496709 7046;
171 521 9 398 6f
781490 305 9f
41466 2367
671875 1742;
44245 1451
US Exports
(metric tons! It mill Ions)
98 747 1 866 49
15 921 0 103 01
304772 2999!
303175 29719
1597 28(
122651 1030)
23503 103
5346 141
21702 98;
45516 388!
4986297 13738
847187 19561
784 01:
1142817 466 3J
00 000
680136 1250:
148085 129:
10665 1161
56037 64<
1303590 4681
418705 1326!
S3 488 5 33553
609582 408!
725 022
41048 123'
00 000
00 000
144640 3408
00 000
14 464 1 34 0!
000
198 9 07!
20835 34!
5275 16!
Pacific Landings
(metric tontl (tmllllontl
998589 42433
455234 21303
3113 471
00 000
43 012
486266 14551
1 644 6 34 77
1202 269
3 483 1 22 20
1454 1 29
21910679 6755
2326190 9621
00 000
15158461 20373
620 4 2 61
627291 14358
34936 049
35 002
00 000
2571203 15497
1000000 6027
157 120 3D 94 70
968405 1060
25 001
1 606 4 1 23
181 4 1 76
20178 770
179873 5256
00 0
179873 5255719
00 0
02 0
1458 0153
Atlantic/Gulf Landings
(metric tons! (t mlllionsl
5368646 164966
962343 18671
387795 30862
367443 28778
20263 2067
117^626 43955
573620 13318
21278 405
1624061 6669
239217 202 46
1022169 2319
131038 3068
75532 1908
35807 620
3734 1 09
230023 6479
254078 1462
75646 1093
17078 423
1609 011
00 000
1609 011
6476 023
19166 689
60446 664
47149 2379
18927 1021
45460 324C
00 OOC
45460 3246
70761 807
1 835 2 1 29
00 000
Total US. Consumption
(metric lonil II millions!
12346264 773854
2261885 12591
592507 9438
448574 5756
140774 3634
5829268 39041
707864 2075
224407 575
1723756 1158
417229 3116
25700598 22495
2274784 2749
316236 1236
14841816 00
101993 324
636716 2874
33 584 4 31 5
67617 00
5.207 1 28 1
3405961 6070
620115 597
2785846 6667
613240 245
20243 79
57112 01
203339 798
196225 1062
2577402 7556
1715219 3987
862162 3569
110238 310
669394 1720
40428 130
All/Gulf Landings as a Pet of Total
(metricians) (1 millions!
43 48% 21 32*,
4255% 1483*
6645% 3270"
8191% 49 9K
1439% 5.69*
2012% 11 261,
61 04% 64 16",
948% 704-
9422% 5751'
57 33% 6437*,
398% 1031'
576% 1116-
2388% 15431
024% N
366% 337-
3613% 2254'
7565% 4645',
10000% N
3280% 1508-
005% 002*
000% OOP!
006% 002-
106% 093*
9468% 8725'
10000% 100 00",
23 19% 29 8?
965% 961'
1 76% 4 30«,
000% 000»,
527% 910-
6419% 2605'
274% 075«
000% 000"
Imports of Caribbean spiny lobster mdude mports of rock lobsters (crayfish)
National Marne Fishenes Service website hltp //www st nmfs gov/st1 /commercial/ and
htto ffmnt st nmfc ocvfetlArade^ndex html
4.2.2 Impact of Mercury Reduction Scenarios on Mercury Exposure
The following equation calculates mercuiy intake attributable to consuming a given
species, s,
E. = (H / 30) * P, * SAG8 * Hg, * (1 - RAG, ) + ((Ms / 30) * P, * SPI, * Hg, * (1-Rpis),
where Ea = MeHg exposure to individual i from consuming species s in
micrograms per day (lug/day),
M5 = Meals of species s in a 30 day period,
P, = Meal portions in grams for individual i,
SAGS = Share of consumption of species s from Atlantic and Gulf
locations,
SPIS = Share of consumption of species s from Pacific and Import
locations,
34
-------
Hg, = Mercui> concentrations in (ug/g,
RAC« = Perceat reduction in mercury deposition applied to Atlantic and
Gulf landings of species s, and
Rpts = Percent reduction in mercury deposition applied to Pacific landings
and imports of species s,
Summing exposures over all species yields an estimate for each individual
Table 4-5 compares baseline mercury exposures for a coastal population (zone 0 and
zonel) with exposures that result from each of the three mercury reduction scenarios
presented in Section 3 The three mercury reduction scenarios result in lower exposures
relative to the baseline However, the reductions are not stncti> proportional to the
changes in deposition because of 1he different mix of domestic and imported species
consumed by individuals
Table 4-5
Mercury Exposure for Key Study Area Populations
(Jig/day)
Scenario
Baseline
Scenario 1
Scenario 2
Scenario 3
Source:
Measure
Median
Mean
Median
Mean
Median
Mean
Median
Mean
Pregnant Women
7 17
1183
700
11 45
659
1063
694
11 16
Men: 40-59
9 17
1308
874
1258
782
11 42
843
1221
Adult Men
686
1158
657
11 08
609
1004
630
1071
Adult Women
558
946
537
918
494
839
514
891
Calculated as explained in the text of this report
Mercury Dose-response Functions
In the past decade the World Health Organization (WHO), EPA, The National Academy
of Sciences, and other research organizations have conducted several exhaustive reviews
of the mercury health effects literature in efforts to update the reference dose (RfD) for
methylmercury In 1990 WHO's International Program on Chemical Safety (IPCS,
1990) reviewed the available lexicological and epidemiological literature on
methylmercury EPA's Mercury Report to Congress (1997) summarized the health
effects literature available at that time 2 In the late 1990s additional data from the
Seychelles and Faroe Islands studies became available The Faroe Islands study
(Grandjean et al, 1 997, 1 998) found that children subject to low doses of mercury
developed subtle developmental deficits, whereas the Seychelles study (Davidson et al ,
2 EPA's Mercury Report to Congress was first released for review as a draft in 1995 and finalized after
much review in 1997
35
-------
1998,1999) found no adverse effects that could be attributed to chronic low doses of
methylmercury In 1998 an expert panel reviewed these and other studies in a workshop
that met in Raleigh, NC (NIEHS, 1998) In 1999 Congress directed EPA to request that
the National Academy of Sciences perform an independent study on Ihe lexicological
effects of methylmercury The Academy's National Research Council (NRC, 2000)
reassessed the literature and provided guidance on how to interpret the sometimes
inconsistent research results More recently, a panel of experts convened by EPA (Abt
Associates, 2002) evaluated the available dose-response functions for developmental
neurotoxicity effects and concluded that an assumption of linearity was reasonable for
chronic low doses from consuming fish and that there is no evidence to assume an effects
threshold
As noted above, the health effects literature on mercury is extensive and has been subject
to detailed reviews by EPA (1997), NRC (2000), and others The purpose of this section
is to select the most appropriate dose-response parameters, to apply them to the exposure
data, and to characterize the scientific uncertainties that underlie the resulting estimates
of health effects The subsections below evaluate the most reliable dose-response
relationships for each of the mercury health endpomts
paraesthesia (pnckling, tickling, itching sensation) as a first symptom of acute
mercury poisoning,
neurodevelopmental deficits from fetal exposure affecting attention, intelligence,
vision, and coordination, and
cardiovascular nsks including elevated blood pressure in children, acute
myocardial infarctions (heart attacks), and pre-mature death
All the dose-response functions require mercury exposure, as measured by mercury
concentrations in hair or blood, as input Thus, translating mercury intake into mercury
hair and blood concentrations is the first task in applying mercury dose-response
functions The second task is to select the dose-response functions for each endpomt, and
the third task is to apply those dose-response functions to quantify the effects for each
endpomt under baseline and reduced mercury exposures
4.3.1 Converting Mercury Intake to Body Burden
Mercury dose-response functions require a measure of body burden as input Mercury
body burden is typically expressed either in terms of concentrations in blood or hair
Mercury in blood is expressed in micrograms per liter (ug/L), and mercury in hair is in
rmcrograms per gram (ug/g) or parts per million (ppm)
NRC (2000) evaluated both the one-compartment pharmacokinetic and physiologically
based pharmacokinetic (PBPK) models3 and concluded that the simpler, one
compartment model was acceptable for use in converting mercury intake into
concentrations in blood The one-compartment model predicts the steady state
1 NRC characterized the PBPK model as conceptually more accurate and flexible than the one-
compartment model, it i!> also more complex, and thus, more difficult to evaluate " (NRC, 2000, p 84)
36
-------
melhylmercuiy concentrations in blood under the assumption that the daily mass of
methylmercuiy entering from ingestoon is equal to Ihe amount exiting by excretioa NRC
(2000) cautioned that each of the parameters in the one- compartment model are random
variables subject to a probability distribution so that evaluating the parameters at their
means may not protect the most sensitive individuals in a population However, for
evaluating the most likely eflects of mercury intake on blood mercury levels in
populations the one- compartment model using parameter means is appropnate The
equation below estimates blood mercury concentrations following the one compartment
model4
Where
C = concentiabon of methylmercury in the blood (ng/L),
d = ingested dose dietary intake (ug/day),
A = absorption factor of ingested mercury (0 95, umtless),
f = fraction of absorbed methylmercury that is taken up by the blood (0 05,
unitless),
b = elimination rate per day of methylmercury (0 0 1 4 day s -1 ), and
V = volume of blood (5 L)
Converting mercury levels in blood to mercury concentrations in hair makes use of a
standard ratio Mercury m blood at a concentration of 250 ug/L is approximately equal to
mercury in hair at a concentration of 1 jig/g (NRC, 2000)
4.3.2 Dose-Response Functions
EPA (1997 and NRC (2000) provide extensive reviews and analysis of the dose-response
functions from various studies This section does not attempt to reanalyze these studies,
but rather to briefly summan/e the basis for the dose- response functions selected and to
point out any uncertainties th.it underlie the analysis
4.3.2.1 Paraesthesia
In acute doses methylmercury is highly toxic, and the methylmercury poisoning events in
Japan and Iraq resulted in overt symptoms that ranged from paraesthesia (pnckhng,
tickling, or itching of extremities) to blindness and death WHO (IPCS, 1990) reviewed
the studies that relied on data from these events and presented dose-response information
from both a "hockey-stick" model with a threshold and an S- curve model with no
threshold The data from the two poisoning incidents suggest a threshold of paraesthesia
effects at concentrations of mercury in blood of about 200 ppb, or an intake of about
<250 to 400 u,g/day of methylmercury At the 200 ppb blood- mercury lev el about 5
percent of exposed adults exhibited paraesthesia
4 Equation and units are from NRC (2000)
37
-------
Additional evidence from studies offish-eating populations reviewed by NRC (2000)
indicate both a higher frequency of neurologic effects at the 200 ppb Hg in blood
threshold and the prevalence of more subtle deficiencies in visual and motor functions
below that threshold This additional evidence suggests the need to reesnmate the dose-
response function for adult central nervous system effects based on more subtle
endpoints However, at present there are no studies that quantify a dose-response
relationship based on these visual and motor function deficiencies
43.2.2 Neuro-developmental Effects
There exist four major chronic, low-dose, prospective epidemiological studies offish-
eating populations New Zealand (Kjellstrom et al, 1986,1989), Faroe Islands (Grandjean
et al, 1997,1998), the Seychelles Islands (Davidson et al, 1998,1999) and the Amazon
(Grandjean et al, 1999) All tested the hypothesis of a relationship between fetal
exposure to methylmercury and neuropsychological and neurodevelopmental effects in
children.
NRC (2000) provides an extensive review and comparison of the data, methodologies,
and health endpoints evaluated in these studies The main health endpoints evaluated
include status on a neurological examination, age at achievement of developmental
milestones, infant and preschool development, and childhood development The
childhood developmental endpoints include a battery of psychological, neuromotor, and
scholastic tests administered to the test subjects at various ages In both the New Zealand
and Faroe Islands studies higher mercury exposures were correlated with poorer
performance on various tests The New Zealand study included IQ tests, and the results
showed IQ deficits to be correlated with higher methylmercury exposures
On NRC's recommendation EPA used the benchmark dose for the lower confidence limit
(BMDL) from the Faroe Islands study's Boston Naming Test as the basis for calculating
an RfD NRC acknowledged that BMDLs for several endpoints in the New Zealand
study, including IQ tests, were actually lower, but had some reservations about relying on
that study due to the smaller sample size and its sensitivity to one outlier value5
However, the Boston Naming Test is not a metnc for which economists have developed
\ alues, although in theory some of the tests measured in the Faroe Islands study could be
linked to IQ (Bellinger, personal communication, 9/19/03) Consequently, the New
Zealand study provides the only dose-response function whose endpomt, IQ, is
quantifiable in dollars
IQ Deficits
In the ongmal New Zealand data, Kjellstrom et al (1989) used a categoncal variable for
hair mercury and reported a-4 41 change in IQ score for subjects whose maternal hair-
mercury exceeded a threshold of 6 ppm Below the threshold no effects on IQ were
5 One Maori mother had mercury in hair ot 86 ppm, which waj> more than four time higher than the next
highest value of 20 ppm Omitting this observation resulted in a more negative coefficient on mercury
exposure with greater significance
38
-------
observed In a reanalysis of the New Zealand data Crump et al (1 998) estimated a dose-
response function using mercury in hair concentrations as a continuous variable in a
linear specification with no threshold After dropping one outlier value, the reanalysis
estimated that a 1 ppm increase in hair mercury levels resulted in a 0 54 loss in IQ as
measured by the Wechsler Intelligence Scale for Children - Revised Performance IQ
(WISC RP) test and a 0 53 Joss in IQ as measured by the Wechsler Intelligence Scale for
Children - Revised Full Scale IQ (WISC RF) test 6
To quantify the impact of mercury exposure on IQ the analysis employs both the original
Kjellstrom et al coefficient, -441, with a 6 ppm hair mercury threshold and Crump et
al 's WISC RF coefficient, -0 53, without a threshold Section 433 summarizes the
results of applying these two IQ dose- response functions to estimate the effects on IQ for
mercury exposure from fish consumption by pregnant women under baseline conditions
and for the three scenarios of reduced mercury intake
IQ<70
Fetal exposure to mercury may result in some infants being bom with IQs less than 70,
which categorizes them as mentally handicapped IQ scores are standardized to a
distribution with a mean of 1 00 and a standard deviation of 1 5 An IQ score of 70 is 2
standard deviations below the mean of 100, and specialists regard this score as the point
below which children require significant special compensatory educatoa
Assuming a normal distribution, an IQ score of less than 70 that is 2 standard deviations
below the mean occurs in about 2 3 percent of the population 7 If there is a change in
mercury exposure, the result Mil be to shift the mean as given by Crump et al 's dose-
response coefficient
AMean IQ = -0 53 * AMean HgB (ug/L)
Assuming a distnbubon with the same standard deviation, it is relatively straightforward
to compute the change in the probability of children with IQs below 70 by comparing the
baseline distnbution, O(z&), ^vrth the post- compliance distribution, O(zpC), that reflects a
change in mercury exposure,
where
hP(lQ< 70) is the change in die probability of a child having an IQ<70,
is the percentage of the post-compliance distnbuhon,
where
ZPC = ((1 00-0 53*AMeanHgB) -70) / 15, and
6 NRC (2000) reported these coefficient values as-0 54 and -0 55, respectively, in Table 7-1, p 280
The -055 value appears to be a misprint
7 See any statistical textbook for probabilities in the tall of a normal distribution
39
-------
is the percentage of the baseline distribution,
where
zfci = (100-70)/15 =
The decrease in the number of children bom with an IQ below 70 is then simply the
change in the probability times the number of births in the population
43.2.3 Hypertension
Mercury is known to accumulate in the heart, and reports from methylmercury poisoning
events diagnosed hypertension and abnormal heart rate in some victims Two recent
epidemiological studies reviewed by NRC (2000) found associations between low-dose
mercury exposure from fish consumption and cardiovascular effects on middle- age men
and on heart function in children at age 7
Adults
In a study of 1 ,833 Finnish men, aged 42-60 Salonen et al (1 995) compared dietary
intake of fish and mercury in hair and unne with the prevalence of acute myocardial
infarction (AMI) and death from coronary heart disease (CHD), cardiovascular disease
(CVD), and all cause mortality (ACM) Mean fish intake was 46 5 g per day (ranging
from 0 to 61 9 2 g per day) and mean mercury in hair was 1 92 ppm (ranging from 0 to
15 67 ppm) The study followed this cohort from March 1984 to December 1991, but
indicated the maximum follow-up period was 7 75 years for AMI events and 8 75 years
for deaths 8
Salonen et al employed a Cox proportional hazards model and reported their findings in
terms of relative nsks 9 The authors estimated models for AMI (fatal and non- fatal
combined), fatal coronary heart disease (CHD), fatal cardiovascular disease (CVD) and
all cause mortality (ACM) For each health endpomt, the authors estimated separate
models as a function of mercury concentrations in hair, fish intake, and mercury intake
The models controlled for a number of potential confounding effects, including age,
examination year, ischemic exercise ECG, maximal oxygen uptake, family history of
CHD, cigarette-years, mean systolic blood pressure, diabetes, socioeconomic status,
urban v rural residence, dietary iron intake, serum apolipoprotein B, HDL2 cholesterol,
and femtin (> v <200 ng/L) concentrations
The authors reported results for two model specifications that related hair mercury
concentration to nsks of AMI and premature death from CHD, CVD and ACM, but not
all coefficients were significant at p<0 05 In models that specified hair mercury as a
s The authors reported that the average time to an event was 5 years, but reported following some patients
di> long a 7 75 years lor AMI events
-------
continuous variable the authors reported that the nsk of non-fatal and fatal AMI increased
by 6 8 percent (p=0 175) and the nsk of all cause mortality (ACM) increased by 9 0
percent (p=0 043) for each 1 ppm increase in hair mercury In a second model they used
a dichotomous variable to compare risks of men with >=2 0 ppm mercury in hair to men
with mercury levels <2 0 ppm The authors reported a 69 percent (p=0 038) greater nsk
of fatal or nonfatal AMI associated with hair mercury above 2 0 ppm and a 93 percent
greater (p=0 007) nsk of pre-mature death from all causes (ACM) Because of the
greater significance of the coefficients the result from the latter specification has been
more widely reported (see NRC, 2000) However, it is more plausible to expect that
mercury nsks increase with increasing levels of mercury exposure, and this analysis
evaluates nsks using the specification in which mercury concentration is a continuous
variable
There are a number of concerns about using the results of the Salonen et al study Due to
the relatively small sample size (n=l,833) of the cohort the power of the models to
estimate effects of small probability events is less than optimal As a result, some of the
coefficients on mercurj exposure are insignificant, and others have relatively wide
confidence intervals In the AMI model in which hair mercury enters as a continuous
vanable the coefficient is not significant at p<0 05 (95 percent confidence interval)10
In addition, the AMI dose-response relationship includes both non-fatal and fatal AMI
events so that there is some overlap with the separate dose-response relationships for
mortality effects Consequently, to use the AMI dose-response function requires
separating out the fatal from non-fatal AMI events and assuming fiat the coefficients for
the joint model apply to the non-fatal component
Despite these concerns about the Salonen et al dose-response functions, more recent
research provides additional evidence of a relationship between mercury and heart
disease Guallar et al (2002) found a statistically significant relationship between
mercury burden, as measured in toenails, and nsk of AMI for men up to age 70
Yoshizawa et al (2002) studied the association between total mercury concentration and
nsks of coronary heart disease in men age 40 to 75 After eliminating from the sample
dentists, whose occupation may expose them to elemental mercury, the authors observed
an association of toenail mercury with nsk of coronary heart disease However, the
sample size was not large enough to conclude the association was statistically significant
These two studies suggest that the finding of an association between mercury and
cardiovascular disease in men age 42 to 60 from the Salonen et al study may extend to
men as old as 70 or 75 How ever, there is no evidence, as yet, to extend these findings to
women.
10 Harvard University's Center for Risk Analysis in a forthcoming study ot the nsks of consuming fish
decided not to use the Salonen et al results as a basis for establishing a quantitative dose-response
relationship for the association oi methylmercury and the nsk of coronary heart disease They listed three
reasons for this decision 1) it is the only study that quantifies effects with reference to methylmercury
intake, hair or unne content, 2) the magnitude of the effect is remarkably large, but the confidence levels
are wide, suggesting low preusion, 3) fish consumption patterns, types offish, and levels of pollution in
fish consumed in Finland may nol be representative offish consumption in a U S population (Koemg, A ,
personal communication, 12/3/03)
41
-------
The analysis applies the dose-response for non-fatal AMI and ACM from the Salonen et
al study to estimate the incidence of these events However, there is a high level of
uncertainty associated with these estimates, since they denve from one study that has not
been replicated in the United States Readers should also remember that the dose-
response coefficient that relates mercury exposure to AMI events is not significant at
p<005
Children
In addition to the adult cardiovascular effects of mercury found in adult men, there
appears to be a link between fetal exposure to methylmercury and cardiovascular health
S0rensen et al, (1999) studied 1,000 children from the Faroe Islands and found an
association between prenatal mercury exposure and a child's blood pressure at age 7
Diastolic and systolic blood pressure increased by 13 9 and 14 6 mmHg,1' respectively,
as maternal cord-blood concentrations rose from 1 to 10 ug/L12 Above 10 ng/L in cord
blood the authors reported that the increase in blood pressure leveled off at 13 9 and 146
mm Hg Also, heart rate variability, a marker of the effectiveness of cardiac autonomic
control, decreased by 47 percent as cord-blood concentrations increased from 1 to 10
(jg/L Surprisingly, at maternal exposures above 10 u,g/L the authors reported finding no
additional increase in children's blood pressure
The analysis uses this dose-response function by treating the 10 [4g/L exposure level as a
threshold It applies the 13 9 and 14 6 point increase in diastohc and systolic blood
pressure to all fetal exposures at or above 10 ug/L and evaluates the percentage that
exceed the 95th percentile threshold for diagnosing hypertension Since doctors often
don't detect or don't act on small changes in blood pressure, this approach captures most
of the likely effect'3 Applying the 10 ug/L level as a threshold is conservative in that it
only applies the dose-response function to the most exposed population It also simplifies
the estimation of the fraction of fetally-exposed infants likely to exceed the 95th
percentile BP level, but ignores the potential risks of less-exposed infants below that
level
This Sorensen et al study provides quantitative information to establish a relationship
between prenatal mercury exposure and blood pressure at age 7 This association is quite
plausible, given the well-established effect of methylmercury on kidney and heart
function (NRC, 2000) However, this is the only study to report this effect on children,
and there is a large amount of uncertainty in relying on results from only one study
1' Blood-pressure is measured in millimeters ot mercury (mm Hg)
12 There is uncertainty in the relationship between mercury levels in maternal cord blood and blood (see
NRC, 2003, pp 93,112-117, for a discussion) This analysis assumes all relationship
n Blood pressure can vary greatly from reading to reading, and doctors usually require evidence from
multiple readings before concluding that blood pressure is elevated
42
-------
4.3.2.4 Summary of Dose-Response Functions
Tables 4-6a and 4-6b summarize dose-response functions for \anous health endpoints
The tables include the endpoint evaluated, the study source, the slope, threshold, and
applicable population The Table 4-6a summarizes the endpoints for mercury poisoning
and neurological deficits from fetal exposure These neurological dose-response
functions have been subject to extensive peer review by the IPCS (1990) and National
Research Council (2000) and provide a reasonable basis for calculating incidence of
mercury health effects
Table 4-6a
Summary of Mercury Dose-Response Functions for Neurological Endpoints
Neurological Endpoints
Endpoints
evaluated
Paresthesia
Weschler
ntelligence Scale
for Children,
Revised full-scale IQ
Weschler
intelligence Scale
or Children,
levised full- scale IQ
Dose-response
Study
WHO (IPCS,
1990)
Kjellstrom et al ,
1989
Crump et al ,
1998
Slope
0 05 increase in baseline
nsk above threshold
-4 41 IQ points for
exposures above
threshold
-0 53 IQ points for each
1 u,g/g increase in
maternal hair
Threshold
200 (ig/day
Hg intake
6mg/kgHg
in maternal
hair
None
Pop
Adults over 15
Pregnant women
age 15 to 44
Pregnant women
age 15 to 44
Table 4-6b summarizes the dose-response functions for cardiovascular endpoints These
dose-response functions are based on a limited number of studies, which have not been
subject to extensive peer review As a result, there is considerably greater uncertainty in
applying these relationships to quantify the impact of mercury exposure to cardiovascular
health effects in adults and children.
43
-------
Table 4-6b
Summary of Mercury Dose-Response Functions for Cardiovascular Endpoints
Cardiovascular Endpoints
Endpoints
evaluated
Acute myocardial
infraction (AMI)
All cause mortality
(ACM)
Average change in
systolic and diastohc
)lood pressure (BP)
Dose-response
Study
Salonen et al ,
1995
Salonen et al ,
1995
S0rensen et al ,
1999
Slope
0 068 increase in risk for
each l^g/gHginhair
0 090 increase in nsk for
each 1 ug/g Hg in hair
Increase in systohc=14 6
and diastohc=l 3 9 for
an increase in maternal
cord blood from 1 to 10
Hg/L
Threshold
None
None
None, but no
increase in
BP above 10
u,g/Lcord
blood
Pop
males 42-60
males 42-60
children age?
4.3.3 Populations in At-Risk Age Cohorts
To apply the dose-response functions for methvlmercury exposure requires estimates of
populations in at-nsk age cohorts The 2000 Census of Population provides the data
necessary to estimate the various gender and age cohorts necessary to match the dose-
response functions The sections below describe supplementary analyses or adjustments
necessary to estimate other at nsk populations
The analysis interprets the dose-response functions narrowly and applies them only to the
specific genders and age groups for which evidence of an effect exists This is a
conservative approach, but may underestimate the health effects of methylmercury
exposure
4 3.3.1 Live Births
Application of the dose-response function for fetal exposure requires an estimate of live
births in the study area The 2000 U S Census of Population provides population
estimates for women age 15 to 44 for the four states in the study area The population
data are available by county, which permits estimates for the four distance zones from the
coast and the portion of Flonda included in the study area Multiplying the fertility rate,
defined as the number of live births, by the number of women of childbeanng age, 15 to
44, yields an estimate of the number of live births Table 4-7 below summanzes fertility
rates for 2001 for the four states in the study area.
44
-------
Table 4-7
Fertility Rates in the Study Area
State
Florida
Georgia
North Carolina
South Carolina
Fertility Rate
670
71 1
704
628
4.33.2 Adult Men Age 42-60
The cardiovascular dose-response functions pertain to a cohort of men ages 42 to 60
years However, the age ranges reported in the U S Census do not correspond exactly to
this cohort The analysis substitutes an age cohort, 40 to 59, and assumes that the dose-
response function applies equally to this population The 40 to 59 age cohort is slightly
younger and slightly larger than the 42 to 60 age cohort studied by Salonen et al
Evidence from other studies suggests that the cardiovascular effects of mercury exposure
may extend beyond age 62, perhaps to to age 70 or 75 Also, it is likely that women of
similar ages would suffer similar effects, but not necessarily at the same response rate
Thus, narrowly restricting the AMI and ACM effects to adult men, age 40 to 59, is likely
to underestimate the benefits
43.4 Estimation of Impact of Mercury Intake on Health Endpomts
This section summarizes the impact of reductions in mercury exposure on the vanous
health endpoints As a reminder, scenario 1 represents a 30 percent reduction in U S
deposition, scenario 2 represents a 100 percent reduction, and scenario 3 represents a 51
percent reduction The analysis also reports the impacts across four population zones
Zone 0 and zone I include populations within 100 miles of the coast, and the impacts on
tins population reflect fish consumption at a level comparable to Flonda residents Zones
II and HI represent more distant populations and reflect fish consumption at the level
reported by the NHANES survey
Paraesthesia
As noted above, analysis of the data from the Iraqi poisoning event indicated the first
symptoms of mercury poisoning, paraesthesia, occurred at a body burden of 200 ppb
blood mercury, which results from a mercury intake of about 250-400 u.g/day The
maximum exposed individual under baseline conditions consumed at a rate of about 205
Hg/day, which is below the paraesthesia threshold Since it is possible that symptoms
could occur below this threshold and since NHANES coverage of coastal populations
may not be sufficient to charactenze fully the extreme upper end of the distribution, it is
possible that cases of paraesthesia do exist But dose-response functions that demonstrate
45
-------
effects below this threshold do not currently exist, and based on the data available to this
study consuming fish at current exposures results in no incidence of paraesthesia
IQ Deficits
Tables 4-7a and 4-7b summanze the IQ points saved under each of the mercury reduction
scenarios Table 4-7a summanzes the results of applying Crump et al 's dose-response
function without a threshold, and Table 4-7b summanzes the results of applying
Kjellstrom et al's dose-response function with a threshold of 6 ppm mercury in maternal
hair Applying the dose-response function without a threshold, Table 4-7a shows that
scenario 1 results in 7,999 IQ points saved (compared to the baseline), scenario 2 results
in 26,617 IQ points saved, and scenario 3 results in 13,798 IQ points saved
Applying the dose-response function with a threshold in results in considerably fewer IQ
points saved, as shown in Table 4-7b The 6 0 ppm mercury in hair threshold is
sufficiently high that only those women in the highest percentles are affected For non-
coastal populations that consume less fish, the mercury reduction scenarios do not shift
the distributions enough to affect the highest percentiles, and there is no impact of
reduced mercury intake in zones 2 and 3 Scenarios 1 and 3 save 10,985 IQ points, and
scenario 2 saves 16,477 IQ points
Table 4-7a
Impact of Mercury Exposure Reduction on IQ: No Threshold
Reduction Scenario 1
Reduction Scenario 2
Reduction Scenario 3
IQ points saved - No threshold
ZoneO
1,627
5,200
2,752
Zone I
2,660
8,502
4,499
Zone II
2,118
7,369
3,736
Zone III
1,594
5,545
2,811
Zone Total
7,999
26,617
13,798
Table 4-7b
Impact of Mercury Exposure Reduction on IQ: With Threshold
Reduction Scenario 1
Reduction Scenario 2
Reduction Scenano 3
IQ points saved - With threshold
ZoneO
4,168
6,253
4,168
Zone I
6,816
10224
6,816
Zone II
0
0
0
Zone III
0
0
0
Zone Total
10,985
16,477
10,985
IQ<70
Reducing fetal exposure to mercury results in a shift in the distnbution of IQ scores The
result is that fewer children fall below the IQ threshold of 70, where special education is
required Table 4-8 shows that in scenano 1 there are 8 fewer children below 70 IQ
Scenarios 2 and 3 result in 27 and 14 fewer children below 70 IQ, respectively
46
-------
Table 4-8
Impact of Mercury Exposure Reduction on Children Whose IQ is Less than 70
Reduction Scenario 1
Reduction Scenario 2
Reduction Scenario 3
Reduction of Incidence of IQ less than 70 by zone
ZoneO
2
5
3
Zone I
3
9
5
Zone II
2
7
4
Zone in
2
6
3
Zone Total
8
27
14
Adult Hypertension
Tables 4-9 and 4-10 summarize the results of applying Salonen et al 's dose-response
functions for acute myocardial infarction (AMI) and all cause mortality (ACM)
Estimating the cases avoided from coefficients derived from the Cox proportional
hazards model uses die following relationship
where
Ay = B*(exp(P*Ax)-l),
Ay = change in number of cases,
B = baseline nsk (i e nsk level before the reduction in mercury exposure),
p = coefficient (=tn(relatve nsk/Ax)),
Ax - the change m exposure (hair mercury in ppm)
Since the change in exposure will be negative, reflecting a reduction in mercury intake,
the change in cases will be negative, which represents cases avoided
To apply the cardiovascular dose-response relationships requires the baseline risks of
non-fatal acute myocardial infarction and all cause mortality CDC provides estimates of
the ACM value on-hne, and Appendix 4-1 calculates a value of 0 0056 (560 per 100,000
population) as a baseline nsk of all cause mortality for men, age 40-59 The estimate for
AMI is 0 0067, which is an interpolation from estimates given for non-fatal heart attacks
in ages 18-64 and 65+ (Apelberg et al, 2003)
Table 4-9 provides estimates of the cases of non-fatal AMI avoided as a result of
reducing mercury exposure Scenano 1 results in 73 cases avoided, scenario 2 results in
247 cases avoided, and scenario 3 results in 121 cases avoided
47
-------
Table 4-9
Impact of Mercury Exposure Reduction on Incidence of Acute Myocardial
Infarction
Reduction Scenano 1
Reduction Scenano 2
Reduction Scenano 3
Non Fatal AMI Incidence- Number of cases saved in each scenano
ZoneO
14
48
24
Zone I
23
75
37
Zone II
21
72
35
Zone in
15
52
25
Zone Total
73
247
121
Table 4-10 shows that under Scenano 1 there would be 80 fewer pre-mature deaths
attributable to mercury exposure Scenano 2 results in 272 fewer deaths, and scenano 3
results in 133 fewer cases
Table 440
Impact of Mercury Exposure Reduction on Incidence of All Cause Mortality
Reduction Scenario 1
Reduction Scenano 2
Reduction Scenario 3
ZoneO
16
52
26
Zone I
25
82
41
Zone II
23
79
39
Zone HI
16
57
28
Zone Total
80
272
133
Hypertension in Children
Interpreting the medical significance of an increase in a child's systolic and diastohc
blood pressure depends on a number of factors, including gender, age, weight, and height
In general, blood pressure protocols indicate anti-hypertensive, drug treatment for
children whose blood pressure exceeds the 95th percenule for their gender, height and
weight (National High Blood Pressure Education Program, undated) The difference
between the 50th and 95th percentoles for systolic blood pressure in both boys and girls
age 7 is about 19 points For diastohc pressure the difference is about 17 points for both
boys and girls This information enables calculation of the z-score for the distribution
and back calculation of the percentjle at which an increase in systolic and diastohc blood
pressure due to fetal exposure to mercury \\ould cause a child's blood pressure to exceed
the 95th percentle Appendix 4-B summarizes this calculation, which indicates that an
increase of 14 6 points in systolic blood pressure would cause children age 7 in the 65
percentile and above to exceed the 95th percentle BP value u Similarly, an increase of
13 9 points in diastohc blood pressure would cause children age 7 in the 62 percenule and
above to exceed the 95th percentle BP value Using the slightly more stringent diastohc
" The calculation uses 50th and 95th percentile data from 1987 blood pressure tables More recent data
(NHBEP, undated) show a slight increase in the 95lh percentile values Lacking data on the 50th percentile
Irom the updated data, the calculation assumes the distribution has shifted to the right with no change in the
standard deviation
48
-------
measure, the dose-response function for children indicates that about 38 percent of the
children exposed in ulero at 10 ug/L or above will exceed the 95th percentile for diastohc
blood pressure and become i andidates for treatment for hypertension Whether all or
only a fraction of these children would require regular treatment for hypertension and for
how long is uncertain. Ihus, the analysis applies Ihis dose-response function to estimate
cases of childhood hypertension by calculating 38 percent of all children whose mother's
mercury exposure exceeds 10 |ig/L
In addition, it seems likely that hypertension in children is likely to cause increased nsks
of cardiovascular disease in later life Recent research by Raitakan et al (2003) found
that elevated blood pressure in childhood is associated with elevated nsks of carotid
artery intima-media thickness, which is a marker of pre-clinical athlerosclerosis'5 This
study provides important evidence of the significant nsks associated with childhood
blood pressure However, there is still no direct link between childhood hypertension and
nsks of final endpoints, such as AMI and ACM The Raitakan et al study also
underscores the fact that values based solely on treating childhood hypertension may
greatly underestimate the benefit of reducing mercury exposure to avoid other nsks
associated with this health endpoint
Table 4-11 summanzes the results of applying Serensen et al's dose-response functions
relating mercury exposure to blood pressure in young children In Scenario 1 the
reduction in mercury exposure is not great enough to shift the distnbution from the
baseline distnbution in which blood-mercury levels exceed 10 ug/L in the 71st percentile
for coastal populations and in the 90th percentile for inland populations However, in
scenanos 2 and 3 tiie reduction in mercury exposure is large enough to shift the
distnbution and generate a reduction of 2,277 and 947 cases, respectively
Table 4-11
Impact of Reduction in Mercury Exposure on Incidence of Elevated Blood Pressure
in Children
Reduction Scenario 1
Reduction Scenario 2
Reduction Scenario 3
Cases of Hypertension Prevented (Defined as BP above the 95th
percentile)
ZoneO
0
539
359
Zone I
0
881
587
Zone II
0
489
0
Zone in
0
368
0
Zone Total
0
2,277
947
The study lound that systolic blood pressure measured at ages 12 to 18 was significantly associated with
adult intima-media thickness, but blood pressure measured at ages below 12 was not
49
-------
4.4 Economic Benefits of Reducing Mercury Exposure
Where mercury exposures cause health impacts, quantifying the changes in economic
welfare requires valuing 1he changes in nsks for each health endpoint Section 441
bnefly discusses the economic approaches and studies used to value each mercury health
endpoint Other regulatory documents review these studies in detail, and the intent of this
section is to provide the reader with sufficient information to understand the economic
approach and the biases and uncertainties that underlie die values Section 442 then
applies those valuations to quantify the benefits of reducing mercury exposure in the
study area
4.4.1 Valuing Health Endpomts
To value reductions in adverse health effects economists have relied on two main
approaches measuring the costs associated with a health effect and measuring the
willingness to pay to avoid the risk Where medical treatment can reverse the injury or
illness and where other effects, such as pain and suffenng or side effects are minimal,
estimating of costs of medical treatment and lost income provides a reasonable
approximation of willingness to pay However, where nsks result in changes that are not
fully reversible, such as premature death, loss of IQ, or lifestyle changes after heart attack
or stroke and/or where there is significant pain and suffenng, cost of illness (COI)
measures may substantially underestimate actual values Consequently, economists
prefer approaches that estimate willingness to pay (WTP) to avoid the nsk However,
willingness-to-pay values exist only for valuing pre-mature death
This section converts all costs and willingness-to-pay estimates into 2003 dollars Where
the discussion reports economic valuations in other than 2003 dollars, the analysis uses
various Consumer Price Indexes (CPI) to inflate estimates to current dollars
4.4.1,1 Paraesthesia
Exposures at levels that begin to exhibit symptoms of mercury poisoning require medical
diagnosis and care Medical protocols for treating mercury disease exist and largely
consist of eliminating the source of the mercury exposure However, data from actual
cases on the costs of treatment are difficult to obtain, since it occurs rarely in the United
States Nevertheless, the treatment protocol requires diagnosis, testing, and momtonng
until mercury levels have declined to normal In addition to medical costs additional
losses may pertain if the disease or medical care requires time off from work or
necessitates changing occupations
A search of the literature did not find any studies that have developed information or data
on the costs of diagnosis and treatment of mercury poisoning
4.4.1.2 Valuing Neurodevelopmental Effects on IQ
50
-------
Researchers have detected and quantified a number of effects resulting from fetal
exposure to methylmercury Infants exposed to mercury in utero ha\e displayed a range
of neuropsychological and neurodevelopmental deficiencies, as measured in a variety of
tests of visual acuity and psy chomotor skills In addition, the New Zealand study
(Kjellstrbm et al 1989) four dan impact of mercury exposure on IQ While all the
neuropsychological and neuiodevelopmental deficiencies represent a loss in well-being to
affected children and their families, economists are able to value only the loss in IQ
Valuation of IQ deficiencies combines estimates of two cost elements
1 foregone earnings and
2 costs of compensate ly training or special education
Studies are available to estimate the effect of IQ deficits on earnings and education, but at
the current time no one has developed an educational program to compensate for the
neurodevelopmental effects specific to mercury Instead, estimates of the costs of
compensatory training are based on programs developed for children exposed to lead on
the assumption that the special educational needs and costs would be similar Since the
loss of IQ affects more lhan just earnings, ihe monetary estimates of earnings loss and
costs of compensatory training represent only a lower bound, and economists
acknowledge (hat parents' willingness to pay to avoid die loss of IQ in their children is
likely to be significantly greater
Foregone Earnings Due to IQ Loss
The effect of IQ on lifetime earnings is well established in the economic literature, and a
number of regulatory impact analyses (RIAs) of lead exposure have valued IQ effects
This discussion summarizes the methodology and results reported in a recent RIA for Ihe
Metal Products and Machinery (MP&M) regulation (Abt Associates, 2003)
A person's intelligence affects earnings in two ways The direct effect is the effect of IQ
on employment and earnings with years of schooling held constant, and Ihe indirect effect
is Ihe effect of IQ on years of schooling and Ihe subsequent effect of years of schooling
on probability of employmen t and earnings Salkever (1995) quantified both of these
effects based on the data from the National Longitudinal Survey of Youth (NLSY)
Salkever (1995) found that Ihe combined effect of a 1 point change in IQ amounts to a
change in lifetime earnings of 1 93 percent for men and 3 22 percent for women. Using
data from the 1992 Current Population Survey and applying a 3 percent discount rate and
a 1 percent increase in productivity over an earnings penod from age 18 to 67, Abt
Associates (2003) estimated the present value of average lifetime earnings Assuming
the distribution of earnings for employed persons and labor force participation rates
remain constant over time, the expected value of foregone earnings for a child (averaged
over male and female) bom in the U S amounts to $448,957 ($2001) Applying the
percent of wage loss per IQ point, yields a value of $10,675 ($2001) per IQ point16
16 This value ib an average for mal es and lemales
51
-------
Salkever (1995) also found tiiat IQ affects educational attainment, and therefore IQ's
effect on ihe cost of education must be subtracted from its contribution to lifetime
earnings to generate a net return This educational cost includes both the direct cost of
educating a child and the opportunity cost of lost income during the years of schooling
Abt Associates (2003) reported that the marginal cost of educating a student (current plus
capital expenditures) amounts to $8,898 ($2001) per year Salkever estimated that the
estimated effect of IQ on educational attainment amounts to 01007 years per IQ point,
which amounts to $896 per IQ point Assuming this cost occurs at the end of a student's
education, age 19, discounting this cost to age 0 results in a present value of $511
($2001) The opportunity cost of lost income is the difference between full-time and
part-time earnings Abt Associates (2003) reported a value of $746 for an additional
0 1007 years of school, discounted to age 0
Combining the effect of IQ on earnings ($10,675) with the IQ's impact on the direct cost
($511) and opportunity cost ($746) of education yields a net value of $9,419 ($2001) per
IQ point
Cost of Educational Resources for Compensatory Training
The approach used to value Ihe cost of educational resources to compensate for IQ
deficiencies resulting from lead exposure requires two steps
Estimate the number of incremental children with IQs less than 70 (the definition
for mental retardation), and
Apply estimates for the annual costs of special education for the years of
schooling required
Section 4 3 estimated the impact of mercury reduction scenanos on the distribution of IQ
scores to estimate the number of children who fall below 70 To value this impact
requires estimates of the cost of special education
The annual cost of special education derives from work done by Kakalik et al (1981),
who estimated the annual costs of part-time special education for children attending a
regular school program This amounts to $6,959 ($2001) Assuming this cost of
compensatory education begins at age 7 and extends through age 18, the expected present
value cost amounts to $58,012 per child (Abt Associates, 2003)
It is likely that mentally handicapped children would require more than a part-time
special education program, a full-time program is likely to be significantly more cosily
As a result the estimate of $58,012 per child is likely to underestimate the costs of special
education for children impaired by fetal mercury exposure Moreover, this education
does not compensate for the loss of all the qualities of life conferred by greater mental
ability, and willingness to pay to avoid the loss in IQ is likely to exceed Ihe estimates of
compensatory education costs
52
-------
4.4.1.3 Valuing Hypertension, Heart Attacks (AMI), and Premature Death
Mercury exposure affects heart function and appears to contribute to abnormal heart
rhythms and high blood pressure that exacerbate the problems of cardiovascular disease
Salonen et al (1995) estimated dose-response relationships for non-fatal acute
myocardial infarction (AMI) and premature death in adults, and Sorensen et al (1999)
found a relationship between mercury burden and blood pressure in children To date the
economic literature that values hypertension and AMI includes only medical costs and
foregone earnings Since the costs of these illnesses encompass a wide range of lifestyle
changes and other effects noi captured by medical costs and foregone earnings, these
economic valuations may be considered a lower bound estimate of willingness to pay to
avoid these health endpomts
The value for premature mortality relies on willingness-to-pay methodologies that
capture all the benefits of av ending mortality nsks However, there exist a range of
estimates for valuing a statistical life, and the estimates used by EPA have a wide range
of uncertainty
Hypertension in Adults and Children
Hypertension is typically diagnosed based on changes in blood pressure Elevated blood
pressure is a major nsk factor in heart attacks and strokes and is normally treated b\ close
medical monitoring, drugs and changes in diet and exercise In the MP&M RLA Abt
Associates (2003) reported an estimate of $1,141 ($2001) for adult hypertension that
included medical costs and work-loss days
This underestimates the WTP to avoid hypertension, since it does not include any
measure of WTP to avoid the pain, suffering, and stress associated with the condition
both for the individual and his family Nor does it value the direct costs of changes in
behavior (salt-free foods, exercise regimes) or the loss in satisfaction associated with
these changes It also fails to account for any side effects from medications, which can
range from nausea and vomiting to impotence, depression, and even cancer
The estimates for hypertension apply to adults However, the study by Serensen et al
(1998) linked mercury to hypertension in children age 7 Although protocols for treating
children exist, studies that evaluate Ihese costs and provide estimates of costs of treatment
for children do not appear to exist For most children treatment for high blood pressure
starts with diet and exercise Ph> sicians are generally reluctant to initiate a course of
drug treatment that would most likely extend into adulthood, and it is unlikely that all
cases above the 95th percentile would qualify Exactly what fraction of the cases would
require drug treatment and for how long is not known
Treating hypertension in children is likely to cost less than for adults The medical costs
would probably be less, since children would typically receive a smaller dosage of drugs
Nor would work-loss days apply to children, although children would suffer the analog in
53
-------
school-loss days The analysis values a case of hypertension in children by making the
following assumptions
Annual medical costs for children are half ihe value calculated for adults
through age 18, about $570 ($2001),
Treatment would extend to age 30 with medical costs after age 18 valued at
the annual cost of medical treatment for adults, $1,141 ($2001),
50 percent of Ihe cases would require medical supervision and drug treatment
Inflating the estimates to 2003 dollars, annual medical costs amount to $616 ($2003) per
year from age 7 to 18and$l,181 ($2003) from age 19 to 30 Discounted at 3 percent, the
present value is $ 12,040 ($2003) Annualized over the period from birth to age 30, it
amounts to about $614 ($2003) per year The analysis applies this annualized estimate to
half ihe cases This value is likely to be an underestimate in terms of WTP for the same
reasons as given above In addition, costs of treatment may extend further into adulthood
accompanied by increasing nsks of heart attacks and strokes A more precise estimate of
Ihe value of avoiding childhood nsks of hypertension requires additional research to
resolve Ihese issues
Acute Myocardial Infarction (AMI)
Cases of acute myocardial infarction (AMI) require treatment in emergency rooms and
intensive care facilities followed by regular monitoring, treatment, and usually lifestyle
changes Medical protocols for AMIs are well developed, and medical insurers have
extensive data on the hospital costs and foilow-up costs of drugs and other treatments for
AMI victims These cost of illness (COI) measures provide a minimum estimate of the
AMI endpomt
In a Technical Support Document to the Off-Road Diesel RIA Apelberg et al (2003)
summarized the COI estimates for non-fatal acute myocardial infarction (AMI) The COI
estimates exhibited a wide range from about $22,000 to $ 109,000 ($2000) The authors
were not able to resolve the inconsistencies in the estimates and averaged the cost
estimates for two studies whose treatment tmeframe extended five years after the event
The resulting midpoint estimate amounted to $65,293 ($2000) In addition, Apelberg et
al (2003) estimated the opportunity cost of lost earnings discounted at 3 percent at
$12,932 for the 45 to 54 age cohort The midpoint of this age cohort approximates the
midpoint for the 40 to 59 age cohort, and combining the medical cost estimate and cost of
lost earnings yields a total cost of $78,834 ($2000) per event
The uncertainties in both the estimates of medical costs and lost earnings are significant
There is a large range in estimates of the costs of treating a heart attack, and this
uncertainly needs to be resolved In addition, the study that Apelber et al (2003) used as
a basis for their estimate of lost earning utilized data from 1977 Given the changes in
both the economy and in treatment of medical care since that time, relying on data from
the 1970s is likely to introduce significant biases
54
-------
In addition to the uncertainties associated with (he COI estimate, relying on costs avoided
as a measure of benefits is h kely to underestimate the change in economic welfare This
is because an AMI event generally requires a change in lifestyle with restrictions on diet
and exercise These changes represent a significant loss in welfare, and COI-based
estimates do not capture the costs associated with these lifestyle changes However, no
CV studies exist that estimate willingness to pay to avoid nsks of AMI
Avoiding Premature Mortality
There are a number of studies lhat provide estimates of an individual's willingness to pay
to avoid premature mortality These estimates represent the value of a statistical life
(VSL) EPA has relied on a number of these studies in vanous Regulatory Impact
Analyses (RIAs) since the 1980s Currently, U S EPA (2000) recommends a value of
$6 5 million ($2001) in its Guidelines for Preparing Economic Analysis, and this study
employs a value consistent wilh that document
This VSL estimate denves from an analysis of 26 studies using both contingent valuation
(CV) methods and wage-nsk approaches (U S EPA, 1999) CV methods employ survey
techniques to elicit information on willingness to pay to avoid some nsk of premature
death Wage-nsk studies estimate the value of a statistical life based on the wage
premium for workers in higher nsk j obs The transferability of these studies rests on the
assumption that WTP for mortal nsk reduction within the range of nsks considered is
linear with respect to nsk reduction This Iineanty assumption means that if a 1/100,000
chance of dying is valued at $10, then a 10-fold increase to 1/10,000 is valued at $100
There is some evidence that the EPA value may overestimate the value of avoiding
premature death A meta-analysis of 33 wage-nsk studies (Mrozek and Taylor, 2002)
found that the VSLs from some revealed preference studies overestimated WTP for nsk
reduction by not accounting for inter-industry wage differentials After controlling for
job characteristics omitted in some studies, the authors estimated a VSL in a range of
about $1 5 to $2 5 million ($1998) However, anumber of studies have also produces
VSL estimates above the current EPA guideline VSL
Another important issue that c onttnues to command attention from VSL researchers is the
question of whether older adults, especially those in poorer health and at greatest nsk due
to circulatory or respiratory conditions, value life differently than younger people whose
values are captured in surveys and wage-nsk studies A recent study by Albenni et al
(2002) using a CV format found no statistically significant evidence that baseline health
negatively affected valuations in the U S sample The study did find that the magnitude
of the coefficients on ihe age dummies for respondents aged 60-69 and 70-79 declined
relative to age 50-59 but the coefficients were not statistically significant
4.4.2 Summary of Economic Valuations
Table 4-12 updates the economic valuation of health effects to 2003 dollars using the
Consumer Price Index (CPI) The valuations for IQ deficits, compensatory education,
55
-------
adult hypertension, acute myocardial infarction, and premature death include a mixture of
medical and non-medical costs Given this mix, the analysis uses the CPI for all items to
inflate these values to 2003 dollars
Table 4-12
Summary of Economic Valuations of Mercury Health Endpomts
($2003 per year)
Health Endpoint
IQ Deficits
Compensatory Education
Adult Hypertension
Child Hypertension
Non-fatal AMI
Premature Death
Valuation (Scurrent/year)
$9,41 9 ($2001) per IQ point
$58,012 ($2001) per case
$1,141 ($2001) per case
$78,834 ($2000) per case
$6,500,000 ($2001) per death
Valuation (S2003/year)
$9,749
$60,042
$1,181
$614
$81,000
$6,700,000
Notes: Consumer Pnce Index (CPI) for all items 2000= 1 72 2, 2001 = 177 1 , 2003 = 1 83 3
4.4.3 Health Benefits of Reducing Mercury Exposure
This section applies the economic valuations in 2003 dollars to the estimates of risks and
cases avoided for the mercury health endpomts Reduced mercury exposure results in a
reduction in the incidence of vanous health endpomts, and the tables below calculate the
benefits of the reduced exposures associated with each of the three mercury reduction
scenarios As a reminder, scenario 1 is a 30 percent reduction in U S emissions, scenario
2 eliminates all U S emissions, and scenario 3 reduces average fish methylmercury
concentrations in all but the top predator group to 0 3 ppm, which represents a reduction
in U S emissions of about 51 percent
4.4.3.1 Benefits of Avoiding IQ Deficits
Fetal mercury exposure results in nsks of neuropyschological and neurodevelopmental
impairment The New Zealand study provided the data for estimating dose-response
functions that yielded estimates of IQ points lost Tables 4-13a and 4- 13b estimate the
value of those lost IQ points based on dose-response functions without and with
thresholds Table 4- 13a estimates that the savings in IQ points in scenario 1 results in a
benefit of about $78 million ($2003), scenano 2 estimates a benefit of about $260 million
($2003), and scenano 3 values the IQ points saved at about $135 million ($2003)
Evaluating the dose-response function with a threshold, Table 4-13b yields estimates of
benefits of about $107 million ($2003) for scenanos 1 and 3, and about $161 million
($2003) for scenano 2
56
-------
Table 4-13a
Valuation of IQ Points Saved - Without Threshold
(millions $2003 per year)
Reduction Scenano 1
Reduction Scenario 2
Reduction Scenano 3
Valuation of saved IQ points - No threshold ($2003)
ZoneO
$159
$507
$268
Zone I
$259
$829
$439
Zone II
$206
$718
$364
Zone III
$155
$541
$274
Zone Total
$780
$2595
$1345
Table 4-13b
Valuation of IQ Points Saved - With Threshold
(millions $2003 per year)
Reduction Scenano 1
Reduction Scenano 2
Reduction Scenano 3
Valuation of saved IQ points - With threshold ($2003)
ZoneO
$406
$610
$406
Zone I
$664
$997
$664
Zone II
$00
$00
$00
Zone 111
$00
$00
$00
Zone Total
$107 1
$1606
$107 1
4.4.3.2 Benefits of Avoiding the Costs of Compensatory Education
The mercmy reduction scenarios result in fewer children with IQs less than 70, who
require compensatory education Table 4-14 summarizes the benefits of avoiding those
costs of special education, \\ hich amount to about $60,000 ($2003) per year Under
scenario 1 the costs avoided amount to about $0 5 million ($2003) Scenano 2 results in
benefits of about $1 6 million ($2003), and scenano 3 totals about $0 8 million ($2003)
Table 4-14
Valuation of Cost of Special Education Avoided fot Children with IQ<70
(million $2003 per year)
Reduction Scenario 1
Reduction Scenario 2
Reduction Scenario 3
Valuation of people with IQ less man 70 by Zone
ZoneO
$010
$031
$017
Zone I
$016
$051
$027
Zone II
$013
$045
$023
Zone III
$010
$034
$017
Zone Total -
$048
$1 61
$084
4.4.3.2 Benefits of Avoiding Adult Hypertension and Cardiovascular Events
Mercury contributes to coronary and cardiovascular disease, including AMI and ACM
endpoints Estimates of cases avoided reflect dose-response functions based on one study
so that the resulting benefits must be treated with extreme caution
57
-------
Table 4-15 summarizes the benefits of avoiding nan-fatal AMI These estimates are
highly uncertain, since the AMI coefficient was not significant at p< 05 The purpose of
presenting these benefits is to provide some indication of the possible magnitude of the
AMI endpomt as a means of prioritizing future research In scenario 1 the annual
benefits of avoiding non-fatal AMI cases amounts to about $6 million ($2003) Scenarios
2 and 3, yield annual benefits of AMI cases avoided of about $21 million (($2003) and
$10 million ($2003), respectively
Table 4-15
Valuation of Non-fatal Acute Myocardial Infarction Cases Avoided
(millions $2003 per year)
Reduction Scenario 1
Reduction Scenario 2
Reduction Scenano 3
Non Fatal AMI Incidence
ZoneO
$1 2
$40
$20
Zone I
$19
$63
$3 1
Zone II
$17
$60
$29
Zone III
$13
$44
$21
Zone Total
$61
$207
$102
Table 4-16 summarizes the results of applying die economic valuation of pre-mature
death to the number of deaths avoided by men age 40 to 59 EPA currently recommends
a value of about $6 7 million ($2003) per life lost Scenano 1 generates benefits of about
$0 5 billion ($2003) per year Scenano 2 yields annual benefits of about $1 8 billion
($2003), and scenano 3 generates about $0 9 billion ($2003) in annual benefits
Table 4-16
Valuation of Pre-mature Deaths Avoided
(millions $2003 per year)
Reduction Scenario 1
Reduction Scenano 2
Reduction Scenano 3
All Cause Mortality
ZoneO
$1054
$3513
$1733
Zone I
$1657
$5524
$2726
Zone II
$1529
$531 9
$2583
Zone HI
$1105
$3844
$1867
Zone Total
$5346
$1,8200
$8909
The estimates of benefits of mortality avoided are very large, and readers should
remember that underlying these values is tiie Salonen et al dose-response function with
its high level of uncertainty Nevertheless, the magnitude of the numbers does indicate
the potential significance of this health endpoint Additional benefits would accrue rf
cardiovascular nsks of mercury exposure affect men in other age groups, as is suggested
by Guallar (2002) and if mercury contnbutes to nsks of ACM in women These
uncertainties emphasize the need for additional research to venfy the applicability of
Salonen et al's dose-response functions to a U S population that includes men and
women
58
-------
4.4.3.4 Child Hypertension
Valuing a case of hypertension in children at $614 per case yields fairly small benefits of
cases avoided, as shown in Table 4-17 Scenario 1 yields no benefits, since there are no
cases avoided Scenarios 2 and 3 yield estimate of about $0 7 million ($2003) and $0 3
million ($2003), respectively Limiting the dose-response function to pregnant women
with blood-mercur} concentrations greater than 10 ug/L underestimates impacts so lhat
the values in Table 4-17 are a lower bound However, the dose-response function relies
on a single study so that theie is a wide confidence range around these estimates
Table 4-17
Benefits of Avoiding Hypertension in Children
(millions $2003 per year)
Reduction Scenario 1
Reduction Scenario 2
Reduction Scenario 3
Valuation of Hy
ZoneO
$0
$02
$01
Zone I
$0
$03
$02
pertension Cases Prevented
Zone II
$0
$02
$00
Zone III
$0
$01
$00
Zone Total
$0
$07
$03
4.4.4 Summary of Health Benefits of Reducing Mercury Exposure in the Study Area
Table 4-18 summarizes the benefits of avoiding the neurodevelopmental and
cardiovascular health endpoints for the three mercury reduction scenanos These benefits
apply to the study area with about 22 5 million people, but are not transferable to other
populations due to the mix offish species consumed and the mercury concentrations
found in those species
The benefits of avoiding fetal health endpoints include both the avoidance of 1Q losses
and costs of compensatory education for children with IQs below 70 The benefits for IQ
deficits derive from a dose-response function reviewed by the National Research Council
(2000) The total for these two benefits amounts to about $78 5 million ($2003) in
scenano 1, about $2611 million ($2003) in scenario 2, and $135 3 million in scenario 3
The cardiovascular endpoints acute myocardial infarction and all cause mortality in men
age 40-59, and hypertension in children, are subject to considerably more uncertainty
The estimates rely on single studies for dose-response Junctions, and the studies have not
been subject to the same rigorous peer review as the studies of neurodevelopmental
effects The ACM endpornt dominates these benefits, since the value of avoiding a pre-
mature death is $6 7 million ($2003) The total of benefits of avoiding Ihe cardiovascular
endpoints amounts to about $0 5 billion ($2003) for scenano 1, $1 8 billion ($2003) in
scenano 2, and $0 9 billion ($2003) in scenano 3
59
-------
Table 4-18
Summary of Benefits of Reducing Mercury Exposure
(millions $2003 per year)
Reduction Scenario 1
Reduction Scenario 2
Reduction Scenario 3
Benefits of Avoiding Mercury Health Endpomts
IQ points
Saved
$780
$2595
$1345
IQ<70
$05
$16
$08
AMI
$61
$207
$102
ACM
$5346
$1,8200
$8909
Child
Hypertension
$0
$07
$03
Total
$6192
$2,102 5
$1,036 7
4 5 Assumptions and Uncertainties
The estimates of benefits from avoiding health risks associated with exposure to mercmy
rely on a standard damage function methodology that proceeds from changes in exposure
to changes in nsks to valuation of risks avoided The discussion in the sections above
identifies a number of key assumptions and uncertainties, and this section summanzes the
most important ones
Estimates of mercury exposure from consuming marine fish and shellfish depend on the
data on mercury fish concentrations, by species, from the MML database These data for
the South Atlantic area comprise over 3,000 observations for over 100 species However,
some species are not well represented in the database, and lack of sufficient samples may
result in some bias in the estimates of mercury concentrations for some species hi
addition, the analysis employs estimates of mercury concentrations for groups of species,
such as drums and snappers, and weights these estimates by the number of samples A
more appropnate weighting would be b> consumption, but detailed information by
species was not available from NHANES The estimates of mercury concentrations may
contain some bias, but the magnitude and direction are not known
The greatest uncertainty associated with the data on fish consumption is the accuracy of
respondent recall NHANES asked respondents to recall the number of meals of fish, by
species, over a 30-day period A 30-day period provides a more accurate estimate offish
consumption over extended penods, especially for consumers who eat fish less regularly,
but the one-month recall period makes heavy demands on respondents' memories Nor
did NHANES collect data on portion sizes In contrast, the Flonda fish consumption
survey required respondents to recall fish and shellfish consumption over a shorter penod
and asked for estimates of portion sizes The 7-day recall penod probably results in more
accurate data, but the shorter recall penod requires a larger sample size to yield
equivalent accuracy for estimates of daily fish consumption These two surveys of fish
consumption both have result in some uncertainty, but the magnitude and direction of any
bias resulting from the questionnaire and sampling design are unknown
In addition to the accuracy of the estimates offish meals consumed, baseline mercury
intake depends on a number of assumptions used to translate data on species consumed
60
-------
from NHANES into data appropriate for the study area It is possible that the baseline
amount offish consumed is less in 2003 than in 1999/2000 when Ihe data were collected,
since U S FDA issued national fish consumption advisones in January 2001 Translating
fish consumption into mercury intake relies on a number of assumptions to relate locally
available species to NHANES estimates of consumption by more generic species groups
These assumptions, which are discussed in Section 4 1 3, may result in biases in baseline
mercury exposure, but the magnitude and direction of any bias are unknown
An important assumption is (hat the direct effect of reductions in mercury applies equally
to an Atlantic and Gulf species, regardless of location Output from mercury air
deposition models would provide more precise estimates of the spatial vanabihty in
mercury deposition, although the lack of knowledge of migratory patterns of many
species limits the usefulness of such information The magnitude and direction of any
bias from this assumption is unknown
The analysis uses data on U S landings from Guh7Atlantic and Pacific areas, imports
and exports to estimate the percentage of consumption and corresponding mercury intake
attributable to species harvesi ed in areas affected by reductions in U S mercury
emissions The calculation relies on value rather than quantity, since the former provides
a better comparison with imports and exports, which include a wide variety of processed
fish products Given the limited resources available, this study estimates these
percentages on a highly aggregate basis A more disaggregate analysis might produce
different percentages of consumption attributable to areas affected by U S emissions
reductions However, the magnitude and direction of any bias are unknown
The dose-response functions employed to estimate various health effects are also a source
of uncertainty The mercury dose- response functions for paraesthesia and fetal effects
have been subject to extensive peer review by scientific experts However, the dose-
response functions for hypertension and cardiovascular disease endpoints have not been
subject to the same scrutiny Moreover, the dose-response results of these studies have
not been replicated in the U S or elsewhere Hence, the estimates of cases of
hypertension, AMI, and ACM that derive from these dose-response functions are highly
uncertain
Finally, estimates of the benefits of avoiding mercury health endpoints that rely solely on
medical costs and lost wages undervalue the overall willingness to pay by failing to
include impacts that range from pain and suffering to changes in lifestyle The largest
benefit estimates result from a voiding pre-mature death, which is valued at $6 7 million
($2003) This value of a statistical life is the source of ongoing debate, and benefits
derived from this value are subject to some uncertainty, but die direction of any bias is
unknown
61
-------
Appendix 4-A
Calculation of Baseline Risks of Non-fatal AMI and Pre-mature Death from
Salonen et al. (1995) Study
ALL CAUSE
MORTALITY
Query Descnotion
Data for Years:
Location:
Ages:
Race:
Gender:
Grouped by:
Crude Rate Calculated
pei
Compressed Mortality
1998-1998
The United States (FIPS=00)
35-54 years through 55-64 years
All Races
Males
Age
100,000
Using the following ICD
Codes:
[000.0-999.8]
Age
Description
35-44 years
45-54 years
55-64 years
Death Count
57,139
91,727
140,134
Population
22,231,580
17,273,166
11,021,108
Crude Death Rate
2570
531 0
12715
Total deaths=289,000
Estimated rates based on
the above.
Age
Description
40-44 years
45-54 years
55-59 years
40-59 years
Death Count
28,570
91,727
70,067
190,364
Population
11,115,790
17,273,166
5,510,554
33,899,510
Crude Per capita
Death Rate
00026
00053
00127
0005615524
Source Downloaded from CDC WONDER at
http //wonder cdc gov/mortICD9J html on 11/14/2003
62
-------
Appendix 4-B
Calculation of Fraction of Children Age? Exceeding 95th Percentile as a Result of an
Increase in Systolic and Diastolic Blood Pressure
from
S0rensen et al., (1999) Study
Systolic Blood Pressure for boys and Diastolic Blood Pressure for boys and
girls girls
50th 98 50th 59
95th 117 95th 76
std 1155085 std 1033497
z-score percentole BP Delta BP z-score percentile BP Delta BP
16449 9500% 117 16449 9500% 76
03809 6484% 1024 146 03000 6179% 621 139
0 5000% 98 0 5000% 59
Note: The blood pressures for the 50th and 95lh percentile were estimated from a
graph for age 7
Source: National Heart, Lung, and Blood Institute, 1987, Report of the second task force
on blood pressure control in children, Bethesda, MD
63
-------
5 0 Commercial Fishing
Fish consumption advisories (FCAs) potentially affect consumer behavior and demand
for fish and fish products Any change in demand results in a new equilibrium pnce and
quantity, which affects the entire fishing industry fishermen, wholesalers, retailers, and
consumers Economists measure benefits by evaluating changes in economic surplus that
result from shifts m demand or supply Any changes in FCAs that allow for more
consumption of marine fish, such as king mackerel, may be expected to cause an increase
in demand, which would generate increases in economic surplus for all sectors of the
industry
This chapter has five sections Section 5 1 discusses the structure of the commercial
fishing industry with reference to the South Atlantic area Section 5 2 analyzes the effect
of mercury FCAs on demand and supply and reviews relevant literature Section 5 3
empirically estimates the effect of U S Food and Drug Administration's fish
consumption advisories for king mackerel and swordfish Section 5 4 considers the
welfare effects of FCAs on consumers and suppliers, and section 5 5 discusses the data
gaps and uncertainties
5.1 Structure of the Commercial Fishing Industry
The U S fishing industry compnses three vertically-integrated sectors harvesting,
wholesaling/processing, and retailing The harvesting sector includes fleets with
specialized gear for targeting different species The swordfish fleet employs long line
technology Crab and lobster boats use pots and traps, and commercial reef fishermen
use rods/reels, bandit reels, and trolling lines
In the South Atlantic area commercial fishermen operate out of ports from North
Carolina to south Flonda Beaufort-Morehead City, NC, and Wanchese-Stumpy Point,
NC, ranked 1 Ith and 31st, respectively, in landings by quantity Key West, FL, which is
home to the reef fish fleet, ranked 39th in quantity, but 8lh in value
Wholesalers and processors provide the essential link between port and retail They
purchase raw fish at auctions and at dockside for resale to processors and retailers,
including supermarkets, fishmongers, and restaurants The largest wholesale auction in
the U S takes place at New York City's Fulton Fish market, but smaller wholesale
auctions occur in most major fishing ports
The processing sector includes canneries and other processors Large east coast
processors, such as Coldwater Seafood in Cambridge, MD, and Gorton's in Gloucester,
MA, buy directly from local fishermen, wholesalers, and importers They process raw
fish into a variety of products, such as canned fish, fillets, breaded products, fish sticks,
frozen dinners, and fish meal products
The retail sector ranges from small restaurants and specialty fish shops to national fast
food, restaurant and supermarket chains These enterprises sell a wide variety of
64
-------
products that include fresh seafood dinners, fresh whole fish and fillets, canned fish, and
breaded and frozen packaged fish products
5.2 Impact of Fish Consumption Advisories (FCAs) on Demand and Supply
The timing of FCAs affects >:onsumer demand, and it is important to understand when
information first came to the attention of consumers In January 2001 U S FDA issued
widely publicized mercury advisories for king mackerel, swordfish, sharks, and ulefish
The FDA advisory recommended no consumption for women who are pregnant or may
become pregnant, nursing mothers, and young children Pre-dating this recent advisory,
FDA had issued an advisory in 1995 that recommended' 1 meal per month' of swordfish
and sharks for pregnant women and women who might become pregnant However, as
early as 1971 the United Star.es had banned imported swordfish due to high mercury
content (Lipton, 1986) Consequently, some consumers may have already substituted
away from swordfish and other species well before the more recent state and FDA fish
consumption advisories
State health departments in the study area have also issued separate mercury FCAs
Fbnda Department of Health issued its first marine fish consumption advisory in May
1991, urging limited consumption of all shark species (Flonda Marine Research Institute,
undated) A second advisory recommended limited consumption of five species from
certain portions of Flonda's coastal waters In June 1996 Flonda issued an advisory for
limited consumption of king mackerel caught in the Gulf of Mexico In March 2000
Flonda, Georgia, South Carolina, and North Carolina issued a joint fish consumption
advisory for king mackerel in all coastal waters (NCDHHS, 2000) This advisory
recommends limiting consumption of king mackerel 33-39 inches and no consumption
for fish greater than 39 inches.
In addition, other Atlantic states have issued FCAs for tuna, stnped bass, and bluefish,
and knowledge of these advisories may also affect consumers in the study area
Table 5-1 summanzes the timetable of state and FDA fish consumption advisones How
these advisones affect demand for these species and other fish depends on consumer
awareness of and response to the information
65
-------
Table 5-1
Summary of Federal and State Fish Consumption Advisories for Saltwater Species
Agency
Florida
US FDA
Flonda
Florida
Flonda
Georgia
South Carolina
North Carolina
US FDA
Date
May 1991
1995
October
1995
June 1996
March 2000
January
2001
Species
Sharks
Swordfish
Sharks
Gafiftopsail catfish
Crevallejack
Spotted seatrout
Ladyfish
Spanish mackerel
from certain areas
King mackerel
from Gulf of
Mexico
King mackerel >39
King mackerel
Swordfish
Sharks
Tilefish
Advisory
'1 meal/month'
'1 meal/month'
'1 meal/month'
1 meal/month'
'eat none'
'eat none'
Population
Pregnant
women, nursing
mothers, and
young children
Pregnant
women or
women of
childbeanng age
Pregnant
women, nursing
mothers, and
young children
Pregnant
women, nursing
mothers, and
young children
Pregnant
women, nursing
mothers, and
young children
Pregnant
women, nursing
mothers, and
young children
Source U S Food and Drug Administration, Consumer Advisory, March 2001,
http //vmcfsan fdagov/~dms/admehg html
North Carolina Department of Health and Human Services, press release,
3/23/2000. hito l/\\\\\\ dhh-> stale nc us/resssrel/3-23-00 htm
Flonda!
Marine!
Flonda P
Fishes o
http /Aw
7ish and Wildlife Conservation Commission, undated, Mercury in
:ish. http Vmannehshenes onyPubs/mcrcurv htm.
vianne Research Institute, undated, Mercury in Marine and Estuanne
'Flonda. An Overview,
vw flondamanne org/dg Its/id 9466/features view article htm.
5.2.1 Awareness of FCAs
There is limited information on awareness of FCAs by the general fish-consuming public
Northeast States for Coordinated Air Use Management (NESCAUM) undertook a survey
offish consumption in 1999 (Round, 2002) that included information on awareness of
66
-------
risks of consuming bolh freshwater and saltwater fish The suney targeted three at nsk
groups anglers, ethnic groups with high fish consumption (e g Hispamcs), and women
of childbeanng age Included in the questionnaire was a question that asked 'have you
ever read or been told that people should limit the amount of some fish they eat due to
mercury contamination'7' The question asked specifically about awareness for both
freshwater and saltwater fish
Table 5-2 summarizes the results of this question for various groups of consumers
Nearly 80 percent of men were aware of the nsks of consuming saltwater fish, but only
about 50 percent of women were aware of the nsks About 66 percent of anglers were
aware of the nsks of consuming some species of saltwater fish Among Hispamcs,
awareness was nearly 60 percent However, of the women of childbeanng age,
awareness of die nsks of consuming saltwater fish was less than 50 percent
Table 5-2
Awareness of Risks of Consuming Saltwater Fish
Response
Yes
No
Total
Men
147
38
185
Women
129
128
257
Anglers
57
30
87
Women
CBA
80
85
165
Hispanics
34
29
63
Total
276
166
464
Source M Round, NESCAUM, personal communication, 1 1/27/03
The NESCAUM survey represents the responses of a fairly small sample of 464 adults in
Massachusetts so thai its findings are not necessarily transferable to other populations In
addition, as of 1999 Massachusetts fish consumption advisones applied only to
freshwater fish, and the survey pre-dated FDA's announcement of FCAs for four
saltwater species As a result this survey has limited value as an indicator of current
levels of consumer awareness of mercury nsks from consuming marine fish
5.2.2 Consumer Response to FCAs
Behavioral adjustments by consumers may affect demand for species named in FCAs or
may apply more broadly to all fish from certain areas If concerns about mercury have
caused consumers to shift awjy from a particular species or from consuming fish in
general, reduced mercury loadings should result in increased demand With normally
shaped demand and supply curves any increase in demand and the associated nse in
pnces across some or all species will result in economic surplus gains to both consumers
and suppliers, including fishermen, wholesalers, and retailers
Since FDA announced FCAs in 2001, there has been limited time to assess consumers'
response The available NHANES data, including estimates of number of meals of fish
consumed, only cover the years 1999 and 2000 which pre-date the FDA advisory
However, NHANES data for 2001 and 2002 will be available in 2004, and the data for
67
-------
the penod 1999-2003 should provide a basis for comparing consumption before and after
the FDA advisory
There is some evidence that FDA's advisory has resulted in a reduction in fish
consumption by pregnant women Oken et al (2003) compared dietary surveys by 2,235
pregnant women in Massachusetts prior to and after 2001 to assess changes in fish
consumption The dietary questionnaire asked about fish consumption in the previous
three months, and the authors divided responses into a pre-advisory penod, 1999-2000, a
"wash-in" penod, January-March, 2001, and a post-advisory penod, Apnl 2001-February
2002 They reported lhat overall consumption of fish consumption for pregnant women
declined by 1 4 servings per month, or about 18 percent, from the pre-advisory mean of
7 7 meals per month Moreover, this decline affected all types of seafood
canned tuna declined about 0 6 servings per month,
dark meat fish declined about 0 3 servings per month,
white meat fish declined about 0 3 servings per month, and
shellfish declined about 0 2 servings per month
In addition, time senes analysis of the consumption data revealed that pnor to the
advisory consumption of all types of fish had been increasing over time (+0 8 percent per
month) but that in the post-advisory penod the slope turned negative (-2 1 percent per
month) This trend had not leveled off as of January 2002 (the last month included in the
analysis), and it is not clear at what level consumption offish by pregnant women will
reach a new equilibnum.
This research suggests that among pregnant women the FDA advisory has resulted in a
significant decline in fish consumption Moreo\er, tins decline in consumption appears
to affect all types of fish, including shellfish and tuna that were not included in the FDA
advisory This suggests that the overall effect of FCAs on commercial fishing may
extend to other species not specifically identified in fish advisones
5.2.3 Impact of FCAs on Demand for Fish
Figure 5-1 shows the change in surplus attnbutable to the removal of an FCA that shifts
the demand curve outward with no effect on the supply curve The area, CS (shaded dark
gray), under initial demand curve, D, but above equilibnum pnce, P, is (he baseline
consumers' surplus And the area, PS (vertical lines), above the supply curve, S, but
below pnce, P, is the baseline producers' surplus Expanding demand to D* results in a
new equilibnum at pnce, P*, and quantity, Q* The new consumers' surplus is the area
CS* under the new demand curve, D*, but above the new pnce, P*, and the new
producers' surplus is the area PS* above the supply curve, S, but below the new
equilibnum pnce, P* The increase in consumers' surplus is the light gray area (part of
CS*) above and to the nght of the onginal demand curve, D The increase in producers'
surplus is slant-line area (part of PS*) that lies between the old pnce, P, and the new
pnce, P* The increase in economic surplus is the increase in both consumers' and
producers' surplus
68
-------
Figure 5-1
Consumers' and Producers' Surplus
Changes Resulting from Elimination of an FCA
Q Q!
Note Elimination of FCA increases demand
69
-------
5.2,4 Measuring Changes in Economic Surplus
Economists estimate tone senes econometric models of demand and supply functions to
measure changes in consumers' and producers' surplus In the case offish consumption
advisories or any other policy change the standard approach is to investigate the market
for a species before and after the change This requires estimating the demand curve with
a dummy vanable that reflects changes in FC As and captures any shift in demand and the
resulting change in equilibnum pnce and quantity However, changes in equilibnum
price and quantity at the ex vessel level also affect wholesalers and retailers, and precise
economic measurements of changes in welfare must consider those sectors
Just and Hueth (1979) provide an analytical framework for evaluating consumer and
producer surplus in vertically integrated markets They demonstrate that if normal
(downward-sloping) demand conditions apply at ex-vessel, wholesale, and retail levels
then the consumers' surplus at the ex-vessel level captures the overall benefits across all
sectors Similarly, if normal (upward-sloping) supply conditions apply in all sectors, then
the producers' surplus at the retail level captures the overall benefits for all producers In
practice, data are much more readily available to estimate demand and supply functions
for the harvesting sector using ex-vessel pnces However, supply curves at the ex-vessel
level do not capture producers' surplus for the wholesale or retail sectors, and economists
often must be satisfied with underestimating the producers' surplus or using ad hoc
approaches to estimate producers' surplus for retailers (see Strand et al, 2002)
Another approach to quantifying impacts on consumer demand relies on contingent
valuation surveys that ask consumers how they would react to FCAs on commercially-
important rnanne species Following Jakus et al (2002), this information could be
applied to shift the demand curve and calculate changes m consumers' and producers'
surpluses
A review of the economic literature on commercial fishenes revealed only one study that
attempted to value the welfare effects of FCAs on commercial fishing (Rae, 2002)
Maryland Department of Natural Resources commissioned a study that estimated
potential economic losses to commercial fishermen if an FCA were issued for stnped
bass in the Chesapeake Bay (Jakus et al, 2002) The section below describes this study
in more detai
5 2.4.1 Maryland Stnped Bass Study
For Maryland Department of Natural Resources Jakus et al (2002) estimated the
potential benefits of eliminating mercury FCAs on stnped bass in the Chesapeake Bay
The study considered health effects and welfare effects on recreational and commercial
fishermen
To estimate the impact of FCAs on commercial fishing the authors simultaneously
estimated demand and supply functions for stnped bass in a two-stage model The
demand function estimated stnped bass catch as a function of pnce, household income,
70
-------
CPI of meat substitutes (beef, chicken), and population Supply (catch) was a function of
price, pnce of substitutes (oysters and catfish), stnped bass population, and number of
commercial licenses The quantitative analysis suffered from senous gaps in the data
There were only 13 annual observations due to closure of the fishery for five years from
1985 to 1989 and missing data for several independent variables
The coefficients for variables in both models generally had the correct signs, although not
all were statistically significant The authors used a linear specification to compute
consumers' and producers' surpluses at the equilibnum pnce and quantity Evaluated at
the predicted equilibnum pnce and quantity, annual consumers' surplus amounted to $3 1
million ($2000) and producers' surplus to $3 1 million ($2000) for a total of $6 2 million
($2000)
To estimate the loss in economic surplus attributable to an FCA the authors assumed a 7
percent loss in consumers' surplus, based on a value denved from a study of the impacts
of FCAs on recreational anglers in Tennessee The loss m consumers surplus amounted
to about $0 22 million ($2000) Using the new equilibnum pnce and quantity, they
recomputed the producers' surplus and estimated the change amounted to a welfare loss
of about $0 30 million ($2000), or about 10 percent
The study demonstrates a wcrkable approach to measuring welfare impacts of changes in
FCAs if there exist reliable estimates of the impact of FCAs on consumer demand The
estimate of benefits to commsrcial stnped bass fishing from removal of an FCA rests on
the tenuous assumption that consumers of commercial fish, most of whom are not
anglers, value the loss due to FCAs the same as freshwater anglers As a result,
transferring benefits from this study to commercial fishing in the South Atlantic is
fraught with potential bias
52.42 Other Estimates of Consumers' and Producers' Surplus
National Marine Fisheries Service collects data on quantity and value of landings, by
species and port. These data are readily available back to the 1970s, and it is fairly
straightforward to estimate species-specific demand and supply functions at the ex-vessel
level Wessells and Anderson (1992) provide a useful summary of the state of the art in
estimating demand for seafood and analyzing ex-vessel, wholesale, and retail markets
However, because of changes in the industry due to overfishing, quotas, and other factors
many of these studies are out of date Thus, the available literature is of limited use in
transferring benefits The studies summarized below provide estimates of economic
surplus for some species harvested in the South Atlantic study, but their primary utility is
to demonstrate applicable methodologies
Swordfish Import Ban
Lipton (1986) analyzed the effect of a 1971 announcement by U S FDA that imported
swordfish had exceeded the criterion value for mercury and were subject to inspection
71
-------
and possible seizure News reports warned consumers of the dangers of consuming
swordfish, and consumption declined 96 percent to less than 0 002 pounds per capita in
1971 However, domestic swordfish were not subject to FDA inspection unless shipped
interstate, and the FDA action promoted growth of the domestic swordfish industry
After 1971 U S consumption of swordfish increased steadily and by 1983 had returned to
pre-regulatory levels
Lipton analyzed the impact of the FDA regulatory action by estimating the demand for
swordfish using a specification that included a lag term for the effects of the consumer
response to the dangers of consuming mercury The demand equation regressed per
capita swordfish consumption on swordfish pnce, per capita disposable income, CPI for
beef, and a dummy in 1971 with a five-year lag penod to capture the impact of the FDA
action Lipton reported results for a semi-log specification where all vanables were
adjusted for inflation The results of the econometric analysis showed that the effect of
the FDA action on consumer demand was negative and highly significant, but the
magnitude of the effect declined each year Without publicity that reminded consumers
of the nsks of consuming swordfish the effect dissipated after five years This study is
useful as an approach to modeling the demand for swordfish, but it examines the effect of
a temporary phenomenon, much like pollution spills, whose effects dissipate over tme
Current publicity on mercury nsks is widespread and continuing, and the FCAs provide
detail on the nsks to pregnant women and fetuses
Stnped Bass Landed in New York
Buerger and Kahn (1989) estimated the value of stnped bass caught by New York
commercial fishermen by evaluating consumers' and producers' surplus The authors
estimated demand and supply in a two-stage mode! The demand model estimated catch
as a function of the pnce of stnped bass, pnce indeces for poultry and fish, and per capita
income Supply vanables included an abundance index for both the Chesapeake Bay and
Hudson River stocks and pnces of alternative species yellowtail flounder, flukes,
porgies, bluefish, and lobsters plus a non-fishery wage rate The latter proved non-
significant and was dropped from the final specification Companng estimates of
consumers' and producers' surplus at abundance levels in 1975 versus 1985, the authors
found that the decline in stocks amounted to a loss of $396,000 ($1985) in economic
surplus
Atlantic Stnped Bass Study
Strand et al (2002) estimated the benefits to stnped bass commercial fishenes of
management regulations to preserve and rebuild the stocks As part of this study they
estimated a demand curve for stnped bass The authors estimated an inverse demand
model with ex-vessel pnce as a function of per capita landings and per capita food
expenditures Integrating the demand curve yielded estimates of consumer surplus In
the absence of data to estimate a supply function, the authors relied on information from
other studies that indicated the producers' surplus amounted to about $0 54 per pound
72
-------
They 1hen used these functions to estimate Ihe net benefits in each year attributable to
stock management policies in the fisheiy
Red Drum Study
Thurman and Easley, Jr (1992) demonstrated an approach to valuing changes in
commercial fishery harvests using a general equilibrium framework They applied their
approach to estimating demand and supply functions for red drum harvested in the Gulf
of Mexico The general equilibrium approach treated red drum as one input into
consumer demand for all seafood Thus, their model allows catch or other harvesting
restrictions on red drum to influence the welfare of fishermen supplying substitute
species and the welfare of otier firms in the supply chain The empirical specification
treated the pnce of red drum as a function of per capita quantity of red drum consumed,
price of two substitutes, red snapper and catfish, and per capita disposable income Both
quantity and pnce of red smipper were considered endogenous and estimated using factor
prices for fuel and labor as instrumental variables in a three-stage estimation The model
was estimated on data from 1970 through 1985 The authors compared the estimates of
consumers' surplus obtained from the general equilibrium model and compared it to a
partial equilibrium model and found that economic losses attributable to a 10 percent
reduction in catch were 43 percent larger for the general equilibnum model
Atlantic Swordfish Study
Lee et al (2000) employed a bioeconomic model to analyze the effect of management
alternatives on the Atlantic swordfish fishery The swordfish fishery is managed under
international convention with quotas set to protect stocks The U S share of the quota is
29 percent, but quotas have been reduced from 17,000 metric tons in 1996 to 10,700
metric tons in 1999 in the face of declining stocks This bioeconomic model
simultaneously solved for harvesting effort and stock replacement to yield estimates of
revenue impacts on the fleet under different quota schemes Under current management
quotas the U S fishery yields approximately $47 million ($1997) m producers' surplus
(net returns) over the next five years However, that amount could increase significantly
if a cost-effective way can be found to increase the survival rate of undersized discards
5.3 Estimating Shifts in Demand Due to FCAs
This section uses econometric, tune senes techniques to estimate the changes in demand
that result from the issuance of advisones by FDA in 2001 Of the four species subj ect to
FDA's advisones, king mackerel and swordfish are of most commercial importance, and
the quantitative analysis focuses on these two species Modeling the impact of FCAs
requires estimating a time series econometric model of quantity as a function of species
pnce and other explanatory variables before and after the implementation of the
advisones
53.1 Modeling Demand and Supply
73
-------
Demand for a given species, measured as catch landed, is a function of species pnce,
prices of substitute fish species, pnces of other substitute foods, such as beef or chicken,
household income, population, and other variables Supply of a given species, also
measured as catch landed, depends on species pnce, pnce of substitute fish species,
species abundance, fishing effort, and other variables Since demand and supply jointly
determine pnce, model estimation requires two-stage models using instrumental vanables
(Roy et al, 1991) Typical instrumental vanables for species pnce include factor inputs,
such as pnce of fuel and labor
Following Just and Hueth (1979), consumers' surplus derived from ex-vessel demand
captures the full amount of consumers' surplus effects Thus, there is no need to estimate
demand functions for the wholesale and retail sectors However, estimates of producers'
surplus based on ex vessel pnces understate actual producers' surplus In the absence of
data to estimate supply functions for the wholesale and retail sectors estimates of
producers' surplus base on ex vessel supply are likely to underestimate total surplus
One area of uncertainty is whether there has been sufficient time for consumers to adjust
fully to the advisones issued by FDA in January 2001 Evidence from the study of
pregnant women in Massachusetts (Oken et al, 2003) found that fish consumption was
still declining among pregnant women a year after announcement of advisones by U S
FDA Also uncertain is whether removing an FCA would have an opposite effect of
exactly the same magnitude as the loss associated with issuing one It is possible that
consumers who shift away from a species in favor of some other fish or some other food
may not automatically shift back or may not resume consumption at exactly the same
level as pnor to the change
5.3.2 Data Requirements
The National Manne Fishenes Service (NMFS) collects data on quantity and value of
fish landed These data are available by species and permit calculation of pnces for king
mackerel and swordfish and for other substitute species These NMFS data are available
monthly, but most models employ annual data, since annual cycles of recruitment and
abundance affect the harvest of fish
NMFS also collects data on wholesale pnces and quantities from the Fulton fish market
and other wholesale auction centers Howev er, no data are regularly collected for the
retail sector Consequently, very little economic analysis has been done at the retail level
in the fishing industry
Pnce data on other substitute products, such as chicken or beef, are available m index
form on a monthly basis from the U S Bureau of Labor Statistics Demographic data on
population and income are also readilv available from U S Bureau of the Census
Data on fish abundance are the most difficult to obtain histoncally, since NMFS does not
assess stocks of all species each year However, NMFS' Sustainable Fishenes Division
recently assessed stocks of both king mackerel and Spanish mackerel and estimates of
74
-------
abundance for both are avail able back to 1981 (NMFS, 2003) Regulation of swordfish
stocks is the responsibility ol the International Commissions for the Conservation of
Atlantic Tunas (1CCAT), which assigns quotas to member countries The NMFS Science
Center in Miami is responsible for assessments of the South Atlantic swordfish
population However the 2003 stock assessment (SCRS, 2003) found that
inconsistencies in the available catch per unit effort trends precluded obtaining reliable
assessments of the South Atlantic swordfish stock
Defining a variable to capture the impact of FCAs requires consideration of both the date
of initial public announcements and the level of publicity and advertising accompanying
the announcement Since FDA and the states imposed FCAs on different species at
different times, it is difficult to define one variable that captures the impact of the
advisones FDA's announcement of an FCA for four species in January 2001 received
wide publicity, but it is not known how much publicity attached to various state
announcements in 2000 Since the state-issued FCAs for king mackerel in March 2000
did not include an entire year and may not have received the same level of publicity as
the FDA announcement, the analysis applies a value of 0 5 to 2000 and 1 thereafter For
swordfish the impact of FCAs across the entire study area extends back to the FDA
advisory in 1995, and the analysis employs a dummy variable equal to 1 starting in lhat
year
Given the reliance on annual data, there are a limited number of observations available to
the model Data are a\ ailable for king mackerel and swordfish back from 1975 through
2002 This period provides ?8 observations for use in estimating the models
5 33 Demand Estimates
The demand models for both king mackerel and swordfish regress per capita quantity
landed on real (deflated) species price, a deflated pnce index for meat, real (deflated) per
capita income for the Southeast, and a dummy variable that captures the penod of the fish
consumption advisones Pnce and mcome variables are deflated using the Consumer
Pnce Index (Urban-all items) To avoid the endogeneity problem of species pnce as an
explanatory variable the regression uses two-stage least squares with deflated pnce of
No 2 diesel fuel and a deflated wage index as instruments These cost variables use the
Producer Pnce Index as a deflator
Tables 5-3 and 5-4 summanije ihe results of linear, semi-log, and log-linear specifications
for both species The king mackerel model estimates demand over the penod, 1975 to
2002 The swordfish model estmates demand over the penod, 1980-2002, by which time
swordfish demand had recovered from FDA's regulatory response to high mercury levels
in imported swordfish in 197]
75
-------
Table 5-3
King Mackerel Regression Results:
Dependent Variable: Quantity Landed per Capita
Variable
Adj r- square
Intercept
Species pnce ($/lb )
CPI meat
Per capita income
D FCA
King mackerel
Linear
07252
491 879**
-25704*
25835
-1 6542**
36243
Semi-log
08191
6 669**
-74 966
0801
-0021**
0100
Log-linear
08318
14696**
-0554
0678
-2 727**
0050
as instruments to predict species pnce
2) All prices and per capita income are deflated
3) ** is significant at p<0 05, * is significant at p<010
4) D_FCA - 0 5 in 2000, =1 in 2001 and 2002
Source Estimated using two-stage least squares
The king mackerel demand specifications generated correct signs for key pnce vanables,
but the coefficients were mostly not significant The coefficient on species pnce was
negative, as expected, but only the linear specification was weakly significant at p<01
The coefficient on the meat pnce index vanable is positive, since consumers will shift to
fish if meat pnces nse, but is not significant in any of the specifications The coefficients
on per capita income were all significant but negative, which suggests that king mackerel
is an inferior good (i e people substitute away from king mackerel as Iheir income
increases) Finally, the FCA dummy entered as positive and insignificant, which means
that the models provide no evidence that the fish consumption advisory on king mackerel
resulted in any negative impact on consumer demand Testing different tme penods for
the dummy did not yield any indication of a negative impact
76
-------
Table 5-4
Swordfish Regression Results
Dependent Variable: Quantity Landed per Capita
Variable
Adj r-square
Intercept
Species pnce ($/lb )
CPI meat
Per capita income ($)
D FCA
Swordfish
Linear
08326
39477
-765 321
134116**
-0818**
-1 284
Senu -log
08006
3743
-8799
2436
-0017**
-0247
Log-linear
08016
13 839**
-0099
2161
-2 139**
-0251
Note 1) Estimator uses two-stage least squares with pnce of diesel fuel and wage rates
as instruments to predict species pnce
2) All prices and pei capita income are deflated
3) ** is significant at p<0 05, * is significant at p<0 10
4) D_FCA = 0 5 in 2000, =1 in 2001 and 2002
Source Estimated using two stage least squares
The models of swordfish demand generated coefficients with mostly correct signs, but
the level of significance of most of the coefficients was below p<0 1 Species pnce
entered negatively in all three specifications, but was not significant at p<0 05 in any of
the specifications The CPIjneat variable was positive in all specifications, but only
significant in the linear model Per capita income was negative and significant in all
three specifications, which was unexpected since swordfish commands high prices in
restaurants Since income is generally correlated with education and possibly with
awareness of health nsks and since the risks of eating swordfish have been known since
the early 1970s, it is possible that the per capita income vanable is picking up some of the
nsk effect of consuming swoidfish. It would require individual level data to test this
hypothesis Meanwhile, the dummy for the swordfish FCA was negative in all three
specifications, but not significant in any The negative coefficients on the FCA dummy
in these models provide an indication that the advisones may have had a negative impact
on swordfish demand, but the uncertainty of the estimates is too large to conclude that the
magnitude of the effect is significantly different from zero
It is important for readers to understand that these econometnc investigations were
narrowly focused on the case study area The model specification assumed that the South
Atlantic area was a closed system where quantities landed were also consumed In fact,
landings from other areas, especially the Gulf for king mackerel and imported swordfish,
are likely to affect this region and may affect the results Howe\ er, limited resources
precluded a wider investigation of these effects
5.4 Estimates of Changes in Economic Surplus
77
-------
The modeling analysis proceeded on the assumption that since the U S FDA's advisories
applied to only four species, the most likely evidence of an effect was likely to be found
in analyzing demand for those species with the most commercial significance However,
the modeling analysis did not find significant evidence of any effect on either king
mackerel or swordfish. The coefficient on the FCA dummy in the king mackerel models
was positive and insignificant In the swordfish models the FCA dummy was negative,
but insignificant Given these modeling results, there is insufficient evidence of any
negative effect of FCAs on consumer demand In the absence of an effect of FCAs on
consumer demand for king mackerel and swordfish, there is no basis for calculating
changes in economic surplus for those species
The only evidence of an effect of FCAs on consumer demand comes from the
Massachusetts survey offish consumption by pregnant women (Oken et al, 2003) This
survey indicated that pregnant women had reduced their consumption of all types of
seafood not just the four species named by FDA in its advisory On the one hand, if the
advisory had a general effect on all seafood consumption, the impact of the FCA could be
quite large On the other hand, if the effect of FCAs is limited to pregnant women, the
effect on demand for marine fish and shellfish may be quite small Evidence will soon be
available from the NHANES for 2001 and 2002 that may resolve this uncertainty Until
then, the gaps in information and data are too great to try to estimate the benefits to
consumers of reducing mercury concentrations in manne fish and shellfish.
5.5 Gaps in Data and Information and Uncertainties
At the current time there are no econometric studies of demand for any manne species
that demonstrate an impact of FCAs on consumers This is a maj or gap in knowledge,
and in the absence of such evidence, it is not possible to estimate benefits of reducing
mercury concentrations in manne fish
There are two approaches to bndge this gap One approach is to estimate econometric
models for all major species to test the hypothesis that FCAs may affect all species, not
just the species named in the FCAs A second approach is to rely on NHANES data to
determine the change in fish consumption nationwide by comparing consumption before
and after FDA's announcement of advisones The NHANES data can provide
information on the change in consumption by species for specific gender and age groups
Any changes could then be applied to shift the demand curves for key species and,
following Jakus et al (2002), to estimate the resulting change in economic surplus It is
important to remember that demand models estimated at the ex vessel level capture all
the consumers' surplus, but supplv models do not Including the impact of wholesale and
retail supply would expand the impact on overall producers' surplus, but the relative
amount of any increase in total surplus is unknown
It is also important to remember that in a general equilibrium framework some of the gam
experienced by consumers would be balanced against losses by meat or other agricultural
producers Thus, any changes in surplus due to shifts in demand for fish are likely to
represent an overstatement of changes in overall social welfare
78
-------
6.0 Assess the Benefits to Recreational Saltwater Anglers
Saltwater fishing is a major recreational pursuit and provides a large number of high
value recreational use days The recreational value of a saltwater fishing day depends on
a number of vanables including water body, species targeted, success rate, mode (shore
or boat), gear, and other vanables And for a significant percentage of saltwater anglers
consuming the fish they catch has significant benefit As noted in Section 5 0, high
mercury levels have caused U S Food and Drug Administration (FDA) and state health
agencies to issue FCAs for a number of species highly prized by recreational anglers,
especially king mackerel, swordfish, tuna, striped bass, bluefish, and sharks
Quantifying how these mercury FCAs and the nsks they represent affect the utility of a
recreational fishing tnp is 1he subject of this section.
Where anglers are unaware of or fail to adjust their behavior to conform to a mercury
advisory, consumption offish may result in health nsks, and Section 4 0 quantifie the
benefits of reducing health nsks to all consumers, including anglers Where anglers do
make adjustments in their fishing or consumption behavior, there is likely to be a change
in overall satisfaction and economic utility Quantifying the changes in utility associated
with fish consumption advisones is Ihe focus of this section.
Valuing the welfare effects of reductions in mercury deposition on recreational requires a
benefits transfer approach, since Ihere are no studies that quantify benefits of changes in
FCAs in a marine environment This entails finding and evaluating studies that measure
changes in attributes of recreational fishing that are comparable to that of mercury FCAs
This section reviews the literature to identify comparable studies and applies the results
of these studies to estimate changes in the value of marine recreational fishing trips
6.1 Angler Response to Fish Consumption Advisories
How recreational anglers react and adjust to fish consumption advisones is key to
understanding the impact of mercury on recreational fishing An anglers' response to
FCAs depends on two factois
awareness of and response to the advisones, and
utility denved from consuming the catch
Evidence from a number of studies suggests that the majonty of anglers consume their
catch so that any changes in consumption or angling behavior to reduce or avoid health
nsks are likely to have welfare consequences The literature reviewed below indicates
that awareness of FCAs exhibits a wide range from site to site, but the most recent of
these surveys date from 1995 It is likely that heightened communications efforts by
federal and state agencies in the last five years have greatly increased awareness of
mercury nsks of consuming fish, but there are no studies that document this supposition
6.1.1 Angler Awareness of Fish Consumption Advisories (FCAs)
79
-------
A review of the literature found that awareness of FCAs varied from 19 percent to 96
percent (Jakus et al, 2002) Most of these studies applied to anglers fishing on
freshwater lakes and streams, but six included marine fishing locations
May and Burger (1996) surveyed anglers in 1994 along the NJ coast and found that
awareness of FCAs varied from 28 percent on Raritan Bay to 60 percent on Arthur Kill
Phlugh et al (1999) surveyed anglers on Newark Bay in 1995 and found that 60 percent
were aware of advisones In a 1990 survey Burger et al, (1993) found that only 19
percent of anglers fishing in Jamaica Bay were aware of FCAs And in a 1983-85 survey
of anglers fishing in New York Harbor Belton et al (1986) found an awareness rate of 50
percent The average of these six studies, all of which are in the NY-NJ area, is 41
percent
Five of these six studies in estuanne locations also provided information on the percent of
anglers who consumed some or all of their catch May and Burger (1996) reported
consumption rates of 70 percent, 88 percent, and 82 percent for Arthur Kill, Rantan Bay,
and NJ Shore sites, respectively Burger et al (1993) reported a consumption rate of 85
percent for Jamaica Bay, NY, and Belton et al (1986) reported 58 percent consumed
their catch from New York Harbor These results suggest that a high percentage of
saltwater anglers consume their catch, even from areas with well-publicized pollution
problems However, these studies are all concentrated in the Northeast, and it is possible
that awareness of FCAs and consumption behavior differ for anglers in the South Atlantic
area
Several of the studies of freshwater anglers provided information on consumption
behavior in response to FCAs Connelly et al (1992) found that 47 percent of New York
State anglers would eat more fish if a fish consumption advisory were not in place,
Similarly, Connelly and Knulh (1995) found that 54 percent of Lake Champlam anglers
would eat more fish if a fish consumption advisory were not in place, whereas in Maine
only 15 percent would consume more fish (MacDonald and Boyle, 1997)
In response to an announcement of fish consumption advisones 62 percent of Lake
Ontario anglers made some changes in consumption, and 77 percent of those said they
reduced consumption (Connolly et al, 1993) Similarly, Knuth et al (1993) found that
42 percent of Ohio River anglers reduced consumption in response to an FCA while 13
percent stopped eating all fish, and 26 percent changed target species Breffle et al
(1999) reported that 23 percent of Green Bay fishermen targeted different species, 45
percent changed the species kept for consumption, and 47 percent changed the size of the
species kept for consumption In Michigan Kneger and Hoehn (1998) found that in
response to a hypothetical question, 'if your favonte site had an FCA', 25 percent would
not eat any fish and 14 percent would eat fewer fish
Se\eral of the studies asked questions about changes in mp behavior in response to fish
advisones MacDonald and Boyle (1997) found that in Maine if there were no advisory
in place 10 percent would fish more days, 5 percent would fish more waters, and 5
percent v\ ould fish different waters In response to the present advisory Connolly and
80
-------
Knuth (1995) found that 7 percent of Lake Champlain anglers stopped fishing in the
contaminated area, and Breffle et al (1999) found that a similar percent of anglers
stopped fishing in the contaminated area of Green Ba> Connelly et al (1993) found that
37 percent of Ohio River anglers would take fewer tnps and 26 percent would change
fishing sites Breffle et al (1999) found similar percentages for Green Bay fishermen
Connelly et al, (1992) also reported that about 25 percent of New York State anglers
made changes in their recreational fishing tnps, including changing fishing location,
changing species eaten, or changing size of fish eaten Kneger and Hoehn (1998) found
that 36 percent of Michigan fishermen would change fishing sites if their favorite site had
anFCA
6.1.2 Angler Adaptations to Mercury FCAs
A wide variety of adaptation;, and adjustments in fishing behavior and consuming
behavior are possible, and most are likely to affect the overall benefit, or utility, of a
recreational fishing tnp Since mercury affects species over wide areas, possible
adaptations in behavior focus on changes in consumption, changes in species targeted,
and changes in number of fishing tnps The following are some of the possible
behavioral adjustments in response to an FCA for mercury
» No change m fishing or consumptian behavior,
No change in fishing behavior, but reduce or cease consumption following
recommendations for listed species,
No change in fishing behavior, but reduce or cease consumption of all species,
No change in species, targeted, but reduce the number of days fishing with
resulting reductions in consumption,
No change in fishing effort or consumption, but target different species,
Reductions in mercury levels in fish or removal of an FCA may be expected to result in
the same range of adjustments, but in the direction of more frequent targeting of affected
species However, it is not known if an anglers' response to eliminating an FCA will be
perfectly symmetrical to the lesponse when the FCA was first issued
All these adaptations are likely to result in a change in the level of satisfaction from a
fishing tnp Adjustments to fishing behavior can be valued using random utility models
(RUMs) that relate the change in angler satisfaction, or utility, to changes in species
availability, success rates, area closures, and FCAs RUMs use trip-specific information
from anglers' dianes or survey recall to estimate the value of a trip as a proxy for utility
and then apply information on species targeted, success rates, gear used, and other
information, including whether the fish is safe to consume, to estimate recreational
demand
Where anglers change consumption but not fishing behavior so that tnp time and costs
remain unchanged, RUM models will not capture the loss in utility Thus, RUM models
may undervalue welfare losses where the only behavioral change is a change in utility
associated with consumption To capture the full value of the loss in utility from altering
81
-------
consumption behavior may require a stated preference approach, especially WTP
surveys
6.2 Manne Recreational Tnps
The National Manne Fisheries Service (NMFS) collects and maintains a database of
information on recreational fishing trips Recent data for the South Atlantic provide a
snapshot of recreational fishing at current levels of mercury deposition This snapshot
represents a baseline estimate of marine recreational fishing in the study area
6.2.1 Marine Recreational Fish Statistical Survey (MRFSS)
NMFS has collected data on saltwater recreational fishing trips since 1979 The Marine
Recreational Fishing Statistical Survey (MRFSS) data include results of both creel
interviews and telephone interviews (Gray et al 1994) The sampling frame is a coastal
county, and data collection takes place in two-month waves
The creel survey provides estimates of the number and species offish caught per tnp, and
the telephone survey estimates the number of tnps This information allows NMFS to
estimate total recreational catch, by species Data collected from the creel interviews
reflect the intercepted ttip and include information on fishing area and mode, hours
fished, gear used, species targeted, catch by species, angler residence, and information on
previous tnps The telephone survey provides data on tnps in the last two months,
including date of tnp, fishing mode, gear used, area fished, boat access site
characteristics, and state and county where the tnp occurred
6.2.2 Recreational Fishing Tnps in the South Atlantic Area
MRFSS data for 2002 provide a baseline for the number of manne recreational fishing
tnps These data show that recreational anglers made a total of nearly 18 million
recreational fishing tnps in the South Atlantic area in 2002 The number of tnps in the
region has increased over the last 10 years from 16 7 million in 1992 to 17 8 million m
2002 However, the annual totals have fluctuated significantly over that penod and tend
to vary inversely with general trends in economic activity More unemployment seems to
mean more time for fishing For example, tnps peaked at over 21 million in 2001 at the
depths of the recent recession and were at a minimum of about 14 4 million in 1999 at the
peak of the recent economic boom The average of annual recreational fishing tnps over
flie last five years is 181 millioa
In 2002 Flonda accounted for the largest number of recreational fishing tnps, about 103
million North Carolina was second with 5 6 million tnps, followed by South Carolina
with 1 3 million tnps and Georgia with 0 6 million tnps
Anglers' saltwater fishing tnps compnse three pnmary modes shore, boat, and charter
About 51 percent of the total tnps in 2002 were from shore, beach, bank or man-made
structure (pier), as shown in Figure 6-1 About 47 percent of the tnps utilized private or
82
-------
rental boats The remaining 2 percent of the trips were anglers who hired charter fishing
boats
Figure 6-1
Marine Recreational Trips by Mode in 2002
Boat
46 5%
MRFSS data show that anglers targeted a wide variety of species Table 6-1 summarizes
recreational trips by species targeted Recreational anglers targeted 2 48 million taps to
catch drums, which amounted to almost 14 percent of total trips Anglers also targeted
nearly 1 million trips, about 5 5 percent, to catch mackerels and tunas Other species
targeted by recreational anglers included dolphins, flounders, snappers, bluefish, porgies,
temperate basses (primarily striped bass), and jacks However, these species account for
less than 40 percent of all tips, and other species targeted account for almost 12 million
marine recreational fishing trips
83
-------
Table 6-1
South Atlantic Marine Recreational Trips by Species Targeted in 2002
Species
Drums
Mackerels and Tunas
Dolphins
Flounder
Snappers
Bluefish
Porgies
Temperate Basses
Jacks
Olher Fishes
Total
Tnps (millions)
248
099
071
067
029
026
024
021
021
1168
1776
Percent
1396
557
400
377
1 63
146
1 35
1 18
1 18
6570
10000
Note Temperate basses are mostly stnped bass
Source National Marine Fisheries Service, available online at
ftp //cusk nmfs noai«ov/mrfbs/pubbcaiKWtbl4G20027 np
6.3 Value of a Saltwater Recreational Fishing Tnp
There are a large number of studies that estimate the economic value of a recreational
fishing trip, and several authors provide useful reviews of the literature Freeman (1995)
cited 27 studies in his review of marine recreational fishing And an earlier review by
Walsh, Johnson, and McKean (1992) assessed a large number of recreational fishing
studies The focus of much of this recreational fishing literature has been on valuing
changes in pollution or other problems lhat affect either access to sites or stock
abundance and catch rates
To value Ihe effects of FCAs using a benefit transfer methodology requires estimates of
the value of trips for marine recreational species affected by FCAs Since mercury FCAs
pertain to an entire species, the theoretically correct value would reflect the welfare effect
of access to that species over the entire study area A number of marine recreational
fishing studies provide estimates of access to species or species groups at the county or
state level However, restncting access to king mackerel in the South Atlantic results in a
greater loss in welfare than the sum of Ihe losses for each state individually This is
because restricting access to one state still allows anglers to access king mackerel if they
are willing to travel farther to other states Similarly, because of the loss in substitution
possibilities eliminating access to the four species that are subject to FDA's advisones
would result in greater welfare losses than the sum of the loss associated with each
species by itself
84
-------
The section below summarizes several useful studies Some of these studies estimate the
value of access to one species or group of species, and some estimate the value to all
species in a given area
6.3.1 Studies in the South \tlantic Area
Haab et al (2000) estimated he economic value of marine fishing in the Southeast and
included species-specific estimates The authors estimated separate models for red drum,
spotted seatrout, snapper/grouper and migratory pelagic (e g mackerels and tunas) The
analysis applied a nested random utility model (RUM) methodology to MRFSS data
collected in 1997, including add-on economic data on cost of travel Sufficient data were
available only for boat anglers so the tnp values apply only to boat-trips The authors
employed both a non-nested logit specification and a nested specification The non-
nested RUM estimates site choice conditional on choice of species and mode The nested
RUM allows switching between species and estimates the value of access to targeted
species based on catch and keep rates and other variables The authors tested four model
specifications with different choice sets using both historic and expected catch rates, but
found litfle difference n WTP estimates across the four models They did find
differences between the simple RUM and nested RUM As expected, the nested RUM
produces lower estimates for the loss of access to any given species, since it allows for
wider substitution across species and locations Consequently, the nested RUM welfare
estimates are most applicable to policies, such as FC As, that restrict access to a given
species, smce they reflect the per tnp loss/gain in angler utility from closing/opening
access to a given species or species group in a large area. The models indicated the
largest per tnp values applied to coastal/migratory pelagic species, followed by
snapper/grouper, red drum, and spotted seatrout Table 6-2 summarizes compensating
variation welfare estimates in 1997 dollars per boat-tnp for the targeted species in each
state
85
-------
Table 6-2
Compensating Variation Welfare Estimates for Site Access
($1997/boat-trip)
State
Florida (Atlantic)
Georgia
South Carolina
North Carolina
Total
Red Drum
$1 28
043
078
051
300
Spotted
Seatrout
$132
Oil
061
043
247
Coastal
Migratory/Pelagic
$216
Oil
064
098
389
Snapper-
Grouper
$160
Oil
068
042
281
All Species
$2088
122
901
1287
4398
Note: 1) Economic welfare estimates are based on the nested RUM model, which considers the
full 68 zone choice set and the mean historic catch and keep
2) Totals for individual species/species groups in the four states in the study area
underestimate true welfare estimates, since shutting down access to a species over a large
area limits substitution possibilities substantially
3) A two stage nested model of mode/species targeted/site access generates comparable
values for access by state, but the authors did not report values for access by
species/species groups
Source: Haab et al , 2000, Tables 5-6 and 5-7
In a study of sportfishing on Flonda's Atlantic coast (Bockstael et al, 1989) estimated
the value of recreational fishing tnps using a nested RUM model The authors modeled
site choice conditional on mode/species choice The authors grouped species into big
gamefish (e g billfish, marlin, tuna), small gamefish (bluefish, mackerel, and seatrout),
and bottom fish (sheepshead, snapper, grouper) and modeled six mode-species choices
shore-big gamefish, shore-small gamefish, shore-bottomfish, boat-big gamefish, boat-
small gamefish, and boat-bottomfish Unfortunately, the authors did not report welfare
estimates for each of the three species groups so that the reported per trip values represent
an average value over all species The authors estimated that the welfare benefits of
access by both shore and boat to nine counties along Flonda's Atlantic coast ranged from
about $1 to $8 per trip ($1988) with the mean of the estimates for the 9 counties equal to
$3 32 per tnp ($ 1988) The value for boat access ranged from about 80 to 90 percent of
the total tnp value with shore access accounting for the remainder
In a study of sportfishing in the mid- and south-Atlantic states McConnell and Strand
(1994) employed data from the 1987 MRFSS to estimate the value of access by state over
all species Following the RUM approach used by Bockstael et al (1989), the authors
modeled site choice conditional on mode/species choice They aggregated sites into
counties and species into big gamefish, small gamefish, bottomfish, flat fish, and non-
targeted small gamefish They tested both expected catch rates and historic catch rates
for these aggregated species in 1he conditional site choice model The authors estimated
mean annual per tnp willingness to pay per one-day fishing tnp for access to all species
in Florida, Georgia, South Carolina and North Carolina Based on expected catch rates,
86
-------
Ihe authors reported values of $80 per top for Florida, $42 per tnp for Georgia, $68 per
tnp for South Carolina, and $66 per tnp for North Carolina The values obtained by
McConnell and Strand (1994) from 1987 data are considerably higher than those obtained
by Bockstael et al (1989) using the same RUM approach and the same 1987 MRFSS
data. In part, this is because the former evaluated site access for the whole states whereas
the latter valued site access at the county level
6.3.2 Other Relevant Studies
Greene et al (1997) estimated the demand for recreational fish in Tampa Bay using a
nested random utility model The approach first considered the decision whether or not
to fish and then whether or not to fish in Tampa Bay The authors estimated a repeated,
nested logit model based on MRFSS data supplemented with data from a University of
Florida survey for 765 anglers The authors reported that the value of a recreational
fishing Inp on Tampa Bay amounted to $1 68 ($1992), this represents the value of access
to the Bay to fish for a mix of all species
In a related study of marine sportfishing in New England and mid-Atlant c states Hicks et
al, (1999) applied an identical methodology to that of McConnell and Strand (1994) to
estimate the value of fishing tnps by state However, the study used 1994 MRFSS data
for the econometnc estimation The 1994 data yielded smaller per tnp values for marine
recreational fishing than McConnell and Strand obtained using the 1987 data Hicks et
al estimated mean values per tnp that ranged from <$1 ($1994) for New Hampshire to
$42 33 ($1994) for Virginia Comparatively, Hicks et al (1994) estimated the value of
Maryland sportfishing tnps al about $12 ($1994), whereas McConnell and Strand (1994)
estimated a significantly high er value of about $27 ($ 1987) These \ alues both represent
WTP for one-day tnps averaged over all species
633 Summary of the Value of Recreational Fishing Tnps
Estimates of recreational fishing tnp values in the South Atlantic are fairly consistent
with the exception of the study by McConnell and Strand (1994) The McConnell and
Strand estimates denve from 1987 data, and this analysis elects to rely on studies that
used more recent data The study by Haab et al (2000) used 1997 MRFSS data and
employed a theoretically appropnate RUM approach to estimate values of access to
species
FDA's advisones apply to sharks, swordfish, king mackerel, and tlefish, and an advisory
on tuna is pending All but tlefish may be categorized as coastal/migratory pelagic The
coastal/migratory pelagic species group analyzed by Haab et al include little tunny, king,
spamsh, and cero mackerels, dolphin, and cobia This grouping is a good proxy for the
species subject to FDA's mercury advisones Haab et al (2000) estimated a value for
this group that totaled of $3 87 ($1997) per tnp in the four states that comprise Ihe South
Atlantic area The study also reported that the per tnp value of access to this entire
species group amounted to $') 21 ($1997) This is 5 percent greater than the sum of per
tnp access values for each of the eight states analyzed (including states bordering the
87
-------
Gulf of Mexico) Using this factor to inflate the South Atlantic per tnp value yields a per
tnp value of$4 06 ($1997) for access to the coastal/migratory pelagic group
Haab et al 's values apply only to private/rental boat trips Shore taps would be less and
charter taps would be more Shore taps account for less than 5 percent of the tuna and
mackerel landings, by weight, and most of that catch is probably incidental rather than
targeted taps Bockstael et al estimated that the ratio of the value of a boat tap to a
shore tap amounted to about 6 3 to 1 Thus, shore taps are valued at about 16 percent of
the value of boat taps None of the studies provided values for charter boat taps, and the
analysis assumes that charter boat tap values for coastal/migratory pelagic species are
double that of private/rental boat taps
Table 6-2 summarizes the baseline value of tnps targeted at coastal/migratory pelagic
species subject to FCAs Total value amounts to about $7 million for shore taps, about
$39 million for boat taps, and about $4 million for charter tnps The overall baseline
value of taps across all modes amounts to about $43 million
Table 6-3
Baseline Value of Tnps Targeted at Species Subject to FDA's Mercury Advisories
Mode
Shore
Pnvate/rental boat
Charter Boat
Total
Annual Tnps
(millions)
923
842
045
1808
Value per Tnp
($2003)
$074
$464
$927
Total Value
(millions $2003)
$685
$3907
$417
$4324
Note Value per tap is inflated to 2003 dollars using the Consumer Price Index
1997=1605,2003=1833
Source: Calculated, based on estimates provided in the text
6.4. Impacts of Mercury FCAs on Recreational Saltwater Fishing
As discussed above, saltwater anglers may react to FCAs in many ways ranging from no
change to a mix of changes in fishing and consumption behavior To complicate matters
further FCAs may range from 'eat no fish' to 'X meals per month' and differ between
sensitive populations and the general population Given the likelihood lhat changes in
emissions may result in relatively small changes in fish concentrations for a given
species, it is quite conceivable that FCAs might change from 'eat one meal per month' to
'eat two meals per month' Thus, it is likely to be necessary to quantify the effects of
small changes in FCAs
There are three possible approaches to measuring the change in angler welfare
attributable to FCAs The first is to measure welfare changes from actual tnp data before
and after the imposition of an FCA This requires data on recreational saltwater fishing
tnps using a random utility modeling (RUM) approach. Although recreational angler tap
88
-------
data have been used to estimate RUMs for manne recreational fishing, most of these
studies have focused on the yalue of access to a site or changes in catch rates as a
reflection of stock abundance A search of the literature has not revealed any RUM
studies that estimate the impact of FCAs on benefits of recreational fishing in a manne
environment
Hie second approach is to iu,e stated preference data from a CV survey that quantifies the
WTP to avoid an FCA or to reduce its stringency Coupled with data on fishing
attributes, such as catch rates, the CV data can provide insight into how changes in FCAs
affect overall utility of a recreational fishing tnp Again, however, a search of the
literature found no CV studies that measure the impact of FCAs in a manne environment
In the absence of studies for die first two approaches a third approach relies on
transferring benefits from freshwater recreational fishing studies where changes in angler
behavior are comparable to that of mercury fish consumption advisories This requires an
estimate of the value of a tnp for a targeted species Then it requires some basis for
estimating Ihe impact of an FCA on the value of that tnp and on the number of tnps,
since FCAs are likely to affect (he decision to fish or not This latter estimate, of course,
is affected both by the fraction of anglers who are aware of FCAs and whether they fish
pnmanly for sport, consumption, or both Needless to say, the reliability of this approach
depends on being able to find studies that reflect values and beha\ior in response to
FCAs that are applicable to recreationally-targeted, manne species
6.4.1 Impact of Fish Consumption Advisories
Many recreational fishing studies have attempted to quantify the welfare benefits of
reducing pollution, but few oi these studies are useful in quantifying the effects of FCAs
This is because most of these studies have focused on the effects of point-source
pollution and the resulting impacts on recreational fishing through area closures or fish
mortality and reduced catch rates Where the pollution onginates from non-point sources
and affects large areas but does not seem to affect fish mortality, as is the case with
mercury, the pnmary impact on angler utility is not on catch rates but rather Ihrough
health nsks from consuming the catch
The literature includes a few studies that have attempted to quantify the impact of FCAs
on recreational fishing However, Ihey all evaluate the effect of FCAs on closures of
specific areas Several studies assess the impact of FCAs for particular species in
freshwater lakes Another evaluates the impact of a statewide advisory for certain
freshwater species hi the only study that values the impact of FCAs in a manne
environment the advisory appihes to an area rather than a species Thus, none of these
studies directly address the issue of how FCAs on a marine species over a wide area
affect the choices and utility of recreational anglers
Green Bay Study
89
-------
As a result of PCS pollution from paper nulls,, the states of Wisconsin and Michigan
implemented fish consumption advisones in 1976 and 1977, respectively, for the Lower
Fox River and an area of Green Bay, which forms an arm of Lake Michigan These
advisones apply to yellow perch, walleye, trout and salmon, and smallmouth bass They
range from eating 'one meal per week' to 'do not eat' Breffle et al (1999) combined
catch rate data with CV choice data to estimate the damages associated with the
contamination and the resulting FCAs
The researchers developed a contingent valuation survey instrument that asked fishermen
to choose between paired alternatives that differed in terms of boat launch fees, tme to
catch various species (inverse of catch rate) and the level of FCAs FCA levels ranged
from 'unlimited' consumption for all species (Level 1) to 'one meal per month' for
yellow perch and 'eat none' for the other species (Level 9) Level 4, which consisted of
'one meal per week' for yellow perch and 'one meal per month' for the other species,
approximated the current conditions An additional question, which asked anglers how
many days they would expect to fish under the chosen alternative, provided information
for assessing the impact of FCAs on the number of tnps
The authors employed a telephone survey to gather information on the number of Green
Bay fishing tnps and to recruit participants for the paired comparison choice survey
They mailed the choice survey to 820 anglers and received 647 responses in 1998 Each
respondent chose between eight paired compansons and also provided an estimate of the
number of days they would fish Green Bay under the chosen alternative
Using a probit model, the authors estimated the contnbution to utility for both catching
fish and consuming fish (avoiding FCAs) and the marginal utility of income from trade-
offs with different levels of access fees17 Under current conditions for catching yellow
perch, trout-salmon, walleye, and smallmoulh bass the utility of catching fish evaluated at
the marginal utility of income amounted to $25 57 per fishing day ($1998) In
comparison the value of unlimited consumption relative to 'do not eat' for trout-salmon,
walleye, and smallmouth bass and '1 meal per month' for yellow perch amounted to
$21 71 per fishing day ($1998) Thus, the value of consuming fish accounted for roughly
46 percent of the total value (=$21 71/(25 57+21 71))
The detailed nature of the questionnaire allowed the researchers to explore WTP to move
to different levels of FCAs They found that Green Bay anglers were willing to pay
about $9 75 per angler-day ($1998) to improve from current FCA conditions (level 4) to
unlimited consumption of all species (level 1) WTP to improve from a more restnctive
level 9 to unlimited consumption (level 1) amounted to $21 71 per angler-day ($1998)
In addition, analysis of the allocation of fishing days between Green Bay and other sites
revealed that the elimination of FCAs from Green Bay would result in a 14 5 percent
increase in the number of Green Bay fishing days Not measured was the effect of a
17 The author specified utility as Uld = lp,ci + ZpqFCAq+ (3y(y, - TC, - fee) + eld, where Uld is utility lor
individual i on day d, q is hours to catch a fish of species 1, FCA q is 1 of 9 levels offish consumption
advisories, y, is income for individual i after tip costs and launch fees, 6,d is an error term, and the PS are
estimated parameters
90
-------
reduction in FCA levels on the number of taps to Green Bay b> non- anglers and anglers
not currently fishing Green
This study provides the best approximation of FCA values for transfer to a saltwater
fishing environment Green Bay is part of Lake Michigan and some of the characteristics
of big lake fishing approximate saltwater recreational fishing In addition, this study
provides useful data on the relative values of improving FCAs from one level to another
Table 6-4 summarizes die WIT values of moving from vanous FCA levels to 'unlimited'
consumption (level 1) Focusing primarily on walleye and salmon- trout as target species
for recreational anglers, WTP to move from '1 meal per week' to 'unlimited'
consumption (level 2 to 1) was about $1 81 ($1998) WTP to improve from ' 1 meal per
month' to ' 1 meal per week was about $7 94 (=9 75-1 81) ($1998) To improve from
'do not eat' to ' 1 meal per month' yielded an increment of $1 1 96 (=21 71-9 75) ($ 1 998)
The increment from ' 1 meal per week' to 'unlimited' consumption represents about 8
percent of the value of 'unlimited' consumption. The increment from ' 1 meal per month'
to ' 1 meal per week' comprises about 37 percent of the consumption value, and the
increment from 'do not eat' to '1 meal per month' comprises about 55 percent
Table 6-4
WTP to Reduce the Stringency of FCAs on Green Bay
FCA Change to 'Unlimited
Consumption' from
WTP per fishing day
($1998)
WTP Increment
(S1998)
2YP 'Unlimited'
T-S '1 meal/week'
W ' 1 meal/week'
SB 'Unlimited'
$1 81
$1 81
4 YP '1 meal/week'
T-S '1 meal/month'
W '1 meal/month'
SB '1 meal/month'
$975
$794
9-YP '1 meal/month'
T-S 'Do not eat'
W 'Do not eat'
SB 'Do not eat'
$2171
$11 96
Note 1) YP is yellow perch T-S is trout/salmon, W is waJleye, SB is smalhnoulh bass
2) Level 4 represents the current FCA advisory levels
3) WTP is willingness to pay per fishing day on Green Bay
Source Breffleetal (1999)
Lavaca Bay Study
A chlor-alkali plant in PL Comfort, TX, discharged mercury-containing wastewater
directly into Lavaca Bay from 1966 to 1970 Contaminated mercury materials disposed
of on Dredge Island continued to pollute the Bay after the plant ceased direct discharges
91
-------
in 1970 The Texas Water Quality Board closed the section of Lavaca Bay adjacent to
the facility in April 1988 to fishing and crabbing and the site was placed on CERCLA's
National Pnonty List in 1994 As part of a Natural Resources Damage Claim (NRDC),
the State of Texas, NOAA, and U S Fish & Wildlife Service implemented a study to
assess the damages to recreational fishing from the closure of Lavaca Bay
The Lavaca Bay recreational fishing study (Texas General Land Office et al, 1998)
employed a random utility model to quantify damages attributable to a closure of part of
the Bay due to mercury contamination The study utilized data on actual trips in a
revealed preference model (RP) and data from a hypothetical survey in a stated
preference (SP) model and combined both types of data in a joint model The RP
analysis included data on 1,079 boat trips and 527 shore/pier fishing trips by anglers from
the three adjacent counties during the month of November 1997 The SP model used data
on choice comparisons between alternative fishing sites made by 700 respondents, these
choices resulted in 10,297 usable observations
The RUM model estimates the probability of making a fishing tnp to a particular site,
usually a boat launch or fishing pier, as a function of a site's utility relative to other sites
Site utility depends on site characteristics, especially whether the site is inside the closed
area Normally, RUMs include travel cost as a key explanatory variable so that the
parameter on cost can be used to recover the value of key attributes in dollars In this
study, however, the authors specified the model in terms of travel distance so that the
damage associated with closing the area can not be monetized'8 In the RP, SP, and joint
models for both boat tnps and shore trips the coefficient on the closed area was negative
and significant However, the relative value of the closed area, as expressed in the ratio
of the parameter on the closed area attribute to the distance parameter, varied
significantly between the RP and SP models In the boat models the ratio ranged from
1 3 using RP data to 8 5 using SP data In the shore/pier models the range of variation
was similar, ranging from 0 7 using RP data to 8 5 using SP data
The authors chose to rely on the joint model and calculated the effect of the closure by
comparing the expected utility over all tnps with and without closure They calculated
the expected utilities under both conditions and indexed them to 100 The results from
the models indicated that the mercury contamination and closure of a portion of Lavaca
Bay resulted in a 1 3 percent loss in utility from boating tnps and a 3 2 percent loss in
utility for pier/shore tnps Other reviewers of the study raised important questions about
tiie data analysis and modeling choices, particularly their reliance on the joint model, but
the direction of any bias is unclear (see Parsons, 1998, Smith, 1998, and Adamowicz,
1998)
Transferring Ihe results of this study to the problem of quantifying losses to marine,
recreational fishermen from FCAs is not straightforward This study assesses the impact
of a ban on fishing for all species in an area, whereas mercury FCAs apply to one species
o\ er a wide area Since there are more choices of fishing sites than species to target, this
study probably underestimates the impact of mercury FCAs for most recreational species
' Ubing distance rather than travt.1 cost preUude!> quantifying the utility trade-offs in dollars
92
-------
Studies of the Impact of FCAs on Freshwater Lakes
Several studies of freshwatei lakes address the impact offish consumption advisories
These studies are less comp
-------
The authors used a RUM to capture the effects of FCAs on the choice of site, but their
methodology did not capture the impact of FCAs on tops Specifications included
information on awareness of advisories and catch and release behavior as explanatory
vanables in the model Mean welfare gains from eliminating the contamination and
removing the FCAs amounted to $7 29 per trip ($1997) for all anglers, assuming all
anglers were aware of advisones Restricting the impact to only those anglers with
knowledge of the advisones, resulted in gains of $1 49 per tap ($1997) for all anglers
Restricting the impact to only those angler with knowledge of the advisones and
adjusting impacts by fishing behavior, the authors found that consumption anglers gained
$2 33 per tnp, whereas the catch and release anglers suffered losses of $1 91 per tnp
However, the coefficient on the catch and release coefficient was not significantly
different from zero
The finding that consumption-onented anglers benefit more from eliminating
contamination offish underscores the need to account for awareness and angler response
to FCAs in any models that estimate benefits
Maine FCA Study
MacDonald and Boyle (1997) studied the impact of a statewide FCA on Maine
freshwater anglers The authors used a contingent valuation study to determine the WTP
to remove the advisory From the survey the authors found that 63 percent of the anglers
knew about the advisory, but fewer than 25 percent engaged in any averting behavior
The results of the CV survey found that the seasonal Joss in consumer surplus amounted
to $151 per angler ($1995)
New York Toxics Study
Montgomery and Needelman (1997) used 1989 data on angler tips to estimate the
benefits of removing toxic contamination from all New York State lakes and ponds The
authors relied on data from a survey done for the National Acid Precipitation Assessment
Program (NAPAP) and included data on tnps to lakes or ponds for 266 anglers between
trad- Apnl and October, 1989 The authors also incorporated data from New York
Department of Environmental Conservation (NYDEC) to determine contamination in
lakes The state includes information on the health hazard from consuming fish from
contaminated lakes in a pamphlet that anglers receive along with their fishing license
Out of 2,584 lakes, 23 had toxics advisones, including 9 with 'eat none' and 14 with 'one
meal per month' Lake Ontario and Lake Champlain are listed as contaminated, even
though only a portion of those lakes may be affected This designation operates to bias
downwards the toxics effect since many anglers fishing on these two lakes are not
affected b> the toxics designation
The authors utilized a repeated discrete choice model that estimates the probability of
mabng a fishing tnp and subsequently the probability of fishing at a particular site This
approach allows the presence of FCAs on any site to affect both the choice of site and the
number of taps The choice set of all NY lakes and ponds is too large to incorporate in a
94
-------
model so Ihe authors limited the choice set to the chosen lake and a random selection of
11 other sites within three hours of the angler's residence
The models incorporated the toxics variable both as an indicator of contamination in the
23 lakes with advisones and also in combination with the more restrictive advisory level,
'eat none' The coefficients on the toxics vanable were negative and highly significant m
all specifications, but the coefficient on the more restrictive 'eat none' advisory was
positive and significant, which suggests that this vanable was picking up some
unobserved positive charactenstics of those 9 lakes These coefficients imply that
benefits to eliminate toxic contamination from all 23 lakes amount to about $151 per tnp
($1989) or $0 45 per day, wtoch amounts to a seasonal benefit of about $63 25 per
person This study attempted to quantify the impact of different levels of FCAs in a
random utility framework Unfortunately, the attempt was unsuccessful, as the 'eat none'
vanable did not have the expected negative sign Nevertheless different advisory levels
are likely to have different impacts on angler utility
6.4.2 Summary of Impacts of FCAs
To reiterate, a search of 1he literature did not reveal any studies that estimate the change
in welfare associated with species-specific FCAs for marine recreational anglers The
Green Bay, Lavaca Bay, and Tennessee Lakes I and II studies all evaluated recreational
losses associated with pollution of specific areas In all these studies anglers could avoid
the impact of the mercury advisones by traveling to a more distant, unpolluted site
Unlike freshwater sites species-specific mercury FCAs in an ocean environment greatly
reduce or eliminate the possibility of substitution across sites and force anglers to make
choices without the 'unlimited' consumption of those species in the choice set
Consequently, the impact of species-specific FCAs in an ocean environment is likely to
be greater than in an environment with opportunities to substitute across areas and avoid
the limitations of the advisory One might argue that the ocean is a ncher environment
with a greater variety of species, and this mitigates the effect of the restricted choices
attributable to FCAs On toe other hand, saltwater fishing gear is often mode and
species-specific so that there are additional costs and constraints in targeting different
species
The only study that evaluated angling behavior in a saltwater environment focused on
point-source mercury pollution in Lavaca Bay Because of the point-source nature of the
pollution, recreational fishermen had access to other sites where they could avoid Ihe
limitations of the polluted area The authors found that welfare losses compnsed about 1
percent for boat tnps and 3 percent for shore trips These relatively small losses probably
reflect die fact that anglers found it relatively easy to access other sites with similar
fishing attributes Consequently, this study probably underestimates the impact of
species-specific mercury FCAs by a considerable amount
In the Tennessee Reservoirs I study Jakus et al (1997) determined that an FCA resulted
in a 6 to 8 percent loss in the value of a trip In a later study to evaluate the impacts of an
FCA on striped bass fishermen the Chesapeake Bay Jakus et al, (2002) transferred the 6
95
-------
to 8 percent loss per top for Tennessee anglers to saltwater anglers However, saltwater
anglers with no uncontaminated site alternatives and possibly fewer species alternatives,
depending on mode and gear, might incur significantly greater bsses than freshwater
fishermen (Jakus, personal communication, 8/29/02) The impact of FCAs on trip values
estimated from ihe study of fishing behavior on freshwater Tennessee reservoirs would
seem to represent a lower bound estimate of the impacts of FCAs in the South Atlantic
The Green Bay study used both stated preference and revealed preference data to
estimate the losses associated with toxics pollution in Green Bay, Wisconsin The toxics
resulted in FCAs, and the tradeoff analysis valued changes in FCAs for yellow perch,
walleye, trout/salmon, and smallmouth bass This analysis produced estimates of the
value of changes in FCAs from 'do not eat' to unlimited' consumption Summing utility
from catch parameters and consumption parameters and evaluating it at the marginal
utility of income yielded an estimate of $47 28 per Green Bay fishing day ($1998) as fee
total value of catching and consuming fish '9
Table 6-5 compares the estimated WTP to avoid FCAs to the overall value of a top The
total value of being able to consume the catch ('unlimited') relative to 'do not eat'
amounts to $21 71 per tnp ($1998), which is about 46 percent (=21 71/47 28) of the
value of a tnp Then the change from do not eat' to ' 1 meal per month' is about 25
percent (=11 96/47 28), the change from 'do not eat' to 1 meal per week' is about 42
percent (=19 90/47 28)
Table 6-5
Value of a Change in FCAs as a Percent of the Value of a Green Bay Fishing Tnp
Descriptor WTP Increment Percent of Total
($1998) (%)
Value of Fishing on Green Bay $47 28 100 00%
Value of Catching Fish $25 57 54 09%
Value of Consumption $2171 45 91 %
'Do not eat' to '1 meal/month' $ 11 96 25 29%
'Do not eat' to '1 meal/week1 $ 19 90 42 09%
'Do not eat' to 'unlimited1 $2171 45 91 %
Note 1) Value of fishing is Ihe sum of catching fish and consuming fish.
2) Value of fishing is based on utility from catching fish
3) Value of consumption is based on utility for unlimited consumptioa
Source Estimated from Breffle et al, 1999, Appendix B
The estimated WTP to avoid the consumption impact of FCAs is an underestimate for
some changes This is because Breffle et al did not define each FCA level with the same
consumption descnptor for all species For example, level 9 amounted to 'do not eat' for
salmon-trout walleye, and smallmouth bass but '1 meal per month' for yellow perch
19 This value is not equivalent to a per trip WTP, since it docs not net out the loss in utility from incurring
the launch fee for the fishing tnp
96
-------
Thus, WTP to move from 'do not eat' to 'unlimited' for all four species would be
somewhat greater than $21 71
6.5 Valuation of Changes in Marine FCAs
In January 2001 U S FDA issued a fish consumption advisory for four species It
recommended that pregnant and nursing women and children 'do not eat' swordfish,
shark, king mackerel, or tlefish Table 6-6 shows the resulting mercury concentration
(hat would result from the reductions in mercury under each of the three mercury
reduction scenanos
Table 6-6
Changes in Mean Mercury Concentration for Mercury Reduction Scenanos
For Species Currently Subject to FCAs
(ppm)
Species
King mackerel
Sharks
Swordfish
Tilefish
Mean
095
080
1 00
145
Scenano 1
084
071
088
128
Scenano 2
057
048
060
087
Scenano 3
075
063
079
1 14
Source Compiled from Tables 2-3 and 3-3 in this report
U S FDA currently uses 1 0 ppm as a cntenon value for issuing FCAs However, some
states, health organizations, and EPA use cntenon values that are more stringent
Consequently, it is very uncatain how different regulatory bodies might adjust
consumption guidelines in response to lower mercury concentrations in fish. It is equally
uncertain how consumers react to inconsistent or conflicting advice from vanous health
and regulator} bodies
Acknowledging the uncertainly, the analysis makes assumptions that allow translation of
the changes in mercury concentrations in fish under the vanous mercury reduction
scenanos to changes in FCAs, Since tilefish are not prevalent in the South Atlantic area
and are not a target species for the region's recreational anglers, the analysis focuses on
king mackerel, sharks, and swordfish Under current baseline conditions the three
Atlantic species range in mercury concentration from 0 8 to 1 0 ppm. Scenano 1 reduces
deposition so that the range declines to 0 71 to 0 88 ppm Under scenario 2 the range
declines further to 0 48 to 0 60 ppm, and in scenario 3 the range is from 0 63 to 0 79 ppm
For all three mercury reduction scenanos mercury concentrations for king mackerel,
sharks, and swordfish are below 1 0 ppm FDA's 1 0 ppm cntenon would allow some
consumption, but die recommended level is uncertain For the purposes of this analysis
we assume toe following
Scenano 1 would represent a change from 'do not eat' to '1 meal per month',
97
-------
Scenario 2 would represent a change from 'do not eat' to' 1 meal per week',
and
Scenario 3 would represent a change from 'do not eat' to '2 meals per month'
The values from Table 6-5 associate the change in FCAs under the three mercury
reduction scenarios to a change in tnp value That is, if the change in FCAs implicit in
Scenario 1 approximates a change from 'do not eat' to '1 meal per month' then this
would amount to an increase of about 25 percent in the total per tnp WTP Similarly,
scenario 2 would represent about a 42 percent change in the value of a tnp The change
to '2 meals per monlh' implicit in scenano 3 would range from 25 to 42 percent, and the
analysis uses the midpoint of this range, about 33 5 percent as representative of the
change in FCAs
Table 6-7 combines the percent change in the value of value of a fishing tnp applicable to
each scenano with estimates of the value of a tnp and the number of taps to quantify the
benefits of reducing mercury concentrations in fish King mackerel, sharks, and
swordfish are all coastal/migratory pelagic species, and the value of access to these
species in the South Atlantic area totals about $4 06 ($1997) per tnp Inflating that value
to 2003 dollars using the CPI Index (urban - all items) yields a value of $4 44 per tnp
($2003) This per tnp value applies to tnps for pnvate/rental boats To extend the
analysis to shore tnps the analysis applies the shore to boat ratio from Bockstael et al
(1989), which yields a value of $0 71 per tnp ($2003) It also assumes that the value of a
charter boat tnp is twice the value of a pnvate/rental boat tnp, $8 89 per tnp ($2003)
The benefits to recreational anglers from reducing mercury deposition result from less
restncuve fish consumption advisones As discussed above, the increase in utility from
reducing the stringency of FCAs only applies to anglers who are aware of the advisones
and who value consumption of the catch There are no recent studies that estimate
awareness since FDA announced FCAs in 2001, but it is likely that a high very
percentage of anglers are aware of the advisones Studies reviewed in Section 61 also
indicated that a very high percentage of saltwater anglers consume their catch In the
absence of updated information specific to Soulh Atlantic anglers this analysis assumes
100 percent are aware of advisones and 100 percent consume their catch
Scenano 1, which approximates a change m fish consumption advisones from 'do not
eat' to '1 meal per month', yields benefits of about $3 6 million ($2003) Scenano 2
represents a change in fish advisones from 'do not eat' to '1 meal per week' and
generates benefits of about $5 9 million Scenano 3 reduces U S mercury deposition by
about 51 percent and causes a change m advisones from 'do not eat' to about '2 meals
per month', total benefits amount to about $4 7 million ($2003)
98
-------
Table 6-7
Benefits to Recreational Anglers from Changes in FCAs ($2003)
Mode Recreational Trips WTP per trip Scenano 1 Scenario 2 Scenano 3
(Smiluons) ($2003) (millions} (millions) (millions)
Shore 922 $071 $018 $030 $024
Boat 841 $444 $112 $187 $150
Charter 045 $889 $225 $374 $299
Total 1808 $355 $591 $473
Note: 1) Recreational mps are based on the a\erage over the last 5 years, as shown in Table 6-
2
2) Willingness to pay for boat tnps is based on total for coastal/migratory pelagic
species
3) WTP for shoie tnps is about 16 percent of boat tnps
4) WTP for charter tnps is assumed to be twice that of boat tnps
5) Consumer Pnce Index - Urban, all items 1997=160 5, 2003=183 3
6) Scenano 1 applies a change of 25 3 percent to the value of a top
7) Scenano 2 applies a change of 42 1 percent to the value of a tnp
8) Scenano 3 applies a change of 33 5 percent to 1he value of a tnp
Source: Computed from Tables 6-2, 6-3, and 6-5, and information in the text of this report
6.6 Assumptions and Uncertainties
Several important assumptions and uncertainties underlie these benefits estimates Most
important is the companson between the utility of catching fish with the utility of
consuming fish The inclusion of both catch rate and FCA levels in the Green Bay study
permits estimation of the relative contribution of each to total utility Applying these
relative shares to saltwater species assumes that Ihe ratio of utility of catching fish to
consuming fish is the same This need not be the case, since the vanous species of ocean
fish subject to FCAs may taste differently than freshwater fish and the catch rates and
gear/mode requirements are eilso likely to be quite different The magnitude and direction
of any bias from applying the Green Bay values to saltwater anglers is unknown
It is also likely that reducing ihe stringency of FCAs will generate more recreational
fishing tops Breffleetal (1999) estimated an increase in Green Bay fishing taps of
about 14 percent if FCAs were eliminated However, the behavior of freshwater anglers
with multiple site choice alternatives is not readily transferable to an ocean environment
Hence, the analysis undervalues the changes in FCAs by failing to account for any
increase in angler-tops
The benefits of reducing or eliminating FCAs apply only to those anglers who are aware
of the FCAs and value consumping of the catch In the absence of updated information
on awareness and area-specific information on the percent of anglers who consume their
catch, the analysis assumes the benefits apply to all anglers This assumption probably
results in some upward bias in the estimates, but the magnitude of that bias is unknown
99
-------
The benefit estimates also depend on the value of a tap targeted at coastal
migratoiy/pelagic species Haab et al (2000) estimated per tnp values that totaled $3 87
($1997) for the four states m the study area. Because of the reduced substitution
possibilities ihe loss in utility from eliminating access to this group of species over all
four states is greater lhan the sum for each state individually The analysis attempts to
account for this underestimate by adjusting the estimates based on Ihe ratio of the total
per tnp value for the entire Gulf and Atlantic areas to the sum of Ihe per tnp estimates for
each state However, this area is greater than just the four-state Atlantic area, and the
ratio probably results in an adjustment that may be slightly upward biased
Two other assumptions extend the value per boat-tnp estimates from the Haab et al study
to shore and charter trips To compute shore tnps the analysis applies a ratio from the
Bockstael et al (1989) study This value denves from comparison of the value of shore
tnps and boat taps in Florida from 1987 data It is uncertain whether that ratio applies to
other states and also whether that ratio has remained stable over time, but the magnitude
and direction of any bias are unknown None of the studies reviewed provided estimates
of the value of charter tnps targeted at coastal/migratory pelagic species In the absence
of information the analysis assumed a value for charter tnps equal to twice the value of
boat tnps Since charter tnps are typically expensive, this per tnp value is probably a
conservative estimate, but it is subject to considerable uncertainty
A final area of major uncertainty relates to how reductions in mercury concentrations in
fish translate to changes in fish consumption advisones The analysis based its
determination of changes in FCA levels based on the 1 0 ppm criterion used by U S
FDA, but EPA and state health organizations use different catena for issuing FCAs
Thus, how any agency would respond to the changes in mercury concentrations in fish is
subject to considerable uncertainty Again, the magnitude and direction of any bias
resulting from the assumptions used in the analysis are unknown
The benefits transfer approach requires that the benefits obtained in a study of one site
pertain to another To assess the validity of the approach, as applied to valuing the
impact of FCAs on marine recreational fishing, it is essential to compare the targeted site
to the site of the study Three questions arise
Are the sites similar'?
Are the populations similar?
Are the changes being valued similar?
The answers to questions 1 and 3 suggest considerable uncertainty in transferring benefits
from the Green Bay study to the South Atlantic study area. Green Bay is an arm of Lake
Michigan, and no lake really compares to a broad expanse of ocean The species found in
the South Atlantic from North Carolina to Flonda also differ significantly from
freshwater species, such as walleye, trout, salmon, bass, and perch The fishing modes
are similar, shore and boat, but saltwater gear differs greatly from freshwater gear
because both the fish and the ocean environment differ from lakes
100
-------
The change being valued differs in important ways between the Green Bay study and the
study area The Green Bay study valued a change in angler utility reflected in different
levels of FCAs for four species targeted in the polluted area of Green Bay The change
attributable to mercury FCAs in the Southeast Atlantic reflects the utility to anglers that
results from a change in FCAs for coastal/migratory pelagic species that range across a
large ocean area In the Green Bay case anglers can target the species subject to FCAs by
traveling to other sites, whereas in the South Atlantic there are no substitute sites where
king mackerel, swordfish, and sharks are not subject to mercury fish consumption
advisories
101
-------
7.0 Research Needs
A key objective of tins research exercise is to identify ihe major gaps in knowledge that
undermine the reliability of any estimates of the benefits of reducing mercury deposition
Each section of this report summarized the key assumptions used to bndge those gaps
and enable the benefits analysis to proceed This section revisits those gaps in our
knowledge and identifies research pnonties to enhance our understanding in key areas
Section 1 clanfied two key assumptions that are fundamental to the report
reductions in mercury concentrations in fish are proportional to reductions in
deposition, and
duration of time between changes in deposition and steady state changes in
mercury concentrations in fish is not so long that discounting would significantly
reduce the values
Research on the fate and transport of mercury in airborne and aquatic environments is
ongoing However, much of tins research has been targeted at freshwater aquatic
environments, and there are major gaps in our knowledge of how mercury deposition
affects fish in saltwater environments An important priority is to enhance the state of
knowledge of the fate and transport of mercury in saltwater environments so that it
is possible to quantify the link between mercury emissions and mercury uptake by marine
fish and shellfish in different locations with reasonable precision
Section 2 identified the study area and summarized the baseline concentrations of
mercury in various saltwater species offish based on EPA's Mercury in Marine Life
database This database is a major advance in our level of understanding of how mercury
is affecting a wide variety of saltwater species in different areas However, this database
has relatively sparse coverage for some key recreational and commercial species in some
areas and for most species in key coastal areas such as the West Coast MML's
coverage of species and areas is not sufficient at this point in time to provide a
baseline estimate of mercury intake for key species harvested in all U.S. waters, and
continued data collection must be an important priority for future mercury
research.
Section 3 relied on the hneanty assumption to translate changes in emissions into
changes in deposition and uptake by fish This has two components The direct effect
results from controlling mercury deposition from U S sources, and the indirect effect
results from controlling the fraction of U S mercury emissions that enter the global pool
Because prevailing winds blow pollution eastward, U S emissions of mercury have a
greater impact on Atlantic and Gulf coast ecosystems than on Pacific coast ecosystems
However, more detailed knowledge of the spatial impact of mercury emissions on
saltwater ecosystems is critical to any national mercury benefits analysis
Section 4 estimated the benefits of reducing mercury emissions on human health There
are a number of research pnonties needed to increase the accuracy and reliability of this
102
-------
assessment First, the National Health and Nutrition Examinations Survey
(NHANES) provides an excellent baseline estimate offish consumption, but several
enhancements are required to reduce the uncertainties of using these data. More
current data on consumption since 2001 and 2002 will provide a better baseline offish
consumption that reflects reaait changes resulting from U S FDA's national fish
consumption advisories and e-flbrts by states to educate the public about the nsks of
consuming fish These data should be released in 2004 Looking ahead to future
surveys, NHANES questions on fish consumption need refinement to enable researchers
to identify more precisely the species consumed. This includes more detailed species
vanables that separate the impacts of consuming freshwater from marine species
Changes to the NHANES questionnaire will not provide usable data until the next round
of surveying in 2005, but should be easy to implement Currently, NHANES does not
include geographical vanables that enable researchers to estimate differences in fish
consumption between coastal and inland areas Information on fish consumption in
different regions is cntcal to an accurate estimate of baseline mercury exposure, since
coastal populations consume a greater mix of marine species than inland populations
NHANES is reluctant to release this information to researchers because the sampling was
not designed to compare coa&tal with inland, but it does exist
Second, there is uncertainly about which dose-response function to use for neuro-
developmental health effects. NRC recommended relying on endpomts from the Faroe
Islands study for a mercury reference dose (RfD) However, economists are not able to
quantify the available endpomts, such as the Boston Naming Test, into an estimate of
benefits To use the Faroe Islands study for benefits estimates requires either translation
of the Faroe Islands endpomts into an estimate of impacts on IQ, or estimation of
willingness to pay to avoid all theneurodevelopmental endpomts associated with fetal
exposure to mercury (see below) Only the New Zealand studv includes an IQ endpoint
that economists can translate into benefits Currently, there are two dose-response
functions for this endpoint one estimated with a threshold and one with no threshold
They result in very different estimates of impacts, and EPA should request NRC to
resolve the uncertainty over which functional form is most appropriate EPA should also
request NRC to resolve whether thresholds exist for paraesthesia and other mercury
endpomts
Third, the effect of mercury on cardiovascular function is a major source of
uncertainty. This uncertainly pertains to the impact of mercury on cardiovascular
function in both adults and children. If mercury does affect cardiovascular function, it is
a major pnonty to determine the magnitude of its impact Funding is needed to replicate
the studies of cardiovascular effects on adults, both male and female, and children in U S
populations
Fourth, reliance on foregone earnings and costs of illness as a measure of benefits
for mercury endpoints underestimates the benefits of reducing mercury exposures.
Funding is needed for a willingness-to-pay study of avoiding neurodevelopmental
endpoints of mercury exposure Funding is also needed for a willingness-to-pay study of
avoiding cardiovascular endpoints in both adults and children
103
-------
Section 5 attempted to assess the impact offish consumption advisones and publicity
about tiie health effects of mercury on tfie commercial fishing industry A review of the
literature found no studies that address this issue In the absence of such studies this
analysis employed an econometric approach to determine the effect offish consumption
advisones on swordfish and king mackerel, but did not have sufficient time or resources
to identify an impact EPA and NMFS should undertake an investigation of the
economic impact of mercury fish advisones and publicity on health risks on the
commercial fishing industry, including the harvesting, wholesale, and retail sectors.
Section 6 used a benefits transfer approach to quantify the impact of mercury fish
consumption advisones on manne recreational anglers The economic literature provides
no studies that quantify the response of saltwater anglers to fish consumption advisones
on species across a broad area. Instead, the analysis transferred results from a contingent
valuation study of a polluted area in Green Bay to a saltwater environment, which entails
considerable uncertainty Funding an economic study of how saltwater anglers
respond to mercury fish consumption advisones should be a high priority. The data
exist to use a random utility modeling approach that compares behavior before and after
the announcements of FCAs Alternatively, CV approaches similar to the Green Bay
study could be applied to saltwater anglers
104
-------
8.0 References
Adamowicz, W L, 1998, Review Comments on "Technical Memorandum on Lavaca
Bay Recreational Fishing Assessment," December 29
Apelberg, B, K. Davidson, L Deck, A Hallberg, D McCubbm, and E Post, 2003,
Preliminary Nonroad Landbised Diesel Engine Rule Air Quality Estimation, Selected
Health and Welfare Benefits Methods, and Benefit Analysis Results, U S Environmental
Protection Agency, Office of Air Quality Planning and Standards, Research Triangle
Park, N C, April
Abt Associates, 2002, Mercury Toxicity Workshop Notes, EPA Conference, Washington,
D C , November 4, 2002
Abt Associates, 2003, "Chapter 14 - Lead-Related Benefits," Economic and
Environmental Benefits Analysis Document for the Final Effluent Limitations Guidelines
and Standards for the Metal Products and Machinery Point Source Category, EPA-821-B-
03-002, U S Environmental Protection Agency, Office of Water Washington, D C,
February
Albenra, A, M Cropper, A Krupnick, and N B Simon, 2002, Does the Value of a
Statistical Life Vary with Age and Health Status'' Evidence from the United States and
Canada, Discussion Paper 02-19, Resources for the Future, April
Belton, T, Roundy, R, and Weinstein, N, 1986, "Urban Fishermen Managing the Risks
of Toxic Exposure," Environment, 28 19-37
Bockstael, N E, McConnell K.E, and Strand, IE, 1989, "A Random Utility Model for
Sportfishing Some Preliminary Results for Flonda/'Marme Resource Economics, Vol
6, 245-260
Breffle, W S, Morey, E R, Rowe, RD, Waldman, D M, and Wynnck, S M, 1999,
Recreational Fishing Damages from Fish Consumption Advisones in the Waters of
Green Bay, Boulder, CO St atus Consulting, Incorporated
Buerger, R and Kahn, J R, "New York Value of Chesapeake Stnped Bass," 1989,
Marine Resource Economics, 6 19-25
Burger, J, Staine, K., and Gochfeld, M, 1993, "Fishing in Contaminated Waters
Knowledge and Risk Perception of Hazards by Fishermen in New York City," Journal of
Toxicology and Environmental Health 39 95-105
Carnngton, C D and M P Bolger, 2002, "An Exposure Assessment for Methylmercury
from Seafood for Consumers in the United States," Risk Analysis, Vol 22, No 4 Pp
689-699
105
-------
Centers for Disease Control and Preventon (CDC), National Center for Health Statistics
(NCHS), Division of Health Examination Statistics (DHES), 2003, NHANES 1999-2000
Data, Docs, Codebooks, and SAS Code, available online at
http //www cdc gov/nchs/about/major/nhanes/NHANES99_00 htm.
Centers for Disease Control and Prevention (CDC), National Center for Health Statistics
(NCHS), Division of Health Examination Statistics (DHES), NHANES 1999-2000
Addendum to the NHANES III Analytic Guidelines, Last Update 8/30/02, available
online at http //www cdc gov/nchs/data/nhanes/guidelinesl pdf
Centers for Disease Control and Prevention (CDC), CDC WONDER at
http //wonder cdc gov/mortICD9J html
Cocca, Paul, 2001, Mercury Maps, A Quantitative Spatial Link Between Air Deposition
and Fish Tissue, Peer Reviewed Final Report, Environmental Protection Agency,
Standards and Health Protection Division, Office of Science and Technology, Office of
Water, Washington, D C , September
Connelly, N A, and Knuth, B A, 1995, "Patterns of Harvest and Consumption of Lake
Champlain Fish and Angler Awareness of Health Advisones," Cornell University,
Human Dimensions Research Unit, Ithaca, NY
Connelly, N A, Knuth, B A, and Bisogra, C A, 1992, Effects of Health Advisory and
Advisory Changes on Fishing Habits and Fish Consumption in New York Sport
Fishenes, Cornell University, Human Dimensions Research Unit, Ithaca, NY, Project No
R/FHD-2-PD, Series 92-9, September
Connell), N A, Knuth, B A, and Vena, J E, 1993, New York State Angler Cohort
Study Health Advisory Knowledge and Related Attitudes and Behavior, With a Focus on
Lake Ontario, Cornell University, Human Dimensions Research Unit, Ithaca, NY, Senes
93-9, September
Cropper, M and P Portney, 1994, "Preferences for Life-Saving Programs How the
Public Discounts Time and Age", Journal of Risk and Uncertainty, 1994
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 Analysis, Volume 18 No 6 701-
713
Cunningham, P, W Cooter, and E Sullivan, 2003, Mercury in Marine Life, Draft
Report, Research Triangle Institute, Research Tnangle Park, NC, Prepared for the U S
Environmental Protection Agency, Oceans and Coastal Protection Division, Washington,
D C, January
Davidson, P W, Myers, G J, Cox, C , Axtell, C, Shamlaye, C , Sloane-Reeves, J,
Cermchian, E, Needham, L, Choi, A, Wang, Y, Berlin, M, and Clarkson, T W, 1998,
106
-------
"Effects of Prenatal and 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 280 701-707
Davidson, P W, Myers, G \, Shamlaye, C , Cox, C , Gao, P Axtell, C , Moms, J,
SIoane-Reeves, J, Cemichian, E, Choi, A, Palumbo, D, and Clarkson, T W, 1999,
"Association Between Prenatal Exposure to Methylmercury and Developmental
Outcomes in Seychellois Crrldren Effect Modification by Social and Environmental
Factors," Neurotoxicology, 20(5) 833-841
Degner, R L, C M Adams, S D Moss, and S K Mack, 1994, Per Capita Fish and
Shellfish Consumption in Florida, Flonda Agricultural Market Research Center,
submitted to Flonda Department of Environmental Protection, Contract WM-475, August
31
Flonda Department ol Envronmental Protection, 2003, Integrating Atmospheric Mercury
Deposition with Aquatic Cycling in South Flonda, An approach for conducting a Total
Maximum Daily Load analysis for an atmospherically denved pollutant, November
Flonda Fish and Wildlife Conservation Commission, undated, Mercury in Manne Fish,
http //mannefisheries oiE/Pufas/mercury hun
Flonda Manne Research Institute, undated, Mercury in Manne and Estuanne Fishes of
Flonda An Overview,
http //www flondamanne oig/dg Its/id 9466/features view_aracle htm.
Freeman, A Mynck, III, 1995, "The benefits of water quality improvements for marine
recreation a review of the empncal evidence,' Manne Resource Economics, 10(4) 385-
406
Grandjean, P , Weihe, P, White, R F, Debes, F, Araki, S , Yokoyama, K., Murata, K,
Sorenson, N, Dahl, R, and Jorgensen, P J, 1997 "Cognitive Deficit in 7-Year-Old
Children with Prenatal Exposure to Methylmercury," Neurotoxicology and Teratology
19 417-428
Grandjean, P, Weihe, P, White, R F, and Debes, F, 1998 "Cognitive performance of
children prenatally exposed to "safe" levels of mefliylmercury," Environ Res, 77(2)
165-172
Grandjean, P, White, R., Nielsen, A, deary, D, and de Ohveira Santos, E C, 1999
"Methylmercury neurotaxialy in Amazonian children downstream from gold mining,"
Environmental Health Perspectives , 107(7) 587-591
Gray, Gerry W, Kline, Lisa L, Osbom, Maury F Salz, Ronald J, Van Voorhees, David
A, and Witag, John F, 1994, MRFSS User's Manual, A Guide to Use of the National
107
-------
Manne Fisheries Service Manne Recreational Fisheries Statistics Survey Database,
Special Report No 37 of the Atlantic States Manne Fishenes Commission, December
Guallar, E,M I Sanz-Gallardo, P Van'tVeer, P Bode, A Aro, J Gomez-Araacena, J
D Kark, R A Riemersma, J M Martin-Moreno, and F J Kok, 2002, "Mercury, Fish
Oils, and the Risk of Myocardial Infarction," New England Journal of Medicine, Vol
347, No 22, November 28, 2002, pp 1747-1754
Green, G,C B Moss, and T H Spreen, 1997, "Demand for Recreational Fishing m
Tampa Bay, Flonda A Random Utility Approach," Marine Resource Economics, Vol
12, pp 293-305
Haab, Timothy C, Whitehead, C , and McConnell, Ted, 2000, The Economic Value of
Manne Recreational Fishing in the Southeast United States, 1997 Southeast Economic
Data Analysis, Final Report, July
Hicks, R, S Stemback, A Gautam, and E Thunberg, 1999, Volume II The Economic
Value of New England and Mid-Atlantic Sportfishing in 1994, National Manne Fishenes
Service. NOAA Technical Memorandum NMFS-F/SPO-38, August
IPCS (International Programme on Chemical Safety), 1990, Environmental Health
Cntena Document 101 - Methylmercury, Geneva World Health Organization
Jakus, P M, Dadakas, D, and Fly, J Mark, 1998, "Fish Consumption Advisories
Incorporating Angler-Specific Knowledge, Habits, and Catch Rates in a Site Choice
Model" American Journal of Agricultural Economics, 80(5) 1019-1024
Jakus, P M, Downing, M, Bevelhimer, M, and Fly, J M, 1997, "Do Sportfish
Consumption Advisones Affect Reservoir Anglers' Site Choice9" Agricultural and
Resource Economics Review, 26 196-204
Jakus, P, M McGuinness, and A Krupnick, 2002, The Benefits and Costs of Fish
Consumption Advisories for Mercury, Resources for the Future, Washington, D C July
Just, Richard E and Hueth, Darrell L, 1979, "Welfare Measures in aMdtimarket
Framework," The American Economic Review, December
Kakalik, J Et al, 1981, The Cost of Special Education, Rand Corporation Report N-
1791-ED
Kjellstrom, T, Kennedy, P, Wallis, S, and Mantell, C, 1986 "Physical and Mental
Development of Children with Prenatal Exposure to Mercury from Fish Stage 1
Preliminary tests at age 4," National Swedish Environmental Protection Board Report
No 3080 Solna, Sweden
108
-------
Kjellstrom, T, Kennedy, P, Wallis, S, Stewart, A, Fnberg, L, Lind, B, Wutherspoon,
T, and Mantell, C, 1989 "Physical and Mental Development of Children with Prenatal
Exposure to Mercury from Fish " National Swedish Environmental Protection Board
Report No 3642
Knuth, B A, Connelly, N A, and Shapiro, M A, 1993, Angler Attitudes and Behavior
Associated with Ohio River Health Advisories, Ithaca, NY Cornell University Human
Dimensions Research Unit Series No 93-6
Kneger, D J, and Hoehn, J P , 1998, "Improving Health Risk Informabon for Sport
Anglers The Value of Information About Chemical Residues," North American Journal
of Fisheries Management, 18 411-21
Lamborg, C H, W F Fitzgerald, J O'Donnell, and T Torgersen," 2002, "A non-steady-
state compartmental model of global-scale mercury biogeochemistry with
interherrusphencgradients," Geochimica etCosmochemicaAtca,Vol 66, No 7, pp
1105-1118
Lee, D J, S L Larkin, and C M Adams, 2000, "A Bioeconomic Analysis of
Management Alternatives for the U S North Atlannc Swordfish Fishery," Marine
Resource Economics 15 77-96
Lipton, Douglas W, 1986, "The Resurgence of the U S Swordfish Market," Marine
Fisheries Review, 48(3) 24-27
MacDonald, H F, and Boyle, KJ, 1997, "Effect of a Statewide Sport Fish Consumption
Advisory on Open-Water Fishing in Maine," North American Journal of Fisheries
Management, 17 687-95
May, H, and Burger, J, 1996, "Fishing in a Polluted Estuary Fishing Behavior, Fish
Consumption, and Potential Risk," Risk Analysis, 16 459-71
McConnell, K.E, and Strand, IE, 1994, "The Economic Value of Mid and South
Atlantic Sportfishing, Vol 2 " 134, College Park, MD University of Maryland
Montgomery, M, and M Needelman, 1997, "Trie Welfare Effects of Toxic
Contamination in Freshwater Fish," Land Economics 73 211 -23
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 High Blood Pressure Education Program, undated, Update on the Task Force
Report (1987) on High Blood Pressure in Children and Adolescents, A Working Group
Report, 1996, National Institutes of Health, U S Department of Health and Human
Services publication NIH 95-3790, available online at
hitp //www nhlbi nth gos/hcdth'frrolfliedrt'hbp/hbp ned pdf.
109
-------
N1EHS (National Institute of Environmental Health Sciences), 1998, Scientific Issues
Relevant to Assessment of Health Effects from Exposure to Metiiylmercuiy, Report of
the Workshop on Scientific Issues Relevant to Assessment of Health Effects from
Exposure to Methylmercury, Nov 18-20, Raleigh, N C
National Manne Fisheries Service, Sustainable Fishenes Division, 2003, Stock
Assessment Analyses on Spanish and King Mackerel Stocks, Southeast Fishenes Science
Center, Miami, FL, April
National Research Council, 2000, Toxicological Effects ofMeHg, Committee on
Toxicological Effects of MeHg, Board on Environmental Studies and Toxicology,
Commission on Life Sciences, National Research Council, National Academy Press,
Washington, D C
North Carolina Department of Health and Human Services, 2000, Four Southeast States
Issue Joint King Mackerel Health Advisory NC Sets Public Information Meetings to
Discuss Advisory, March 23, http /-'www dhhs statenc us/resssrel/3-23-00 htm.
Okey, T A, 2001, "A 'Straw-man' Ecopath Model of the Middle Atlantic Bight
Continental Shelf, United States," in Fisheries Impacts on North Atlantic Ecosystems
Models and analyses, eds A Guenette, V Chnstensen, and D Pauley (eds), Fishenes
Centre Reports, University of British Columbia, Canada, Volume 9, No 4, pp 151-166
Oken, E, K P Klemman, W E Borland, S R Simon, J W Rich-Edwards, and M W
Gillrnan, 2003, "Decline in Mercury Among Pregnant Women After a National Mercury
Advisory," Obstetrics and Gynecology, Vol 102, No 2, August
Pacyna, J M, E G Pacyna, F Steenhuisen, and S Wilson, 2003, "Mapping 1995 global
anthropogenic emissions of mercury," Atmospheric Environment, Vol 37 Supplement
No l.pp S109-S117
Parsons, George R., 1998, Comments on the "Technical Memorandum on Lavaca Bay
Recreation Fishing Assessment," December
Phlugh, K K., Lung, L, von Hagen, L A, von Hagen, S, and Burger J, 1999, "Urban
Anglers' Perception of Risk from Contaminated Fish," The Science of the Total
Environment, 228 203-18
Portier, K M, Um, Y, R L Degner, R L, S K. Mack, and C M Adams, 1995,
Statistical Analysis of Flonda Per Capita Fish and Shellfish Consumption Data, Florida
Agricultural Market Research Center Industry Report 95-1, submitted to the Flonda
Department of Environmental Protection, Contract SP332, December
Rae, D, 2002, An Assessment of Information and Data Necessary to Quantify Impacts
and Benefits of Reducing Mercury in Saltwater Fish, Final Report, for Office of
110
-------
Wetlands, Oceans, and Watersheds, U S Environmental Protection Agency, Washington,
D C, October
Raitakan, O, M Juonala, M Kahonen, L Taittonen, T Laitinen, N Maki-Torkko, M
Jarvisalo, M Uhan E Jokirien, T Ronnemaa, H Akerblom, and J Vnkan 2003,
"Cardiovascular Risk Factors in Childhood and Carotid Artery Intma-Media Tluckness
in Adulthood The Cardiovascular Risk in Young Finns Study, Journal of the American
Medical Association, Vol 290 (17), November, pp 2277-2283
Round, M, 2002, Excerpts from the Analysis of the NESCAUM Fish Consumption
Survey, Northeast States foi Coordinated Air Use Management, May
Roy, N, E Tsoa, and W E Schrank, 1991, "What Do Statistical Demand Curves Show?
A Monte Carlo Study of the Effects of Single Equation Estimation of Groundfish
Demand Functions," In Econometric Modeling of the World Trade in Groundfish, (eds
W E Schrank and N Roy), 572 The Netherlands Kluwer Academic Publishing
Salkever, D, 1995, "Updated Estimates of Earnings Benefits from Reduced Exposure of
Children to Environmental Lead" Environmental Research 70 1-6
Salonen, J T, K Seppanen, H Nyyssonen, H Korpela, J Kauhanen, J Kantola, J
Tuormlehto, H Esterbauer, F Tatzber, and R. Salonen, 1995 "Intake of Mercury from
Fish, Lipid Peroxidatton, and the Risk of Myocardial Infarction and Coronary,
Cardiovascular, and Any Death in Eastern Finnish Men " Circulation, 91 645-55
Seigneur, C, undated, "Global Emissions Inventory Activity, Review Mercury,"
http //www geiacenter org/reviews/mercury html
Smith, V Kerry, 1998, Comments on the Technical Report on the Lavaca Bay
Recreational Fishing Assessment, memo to David J Chapman and Douglas MacNair.
December 30
Serensen, N, K. Murata, E Butz-Jergensen, P Weihe, and P Grandjean, 1999 "Prenatal
Methylmercury Exposure as a Cardiovascular Risk Factor at Seven Years of Age '
Epidemiology 10 370-5
Standing Committee on Research & Statistics (SCRS), 2003, Report of the Standing
Committee on Research & Statistics, International Commission for the Conservation of
Atlantic Tunas, website hflp //www iccat es
Strand, 1, J E Kirkley, M M Mamon, R. G Black, J Renshaw, and R. E Unsworth,
2002, Economic Benefits fiomthe Successful Restoration of the Atlantic Stnped Bass
Fishery 1981-1999, June
Texas General Land Office, Texas Parks and Wildlife Department, Texas Natural
Resources Conservation Department, National Oceanic and Atmospheric Administration,
111
-------
and The U S Fish and Wildlife Service, 1998, Technical Memorandum on Lavaca Bay
Recreational Fishing Assessment, Draft for Peer Review, November
Thurman, Walter N, and J E Easley, Jr, 1992, "Valuing Changes in Commercial
Fishery Harvests A General Equilibrium Derived Demand Analysis," Journal of
Environmental Economics and Management, 22 226-240
U S Environmental Protection Agency, 1997, Mercury Study Report to Congress,
Volumes I-VIII, Office of Air Quality Planning and Standards and Office of Research
and Development, Washington, D C, December
US Environmental Protection Agency 1999 The Clean Air Act Amendments (CAAA)
Section 812 Prospective Study of Costs and Benefits (1999) Advisory by the Health and
Ecological Effects Subcommittee on Initial Assessments of Health and Ecological
Effects, Part 1 Prepared by the Health and Ecological Effects Subcommittee (HEES) of
the Advisory Council on the Clean Air Compliance Analysis, Science Advisory Board,
US Environmental Protection Agency Washington, DC EPA-SAB-Council-ADV-99-
012 July 28
U S Environmental Protection Agency 2000, Guidelines for Preparing Economic
Analyses, Washington, D C, EPA-R-00-003, September
U S Environmental Protection Agency, 2003, Clear Skies Initiative, Office of Air
Quality Planning and Standards and Office of Research and Development, Raleigh-
Durham, NC
U S Food and Drug Administration, 2001, Consumer Advisory, Center for Food Safety
and Applied Nutrition, "An Important Message for Pregnant Women and Women of
Childbeanng Age Who May Become Pregnant About the Risks of Mercury in Fish,"
March, http //vmcfsan,fdagov/-dms/admehg html, March
U S Food and Drug Administration, Center for Food Safety and Applied Nutrition, 2001,
"Mercury Levels in Seafood Species," May, available online at
http //\vw\v cfsan fdagovA-trfi sea-mehg html
Walsh, Richard G, Johnson, Donn M, and McKean, John R, 1992, "Benefit transfer of
outdoor recreation demand studies," Water Resources Research, 28(3) 707-713
Wessels, C R and Anderson, J L, 1992, "Innovations and Progress in Seafood Demand
and Market Analysis," Marine Resource Economics, 1 209-228
Wittels, E H, J W Hay, and A M Gotto, Jr, 1990, "Medical Costs of Coronary Artery
Disease in the United States," The American Journal of Cardiology, Volume 65, pp 432-
440
112
-------
Yoshizawa, KL, D B Rimm J S Moms, V L Spate, C Hsieh, D Spiegleman, M J
Stampfer, and W C Willette 2002, "Mercuiy and the Risk of Coronaiy Heart Disease in
Men," New England Journal of Medicine, Vol 347 No 22, November 28, 2002, pp
1755-1760
Telephone Interviews
Atkinson, D, U S EPA, personal communication, 9/27/03, 202-566-1226
Bellinger, D, Boston Children's Hospital, personal communication, 9/19/03,617-355-
6565
Bullock, R., U S EPA personal communication, 10/27/03,919-541-1349
Feener, S , Gorton's, personal communication, 10/3/03, 978-283-3000
Jakus, P, University of Utah personal communication, 8/29/02 435-797-2309
Koemg, A, Harvard Center Tor Risk Analysis, personal communication, 11/14/03,6} 7-
432-6933
Round, M, Northeast States for Coordinated Air Use Management, personal
communication, 11/26/03,6)7-367-8540
Sasiela, R, Coldwater Seafood, personal communication, 10/3/03,410-228-7500
Schober, S, National Center for Health Studies (NCHS), personal communication,
10/1/03, 301-458-4484
Sinks, T, Centers for Disease Control and Prevention (CDC), personal communication,
9/24/03
113
------- |