United States
Environmental Protection
Agency
EPA-452/R-96-001f
June 1996
Air
Mercury Study
Report to Congress
Volume VI:
Characterization of Human
Health and Wildlife Risks from
Anthropogenic Mercury
Emissions in the United States
SAB REVIEW DRAFT
f/EPA
Office of Air Quality Planning & Standards
and
Office of Research and Development
066011-2-6
-------
MERCURY STUDY REPORT TO CONGRESS
*.
VOLUME VI:
CHARACTERIZATION OF HUMAN HEALTH AND
WILDLIFE RISKS FROM ANTHROPOGENIC MERCURY EMISSIONS
\IN THE UNITED STATES
SAB REVIEW DRAFT
June 1996
U.S. Environmental Protection Agency
Region 5, Library
-------
TABLE OF CONTENTS
LIST OF TABLES iii
LIST OF FIGURES iv
LIST OF SYMBOLS, UNITS AND ACRONYMS . v
U.S. EPA AUTHORS vi
SCIENTIFIC PEER REVIEWERS viii
WORK GROUP AND U.S. EPA/ORD REVIEWERS ix
EXECUTIVE SUMMARY x
1. INTRODUCTION '1-1
2. HUMAN HEALTH EFFECTS: HAZARD IDENTIFICATION AND DOSE-
RESPONSE 2-1
2.1 Health Hazards Associated with Mercury Exposure 2-1
2.2 Dose-Response to Methylmercury 2-3
2.2.1 Calculation of Methylmercury RfD 2-3
2.2.2 Human Dose-Response Issues 2-7
2.3 Uncertainty in the Human Health RfD for Methylmercury 2-16
2.3.1 Qualitative Discussion of Uncertainties in the RfD for Methylmercury
Alternate Analyses 2-16
2.3.2 Quantitative Analysis of Uncertainty in the Methylmercury RfD 2-19
3. WILDLIFE HEALTH EFFECTS: HAZARD IDENTIFICATION AND DOSE-
RESPONSE 3-1
3.1 The Framework for Ecological Risk 3-1
3.2 Health Hazards of Methylmercury Exposure to* Wildlife 3-2
3.2.1 Mammalian Species 3-2
3.2.2 Avian Species 3-3
3.3 Dose-Response to Methylmercury for Wildlife Species 3-3
3.3.1 Mammalian Species 3-3
3.3.2 Avian Species 3-4
3.4 Wildlife Criteria 3-5
3.4.1 Wildlife Criteria Methodology 3-5
3.4.2 Bioaccumulation Factors 3-6
3.4.3 Other Exposure Parameters 3-7
3.4.4 Health Endpoint (TD) 3-8
3.4.5 Calculation of Wildlife Criterion Values 3-8
3.5 Uncertainty Around the Dose-Response Assessments for Methylmercury 3-11
3.5.1 Uncertainty in the Wildlife Criteria 3-11
3.5.2 Sensitivity Analysis 3-11
3.5.3 Uncertainties Associated with the GLWQI Methodology 3-12
4. CHARACTERIZATION OF MERCURY EXPOSURE OF SELECTED HUMAN AND
WILDLIFE POPULATIONS 4-1
4.1 The Modeling Analysis 4-1
June 1996 • i SAB REVIEW DRAFT
-------
TABLE OF CONTENTS (continued)
Page
4.1.1 Study Design of the Modeling Analysis 4-1
4.1.2 Uncertainties and Defaults Used in Exposure Modeling 4-7
4.2 Estimates of Methylmercury Exposure Based on Monitoring Data, Dietary
Surveys and Mercury Residue Data 4-11
4.2.1 Non-Human Mammalian Species Exposures to Methylmercury 4-11
4.2.2 Avian Species Exposure to Methylmercury 4-13
4.2.3 Human Intake of Methylmercury Estimated Through Dietary Surveys
and Mercury Residue Data 4-16
4.3 Estimates of Sizes of At-Risk Populations -. 4-35
4.3.1 Human Populations 4-35
4.3.2 Estimates for the Size of the Piscivorous Wildlife Population 4-46
5. INTEGRATP/E ANALYSIS FOR METHYLMERCURY 5-1
5.1 Characterization of Risk: Quantitative Integration of Human and Wildlife
Exposure and Dose-Response 5-1
5.1.1 Introduction 5-1
5.1.2 Description of Subsistence Fishers 5-1
5.2 Integration of Modeled Methylmercury Exposure Estimates for Humans and
Wildlife with the Dose-Response Assessments 5-2
5.2.1 Methylmercury Intake by Humans and Wildlife Based on Modeling of
Fate and Transport of Mercury and Patterns of Fish Intake 5-2
5.2.2 Comparison of Dose-Response Estimates Across Species 5-6
5.2.3 Integration of Modeled Methylmercury Intake Through Consumption
of Fish for Hypothetical Humans and Wildlife with Dose-Response
Data 5-9
5.3 Potential Effects of Mercury Emission Sources on Local Fish Consumers 5-11
5.3.1 Humans 5-12
5.4 Comparison with Other Recommendations 5-14
5.4.1 Human Populations and Subpopulations 5-17
5.4.2 Wildlife Species 5-24
5.5 Risk Characterization Issues 5-26
6. CONCLUSIONS 6-1
7. RESEARCH NEEDS 7-1
8. REFERENCES 8-1
June 1996 . ii SAB REVIEW DRAFT
-------
LIST OF TABLES
Paae
2-1 Density-Based Dose Groupings 2-13
2-2 Uniform Dose Groupings 2-14
3-1 Summary of Bioaccumulation Factors for Trophic Levels 3 and 4 (mean, 5 Percent,
and 95 Percent values) 3-7
3-2 Exposure Parameters for Mink, Otter, Kingfisher, Osprey, and Eagle 3-8
4-1 Liver Mercury Concentration in Common Merganser, Red-Breasted Merganser and
Herring Gulls from Northern Quebec (Langlois and Langis, 1995) 4-15
4-2 Mercury and Methylmercury Concentrations in Tissues (ug per Gram Fresh Weight)
from the Common Loon in Northwestern Ontario (Barr, 1986) .' 4-15
4-3 Average Serving Size (gms) for Seafood from USDA Handbook #11 Used to
Calculate Fish Intake by FDA (1978) 4-18
4-4 Fish Species and Number of Persons Using the Species of Fish. Adapted from Rupp
et al., 1980 4-20
4-5 Fish Consumption from the NPD 1973/1974 Survey 4-21
4-6 Percent of Females By Age* Consuming Fish/Shellfish from SRI (1980) 4-2.1
4-7 Daily Average Per Capita Estimates of Uncooked Fish Consumption from CSFII 89/91 . 4-22
4-8 Daily Average Per Capita Estimates of Cooked Fish Consumption U.S. Population -
Finfish and Shellfish 4-22
4-9 Summary of Mercury Concentrations in Fish Species Micrograms Mercury per Gram
Fresh Weight (ug Hg/g) 4-25
4-10 Consumption of All Fish & Shellfish (gms/day) and Methylmercury per Kg body
weight from Fish among Respondents of the 1989-1991 CSFII Survey. Data for
"Users" Only. Bahnick et al. estimates for fresh-water fish Methylmercury
Concentrations 4-31
4-11 Consumption of All Fish & Shellfish (gms/day) and Methylmercury per Kg body
weight from Fish among Respondents of the 1989-1991 CSFII Survey. Data for
"Users" Only. Lowe et al. estimates for fresh-water fish Methylmercury
Concentrations . . -. 4-32
4-12 Consumption of Freshwater Fish (gms/day) and Methylmercury per Kg body weight ,
from Fish among Respondents of the 1989-1991 CSFII Survey. Data for "Users"
Only. Fish methylmercury concentrations based on Bahnick et al., (1994) 4-33
4-13 Consumption of Freshwater Fish (gms/day) and Methylmercury per Kg body weight
from Fish among Respondents of the 1989-1991 CSFII Survey. Data for "Users"
Only. Fish methylmercury concentrations based on Lowe et al., (1994) 4-34
4-14 Resident Population of the United States and Divisions, April 1, 1990 Census by
Gender and Age; in Thousands, including Armed Forces Residing in Region 4-36
4-15 Resident Population of the Contiguous United States, April 1, 1990 Census by Gender
and Age; in Thousands, including Armed Forces Residing in Region 4-37
4-16 Pregnancies by Outcome for Resident Females by Divisions and States, U.S. 1990, by
Age 4-38
4-17 Number of Pregnant Women Consuming Fish at Various Intake Levels Based on Data
from NPD.Inc. 1973/74 and 1990 U.S. Census Data 4-39
June 1996 iii SAB REVIEW DRAFT
-------
LIST OF TABLES (continued)
Page
4-18 Estimated United States Population Consuming Fish, Excluding Alaska and Hawaii
Estimates Based on the 1990 U.S. Census and the Continuing Surveys of Food Intake
by Individuals, 1989/1991 4-40
4-19 Fish Species by Percent of Consumption in CSFII 89/91, Mean Mercury Concentration
and Mercury Levels in 100 gram Servings of Fish 4-43
4-20 Percent of Dietary Intake by Species-Category of Fish/Shellfish for Persons Consuming
Fish or Shellfish (from CSFII 89/91) 4-44
4-21 Percent of Dietary Intake by Species-Category of Fish/Shellfish by Women of Child-
Bearing Age (from CSFII 89/91) ' 4-44
4-22 Percent of Dietary Intake by Species-Category of Fish/Shellfish by Children Ages 14
Years and Younger (from CSFII 89/91) 4-45
4-23 Total Fish Intake and Intake of Sports-fish by Licensed Wisconsin Anglers as Reported
by Fiore et al., 1989 4-46
4-24 EstimatedNumber of Pregnant Women Consuming Fish at Various Intake Levels Based
on Data from NPDJnc. 1973/74 and 1990 U.S. Census Data 4-46
5-1 Assumed Human Fish Consumption Rates and Body Weights Used in the Exposure
Modeling 5-4
5-2 Assumed Fish Consumption Rates and Body Weights of Piscivorous Birds and
Mammals Used in the Exposure Modeling 5-5
5-3 Assumed Fish Consumption Rates of Piscivorous Birds and Mammals Used in the
Exposure Modeling 5-5
5-4 Incidence of Effects in Iraqi Children by Exposure Group 5-7
5-5 Animal and Human Health Endpoints for Methylmercury in jag/kg bw/day 5-9
5-6 The Concentrations of Methylmercury in Trophic Level 3 and Trophic Level 4 Fish
Which, If Consumed at the Assumed Rates on a Daily Basis, Result in Exposure at the
RfD or the LOAEL 5-10
5-7 Predicted Methylmercury Concentrations in Fish in the Eastern U.S. combining 90th
percentile RELMAP Estimates and Local Source Estimates and the Resulting Human
Exposure Estimates 5-13
5-8 Hair Mercury Concentrations (ug Hg/gram hair or ppm) from Residents of Various
Communities in the United States 5-16
5-9 Association Between Hair Mercury and Frequency of Fish Ingestion 5-18
5-10 Estimated United States Population Consuming Fish, Excluding Alaska and Hawaii
Estimates Based on the 1990 U.S. Census and the Continuing Surveys of Food Intake
by Individuals, 1989/1991 5-19
5-11 Estimated Fish-Consuming Population in the United States, excluding Alaska and
Hawaii Estimates Based on the 1990 U.S. Census and the National Purchase Diary
Inc., 1973/74 Data on Fish/Shellfish Consumption 5-20
5-12 Comparison of Wildlife Criteria Calculated by Great Lakes Water Quality Initiative
and by the Mercury Study 5-25
June 1996 iv SAB REVIEW DRAFT
-------
LIST OF FIGURES
Page
2-1 Density of Data Points Relative to Hg concentration in Hair for Iraqi Cohort Data .... 2-13
4-1 Fate and Transport Models used and Exposure Routes Considered to Examine Exposure
Predictions Using Measured Environmental Concentrations 4-2
4-2 Fate, Transport and Exposure Modeling Conducted in the Long Rante Transport
Analysis 4-3
4-3 Fate, Transport and Exposure Modeling Conducted in the Local Impact Analysis 4-5
4-4 Fate, Transport and Exposure Modeling Conducted* in the Combined COMPDEP and
RELMAP Local Impact Analysis . 4-6
4-5 Distribution of Fish Consumption Rates of Various Populations . / 4-42
June 1996 v SAB REVIEW DRAFT
-------
LIST OF SYMBOLS, UNITS AND ACRONYMS
LOAEL Lowest-Observed-Adverse-Effect Level
NOAEL No-Observed-Adverse-Effect Level
RfD Reference Dose
June 1996 vi SAB REVIEW DRAFT
-------
U.S. EPA AUTHORS
Principal Authors:
Kathryn R. Mahaffey, Ph.D.
National Center for Environmental Assessment-
Cincinnati
Office of Research and Development
Cincinnati, OH
. Rita S. Schoeny, Ph.D.
National Center for Environmental Assessment-
Cincinnati
Office of Research and Development
Cincinnati , OH
Glenn E. Rice
National Center for Environmental Assessment-
Cincinnati
Office of Research and Development
Cincinnati, OH
Contributing Authors:
Robert B. Ambrose,-Jr., P.E.
Ecosystems Research Division
National Exposure Research Laboratory
Athens, GA
William G. Benjey, Ph.D.
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric
Administration
Research Triangle Park, NC
on assignment to the
U.S. EPA National Exposure Research Laboratory
a Deceased
O. Russell Bullock
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric
Administration
Research Triangle Park, NC
on assignment to the
U.S. EPA National Exposure Research Laboratory
Terry Clark, Ph.D.a
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric -
Administration
Research Triangle Park, NC
David H. Cleverly
National Center for Environmental Assessment
Office of Research and Development
Washington, DC
Stanley Durkee
Office of Research and Science Integration
Washington, DC .
Martha H. Keating
Office of Air Quality Planning and Standards
Research Triangle Park, NC
James D. Kilgroe, Ph.D.
National Environmental Research Laboratory
Office of Research and Development
Research Triangle Park, NC
John W. Nichols, Ph.D.
Mid-Continent Ecology Division
Office of Research and Development
Duluth, MN
Jeff Swartout
National Center for Environmental Assessment-.
Cincinnati »,
Office of Research and DevelopmSnf
Cincinnati, OH
June 1996
vn
•SAB REVIEW DRAFT
-------
SCIENTIFIC PEER REVIEWERS
Brian J. Alice, Ph.D.
Harza Northwest, Incorporated
Thomas D. Atkeson, Ph.D.
Florida Department of Environmental
Protection
Steven M. Bartell, Ph.D.
SENES Oak Ridge, Inc.
Mike Bolger, Ph.D.
U.S. Food and Drug Administration
James P. Butler, Ph.D.
University of Chicago
Argonne National Laboratory
Rick Canady, Ph.D.
Agency for Toxic Substances and Disease
Registry
Rufus Chancy, Ph.D.
U.S. Department of Agriculture
Tim Eder
Great Lakes Natural Resource Center
National Wildlife Federation for the
States of Michigan and Ohio
William F. Fitzgerald, Ph.D.
University of Connecticut
Avery Point
Robert Goyer, M.D.
National Institute of Environmental Health
Sciences
George Gray, Ph.D.
Harvard School of Public Health
Terry Haines, Ph.D.
National Biological Service
Joann L. Held
New Jersey Department of Environmental
Protection & Energy
Gerald J. Keeler, Ph.D.
University of Michigan
Ann Arbor
Leonard Levin, Ph.D.
Electric Power Research Institute
Malcom Meaburn, Ph.D.
National Oceanic and Atmospheric
Administration
U.S. Department of Commerce
Paul Mushak, Ph.D.
PB Associates
Jozef M. Pacyna, Ph.D.
Norwegian Institute for Air Research
Ruth Patterson, Ph.D.
Cancer Prevention Research Program
Fred Gutchinson Cancer Research Center
Donald Porcella, Ph.D.
Electric Power Research Institute
Charles Schmidt
U.S. Department of Energy
Pamela Shubat, Ph.D.
Minnesota Department of Health
Alan H. Stern, Dr.P.H.
New Jersey Department of Environmental
Protection & Energy
Edward B. Swain, Ph.D.
Minnesota Pollution Control Agency
M. Anthony Verily, M.D.
University of California ^
Los Angeles *•'—
June 1996
via
SAB REVIEW DRAFT
-------
EXECUTIVE SUMMARY
Section 112(n)(l)(B) of the Clean Air Act (CAA), as amended in 1990, requires the U.S.
Environmental Protection Agency (U.S. EPA) to submit a study on atmospheric mercury emissions to
Congress. The sources of emissions that must be studied include electric utility steam generating
units/ municipal waste combustion units and other sources, including area sources. Congress directed
that the Mercury Study evaluate many aspects of mercury emissions, including the rate and mass of
emissions, health and environmental effects, technologies to control such emissions and the costs of
such controls.
Volume VI presents the risk characterization for mercury emitted to the environment from
anthropogenic sourc.es. Risk characterization is the last step of the risk assessment process as
originally described by the National Academy of Sciences (NAS, 1983) and adopted by U.S. EPA
(U.S. EPA, 1984, 1992). This step evaluates assessments of human health and ecological effects,
identifies human subpopulations or ecological species exposed to mercury, assesses exposures from
multiple environmental media and describes the uncertainty and variability in these assessments. In
addition to the NAS (1983) source, guidance from the recent report Science and Judgment in Risk
Assessment (NAS/NRC 1994) and from the Policy far Risk Characterization at the U.S. Environmental
Protection Agency (issued in March, 1995, by the Administrator of U.S. EPA) were also followed.
This latter document reaffirmed the principles and guidance found in the Agency's 1992 policy
Guidance on Risk Characterization for Risk Managers and Risk Assessors.
Volume VI of this Report summarizes and integrates the exposure and effects information for
mercury presented in Volumes m, IV and V into an overall risk characterization for humans and
wildlife. First, technical characterizations of the human and wildlife health effects of mercury are
described, with accompanying discussion of uncertainty in the quantitative risk estimates. In Chapter
4 a technical characterization of the exposure of selected human and wildlife populations to mercury is
presented, again accompanied by a discussion of uncertainty. Also in Chapter 4 are estimates of the
size of the wildlife and human populations that are exposed to methylmercury. Literature reports on
mercury tissue levels in piscivorous wildlife species and the size of selected wildlife populations are
also presented. An overall characterization of the risk is presented in Chapter 5, taking two
approaches. In the first, RfD values and lowest adverse effect levels (LOAELs) for wildlife and
humans are used to estimate fish tissue concentrations of mercury that are below the risk level for
selected piscivorous wildlife species and hypothetical human populations. Chapter 5 concludes with a
comparison of recommendations from various groups with an interest hi mercury.
During episodes of methylmercury poisoning, both human subpopulations and wildlife species
have been affected by mercury poisoning. Clinical poisoning of humans from methylmercury occurred
in epidemics in Iraq (Bakir et al., 1973; Amin-Zaki et al., 1979) and Japan (Harada, 1968, 1977,
1995), and smaller outbreaks have occurred in additional populations. In the middle decades of this
century, consumption of grain treated with mercury fungicides produced severe and frequent poisoning
among wildlife species (Borg et al., 1979). Consequently, hi the risk characterization both human
subpopulations and wildlife species were considered.
As a chemical element mercury cannot be created or destroyed. The same amount has existed
on the planet since the earth was formed. Mercury, however, can cycle in the environment as a result
of both natural and anthropogenic activities. Both measured data and the results from global modeling
have led to the understanding that anthropogenic mercury emissions equal or exceed those from
June 1996 ES-1 SAB REVIEW DRAFT
-------
natural sources (such as volcanic activity, volatilization from the oceans, etc.). Human and natural
activity has the overall effect of making more mercury biologically available. Emissions of mercury
from human activity are thought to contribute from between 50 to 75 of the current total annual input
of mercury to the atmosphere. There are data and modeling results that indicate that the amount of
mercury mobilized and released into the biosphere has increased since the beginning of the industrial
age.
The Inventory of Anthropogenic Mercury Emissions in the United States (Volume n of this
Report) comprehensively examined mercury sources to the extent supported by available data. The
inventory included many manufacturing processes, a variety of combustion sources (including sewage
sludge burning and crematories) and miscellaneous sources (such as laboratory use, electrical lamp
breakage and dental amalgam preparation). The exposure assessment hi Volume III of this Report
focussed on those source categories having significant emissions in the aggregate, sources with the
potential for individual facilities to have a localized impact on the environment and sources for which
there were sufficient data to support an intelligent use of exposure models. The sources were electric
utility boilers, municipal waste incineration, medical waste incineration, chlor-alkali plants, primary
lead smelters and copper smelters.
In this Mercury Study Report to Congress, three species of mercury were considered:
elemental (Hg°), inorganic mercury or mercuric mercury (Hg2*) and methylmercury. Data in both
humans and experimental animals show that all three forms of mercury evaluated in this Report
(elemental, inorganic and methylmercury) can produce adverse health effects. Except for people
whose occupation brings them hi contact with elemental mercury, humans will be exposed primarily to
methylmercury; that exposure will be largely through consumption of fish.
Methylmercury can produce a variety of adverse effects, depending on the dose and time of
exposure. To present a comprehensive estimation of methylmercury effects hi humans, many different
health endpoints were evaluated by U.S. EPA using established Risk Assessment Guidelines.
Methylmercury has been shown to cause tumors hi mice at doses that produce severe non-cancer
toxicity. The data are limited but lead to the conclusion that low dose exposures to methylmercury are
not likely to cause cancer in humans. Based on data on effects related to mutation formation (changes
hi DNA), there is concern that methylmercury could increase frequencies of mutations in human eggs
and sperm. These data were not sufficient, however, to permit estimating the amount of
methylmercury that would cause a measurable mutagenic effect in a human population. Data in both
humans and animals are sufficient to judge methylmercury to be a human developmental toxicant; that
is, a material that would produce effects during the period of human development (from conception to
sexual maturity). The developmental deficits noted have been associated with nervous system damage,
or neurotoxicity. Neurotoxicity is also the effect of concern when adults are exposed to
methyhnercury, but developmental delays are the critical effect.
Data were sufficient for calculation of quantitative estimates for general systemic toxicity for
elemental mercury (reference concentration, or RfC, of S.OxlO"4 mg/m3), inorganic mercury (RfD of
3xlO'4 mg/kg-day) and methylmercury (RfD of IxKT4 mg/kg-day). These estimates seem to be very
close in magnitude. The endpoints for the methylmercury and elemental mercury estimates are
similar: neurologic deficits. It should be noted, however, that the endpoint for methylmercury was the
observation of developmental delays hi children exposed in utero and the endpoint for elemental
mercury was measurement of sensitive indicators of neurologic damage in adults. There may be route-
specific effects on dose response that have not been investigated. The endpoint for inorganic mercury
was measurement of changes leading to immune-mediated kidney damage in rats. There is evidence
June 1996 ES-2 SAB REVIEW DRAFT
-------
of kidney damage in mercury-exposed humans. The inorganic mercury RfD has a relatively large
uncertainty factor (1000 due to lack of a NOAEL, lack of a life-time study and extrapolation from
animal data to humans); thus, the RfD may not be strictly comparable to the methylmercury RfD in
terms of magnitude.
A previous RfD for methylmercury of 3xlO"4 mg/kg-day had been calculated by U.S. EPA
based on observation of paresthesia in adults who had consumed contaminated seed grain in Iraq in
the early 1970s. Both a quantitative uncertainty analysis and a consideration of reporting errors hi
adult paresthesia have led to the conclusion that this is not the most reliable endpoint for use in a
quantitative estimate of risk. Concern had been raised as to whether the RfD based on effects in
adults was protective of developmental effects. A new RfD based on application of a benchmark
approach to developmental neurotoxicity in children exposed in utero is within a numerical factor of
three of the older estimate based on observation in adults. A quantitative uncertainty analysis of this
RfD indicates that it is likely to be protective for all developmental endpoints.
The exposure assessment made use of computer-based models for long range transport of
mercury (RELMAP) and impact of mercury emissions near the point of emissions (COMPDEP and
IEM 2). Data on measured mercury levels in various environmental media were not sufficient for a
nationwide survey of mercury but were used for comparison with the modeled estimates. For
RELMAP results, measured data corroborate modeled estimates and geographic trends. Exposure
assessments were conducted for nine different hypothetical human receptors hi several settings (near a
lake, urban, etc.). The assessment of exposure pathways consequent to emissions of mercury from
anthropogenic sources indicates that the major exposure to both humans and wildlife is to
methylmercury in fish.
The broad ecosystem effects of mercury are not completely understood. No applicable studies
of the effects of mercury on intact ecosystems were found. Consequently, characterization of risk for
non-human species did not attempt to quantify effects of mercury on ecosystems, communities, or
species diversity. The characterization focused on (1) quantities of mercury that adversely affect the
health of sensitive subpopulations of wildlife species; and (2) the co-location of these populations with
areas of elevated mercury exposure secondary to ambient, anthropogenic emissions of methylmercury.
To this end wildlife criteria (WC) were calculated for three piscivorous (i.e., fish-eating) birds and two
mammals. The WC is a mercury level in water that is expected to be without harm for the species.
The WC considers the bioaccumulation of mercury in the large and small fish eaten by the mammals
or birds. WC calculation used bioaccumulation factors (BAF) to estimate mercury tissue level for
trophic level 3 and trophic level 4 fish, given a concentration of mercury in the water column. The
BAFs were derived by application of two (BAF3) or three (BAF^ methodologies and field data on
fish and water mercury concentrations; derivation of the BAFs and the quantitative uncertainty analysis
are described in Volume V. The effects data for mammals were from a short-term study of
neurotoxicity in mink. The data for piscivorous birds were from a three-generation study in mallard
ducks. The WC are these: mink, 415 pg mercury/L water, otter, 278 pg/L; kingfisher, 193 pg/L;
osprey, 483 pg/L; bald eagle 538 pg/L.
There is uncertainty and variability associated with each WC. These include lack of long-
term studies for mammals, lack of a no adverse effect level (NOAEL) for birds, and extrapolation
from one species to another. It is not known if the species selected for WC development are the most
sensitive or appropriate species, nor if protecting individual animals or species will guarantee
protection of their ecosystem from harmful effects of mercury. There are uncertainties and expected
variability hi the BAF; it was the subject of a quantitative uncertainty analysis.
June 1996 ES-3 . SAB REVIEW DRAFT
-------
Sizes of populations potentially at risk for methylmercury exposure were estimated for both
humans and wildlife species. These were compared to measured levels of mercury contamination.
Women of child-bearing age are one group of concern, because methylmercury is a developmental
toxicant. Even short-term exposures to methylmercury could adversely affect development because of
the sensitivity of the developmental process. Moreover methylmercury persists in tissues; dietary
intakes just prior to pregnancy may be of concern in addition to methylmercury intakes during
pregnancy. Another cause for concern is that using estimates on the number of pregnant women in the
age group 15 through 44 years, 9.5% of women are pregnant in any given year; thus the size of the
impacted population is not negligible. The number of women of child-bearing age were determined in
the 1990 U.S. Census. This census estimated that the total female population ages 15 through 44
years was 58,222,000 hi the 48 contiguous states.
Data on fish consumption for a general population of women hi the United States were
developed from the United States Department of Agriculture's Continuing Surveys of Individual Food
Consumption *for the period 1989-1991 (CSFII 89/91). Cross-sectional data on food consumption
collected over a three year period were used to estimate longer-term dietary patterns. CSFII 89/91
reported that 30.5% of women ages 15 through 44 years consume fish at least once hi a 3-day period.
This does not mean that the other 69.1% of women avoid fish consistently; rather that fish did not
appear as a dietary item during the three days during which the food diaries were kept. There are
17,371,000 women who are fish consumers. If the 95th percentile is determined to be "high end" and
as 9.5% of the female population from ages 15 through 44 years are pregnant hi a given year, the
number of pregnant, "high end" fish consumers in the contiguous United States number about 84,300.
Fish consumption, measured mercury hi fish and the human RfD and LOAEL are compared graphical-
ly hi Chapter 4. Comparisons are made for the general U.S. population, women of child-bearing age
and children 15 years and younger.
Additional data on fish consumption from a longitudinal food survey were also analyzed. The
National Purchase Diary, Inc. surveyed families in 1973 and 1974 and found that 94% of persons
reported consuming fish at least once during a month long period. The top one percent of consumers
ingested fish and shellfish at levels over 100 grams per day. Using these data on fish consumption the
number of maternal-fetal pairs at risk hi any given year was estimated to be hi excess of 50,000
women and developing infants.
A quantitative assessment of risk methylmercury exposure from contaminated fish has been
performed for three hypothetical humans receptors and five wildlife species. Estimated LOAELs and
RfDs were combined with modeled fish mercury levels and amounts of fish consumed; the result was
a level of mercury in fish that if consumed on a daily basis would result in exposure to the RfD or
LOAEL. These numbers are presented for interspecies comparison; if health endpoints are considered
equivalent, then the kingfisher is the most impacted by mercury contamination hi fish. There are a
number of uncertainties hi this analysis, including the lack of comparability in terms of sensitivity (or,
conversely, adversity) of the endpoints used hi the effects assessments.
June 1996 ES-4 SAB REVIEW DRAFT
-------
Conclusions
The following conclusions are presented in approximate order of degree of certainty in the
conclusion, based on the quality of the underlying database. The conclusions progress from
those with greater certainty to those with lesser certainty.
• There is a plausible link between methylmercury concentrations in freshwater fish and
anthropogenic mercury emissions. The degree to which this linkage occurs cannot be
estimated quantitatively at this time.
• Among humans and wildlife that consume fish, methylmercury is the predominant
chemical species contributing to mercury exposure.
• Methylmercury is known to cause neurotoxic effects in humans via the food chain.
• The human RfD for methylmercury is calculatedto be IxlO"4 mg/kg body weight/day.
While there is uncertainty in this value, there are data and quantitative analyses of
health endpoints that corroborate and support a reference dose within a range of an
order of magnitude. A quantitative uncertainty analysis indicates that the human RfD
based on observation of developmental neurotoxicity in children exposed to
methylmercury in utero is likely to be protective of human health.
• The RfD is a confident estimate (within a faedfr^of 10) of a level of exposure without
adverse effects on those human health endpoints measured in the Iraqi population
exposed to methylmercury from grain. These included a variety of developmental
neurotoxic signs and symptoms. The human RfD is for ingested methylmercury; no
distinction was made regarding the food or other media serving as the ingestion
vehicle.
• U.S. EPA calculates that members of the U.S. population ingest methylmercury
through the consumption of fish at quantities of about 10 times the human reference
dose. This amount of methylmercury is equivalent to the benchmark dose used in the
calculation of the reference dose; the benchmark dose was taken to be an amount
equivalent to the NOAEL. The NOAEL was an ingested amount of 1.1 ug per kg
body weight per day. Consumption of mercury equivalent to the NOAEL is predicted
to be without harm for the majority of a population. Individual risks cannot be
determined from the available data.
• Prediction of risk cannot be made for ingestion of methylmercury above the
benchmark dose given the currently available data in humans.
• Concentrations of mercury in the tissues of wildlife species have been reported at
levels associated with adverse health effects in laboratory studies in the same species.
• Dietary survey data based on short-term, cross-sectional sampling periods indicate that
approximately 30 percent of the general U.S. population consumes fish at least once
during a three-day period. Among this group of fish consumers, roughly 50 percent
are predicted to consume methylmercury at the RfD. Consuming methylmercury at
levels equal to the RfD is equated to be without harm.
June 1996 ES-5 SAB REVIEW DRAFT
-------
• Based on longer-term data that recorded fish consumption for one-month periods,
approximately 94% of the population consumed fish at least once during that period.
• Using both^the longitudinal and cross-sectional survey data, it is estimated that 1 to 2
percent of women of child-bearing age consistently consume fish and shellfish at
intakes of 100 grams per day or greater. Whether or not methylmercury intakes are
elevated above the estimated NOAEL depends on the concentration of methylmercury
in the fish and shellfish consumed.
• U.S. EPA estimates that approximately one-third of fish and shellfish consumed are
from freshwater/estuarine habitats that may be affected by local sources of mercury.
• Case reports in the literature document that sick and/or dying animals and birds with
seriously elevated tissue mercury concentrations have been found hi the wild. These
wildlife have mercury concentrations elevated to a level documented in laboratory
studies to produce adverse effects in these species. For a specific case report
concurrent exposure to other sources of ill health cannot- be excluded.
• Modeled estimates of mercury concentration in fish around hypothetical mercury
emissions sources predict exposures at the wildlife WC. The wildlife WC, like the
human RfD, is predicted to be a safe dose over a lifetime. It should be noted,
however, that the wildlife effects used as the basis for the WC are gross clinical
manifestations or death. Expression of subtle adverse effects at these doses cannot be
excluded.
• Data are not sufficient for calculation of separate reference doses for children, in utero
exposure and the aged.
• Comparisons of dose-response and exposure estimates through the consumption of fish
indicate that certain species of piscivorous wildlife are more exposed on a per
kilogram body weight basis than are humans. The implications for wildlife health are
uncertain.
•
There are many uncertainties associated with this analysis. The sources of uncertainty include
the following:
• There is considerable uncertainty and apparent variability in the movement of mercury
from the abiotic elements of the aquatic system through the aquatic food chain.
• U.S. EPA has developed a BAF hi an attempt to quantify the relationship between
dissolved mercury concentrations in the water column and methylmercury
concentrations hi fish. This BAF was developed using a four-tier food chain model
and extant field data. A quantitative uncertainty analysis of the BAF and the
variability of the BAF was examined.
• There is considerable uncertainty in atmospheric processes that affect emitted mercury.
U.S. EPA has attempted to predict the fate and transport of mercury through the use of
atmospheric models. The results of these models are uncertain. For the regional
June 1996 ES-6 SAB REVIEW DRAFT
-------
1. INTRODUCTION
Section 112(n)(l)(B) of the Clean Air Act (CAA), as amended in 1990, requires the U.S.
Environmental Protection Agency (U.S. EPA) to submit a study on atmospheric mercury emissions to
Congress. The sources of emissions that must be studied include electric utility steam generating
units, municipal waste combustion units and other sources, including area sources. Congress directed
that the Mercury Study evaluate many aspects of mercury emissions, including the rate and mass of
emissions, health and environmental effects, technologies to control such emissions and the costs of
such controls.
In response to this mandate, U.S. EPA has prepared a seven-volume Mercury Study: Report to
Congress. The seven volumes are as follows:
I. Findings and Recommended Actions
II. An Inventory of Anthropogenic Mercury Emissions in ttie United States
III. An Assessment of Exposure from Anthropogenic Mercury Emissions in the United
States
IV. Health Effects of Mercury and Mercury Compounds
V. An Ecological Assessment for Anthropogenic Mercury Emissions in the United States
VI. Characterization of Human Health and Wildlife Risks from Anthropogenic Mercury.
Emissions in the United States
VII. An Evaluation of Mercury Control Technologies and Costs
Risk characterization is the last step of the risk assessment process as originally described by
the National Academy of Sciences (NAS, 1983) and adopted by U.S. EPA (U.S. EPA, 1984, 1992).
This step evaluates assessments of human health and ecological effects, identifies human
subpopulations or wildlife species at elevated risk from mercury, assesses exposures from multiple
environmental media, and describes the uncertainty and variability in these assessments.
In March, 1995, the Administrator of U.S. EPA issued the Policy for Risk Characterization at
the U.S. Environmental Protection Agency reaffirming the principles and guidance found in the
Agency's 1992 policy Guidance on Risk Characterization for Risk Managers and Risk Assessors. The
purpose of this policy statement was to ensure that critical information from each stage of a risk
assessment be presented in a manner that provides for greater clarity, transparency, reasonableness, and
consistency in risk assessments. Most of the 1995 Policy for Risk Characterization at the U.S. EPA
was directed toward assessment of human health consequences of exposures to an agent. This
guidance refers to an ongoing parallel effort by the Risk Assessment Forum to develop U.S. EPA
ecological risk assessment guidelines that will include guidance specific to ecological risk
characterization. The 1995 Policy for Risk Characterization at the U.S. EPA makes reference to the
use of data from wildlife species in assessing the consequences of exposure to an agent through
environmental media.
Key aspects of risk characterization identified in the 1995 Policy for Risk Characterization at
the U.S. EPA include these: bridging risk assessment and risk management, discussing confidence and
uncertainties and presenting several types of risk information. Risk characterization is the
summarizing step of the risk assessment process. In this volume of the Report, information from the
three preceding components of risk assessment are summarized, and an overall conclusion about risk is
synthesized that is complete, informative, and useful for decision-makers. One aim of the process is to
highlight clearly both the confidence and the uncertainty associated with the risk assessment. The risk
June 1996 1-1 SAB REVIEW DRAFT
-------
characterization conveys the assessor's judgment regarding the nature and existence (or lack of) human
health or ecological risks that accompany exposures to an agent.
Integration of multiple elements of risk assessment for both human health or ecological
impacts is a complex process that is intrinsically nonsequential. Assessment of the likelihood of
hazard depends on the magnitude of exposure to human or wildlife species, which requires an
understanding of dose-response relationships. For an element such as mercury, which can exist in
multiple valence states and numerous chemical compounds, risk characterization requires a broad-
based, holistic approach to the risk assessment process. This holistic approach encompassing human
health and ecological hazard assessments, as well as analysis of exposures, has been described in
greater detail (Harvey et al., 1995).
In mis Report, three species of mercury are considered: elemental (Hg°), inorganic or
mercuric mercury (Hg2*), and methylmercury. The assessment of exposure pathways consequent to
emissions of mercury from anthropogenic sources indicates that the major exposure to both humans
and wildlife is to organic mercury Qargely methylmercury) in fish. A quantitative assessment of risk
of mercury exposure to both humans and wildlife has been determined for three subpopulations of
humans and for representative piscivorous avian and mammalian wildlife species. Assessments were
made of all three forms of mercury for potential human health effects; because exposure to humans is
likely to be as ingested methylmercury, that form is emphasized in this volume. Estimated Lowest
Observed Adverse Effects Levels (LOAELs) and No Observed Adverse Effect Levels (NOAELs) and
water criteria for wildlife were limited to methylmercury. These assessments were drawn from
exposure modeling and doses of mercury associated with adverse health effects.
June 1996 1-2 SAB REVIEW DRAFT
-------
2. HUMAN HEALTH EFFECTS: HAZARD IDENTIFICATION AND DOSE-RESPONSE
2.1 Health Hazards Associated with Mercury Exposure
The three forms of mercury considered in this Report (mercury vapor, divalent inorganic
mercury, and methylmercury) are characterized by somewhat different health endpoints for human
health risk assessment All three chemical species of mercury have been associated with adverse
human health effects, and human and animal data on all three forms of mercury indicate that systemic
toxic effects (rather than cancer or germ cell mutagenicity) are most likely to occur in humans as a
consequence of environmental exposures. Available information on health endpoints relevant to
human health risk assessment is described in Volume IV. A brief characterization of endpoints other
than systemic toxicity is given in Chapter 2 of Volume IV.
Data are insufficient to support comparisons of innate toxicity among the three forms of
mercury. Human data adequate for quantitative dose-response assessment have not been reported for
inorganic, divalent mercury. The RfD for inorganic mercury is within a factor of 3 of the RfD for
methylmercury; the RfD for inorganic mercury, however, includes a large uncertainty factor (1,000).
Furthermore, the extent to which the endpoints for inorganic and methylmercury are comparable
(based on either the severity or sensitivity) is unknown. The RfD for methylmercury and the RfC for
inhaled elemental mercury were both based on observation of neurotoxicity (from exposure hi adults
for elemental mercury and from exposure in utero for methylmercury). The two quantitative risk
estimates are an RfD of IxlO"4 mg/kg-day for methylmercury and an RfC of 3xlO"4 mg/m3 for
elemental mercury. In order to compare the toxic potency implied by these values, some conversion
to internal dose appropriate to the route of exposure would be necessary. This has not been done for
this Report
Assessment of health end-points, dose-response and exposure suggests that methylmercury is
the chemical species of major concern. Methylmercury is the chemical species of greatest concern
because of the fate and transport of mercury to water bodies and sediments with subsequent
bioaccumulation of methylmercury in the aquatic food-web. In short the exposure assessment in this
Report (as well-as other exposure assessments) indicates that most human exposure is likely to be due
to methylmercury in food, primarily fish. Fish-eating wildlife will also be exposed in the main to "
methylmercury.
Adverse effects on the nervous system and reproduction are the predominant effects of
methylmercury exposure on humans and several wildlife species. In multiple species, the neurological
effects of methylmercury exposure are mainly on the motor and sensory systems, especially in the
areas of sensory-motor integration. The type of information available differs markedly across species
resulting hi gross disparity in the severity of the hazard. For example, marked incoordination hi gait
(ataxia) is the most sensitive endpoint identified in previous research on methylmercury toxicity in
mink. By contrast human subjects can identify altered sensory perception (such as paresthesia), a
much more subtle indicator of neurological effect Nonetheless, me consistent pattern observed across
human and wildlife species is adverse effects of methylmercury on sensory-motor function.
Human epidemics of methylmercury poisoning have occurred in this century. During the
1950s and 1960s in Japan, major epidemics of fatal and nonfatal neurological disease were caused by
methylmercury exposure from consumption of seafood in Minamata and fresh-water fish in Niigata
(Tsubaki and Irukajama, 1977). Additional epidemics of methylmercury poisoning from consumption
of methylmercury on grain occurred in Iraq in the 1960s and 1970s (Jalili and Abbasi, 1961;
June 1996 2-1 SAB REVIEW DRAFT
-------
Kantarjina, 1961; Bakir et al., 1973). These epidemics have provided the strongest possible evidence
linking exposure to methylmercury with human fatalities and neurological disease. The fundamental
question for risk characterization is not whether methylmercury from fish can produce neurological
disease, but rather what quantities of methylmercury in fish and what duration of this exposure
produce neurological disease in humans.
Exposure to high doses of methylmercury in utero has produced neurological sequelae.
Developmental effects in humans consequent to methylmercury exposure have been reported for
offspring of women who consumed contaminated seed-grain in Iraq (Amin-Zaki et al., 1976; Marsh et
al., 1981, 1987) and infants born to mothers who ate contaminated fish from Minamata Bay in Japan
(Harada, 1978). An inverse correlation was observed between IQ in children in New Zealand and
maternal hair mercury level {Kjellstrom et al., 1989). Maternal hair mercury level has been correlated
with abnormal muscle tone in Cree Indian male children (McKeown-Eyssen et al., 1983). These
multiple episodes of disease among numerous groups of people widely separated geographically
provide the basis for high confidence in the association of methylmercury exposure and adverse
developmental deficits of the nervous system. Developmental effects have been reported in three
strains of rat and two strains of mice and in guinea pigs, hamsters, and monkeys. While some studies
are limited in their usefulness to assessment of developmental risk, the database taken as a whole
supports a judgment of Sufficient Human and Animal Data for developmental toxicity of
methylmercury, in the language of the Risk Assessment Guidelines. The RfD of IxlO"4 mg/kg-day
was derived using an estimate of threshold (bench mark) for the Iraqi neurodevelopmental
observations.
The neurological scores used in developing the benchmark dose for effects in children were
based on clinical evaluation for cranial nerve signs, speech, involuntary movement, limb tone-strength,
posture, and the ability to sit, stand and run. A limitation on these data is that the Iraqi mothers did
not know with accuracy the ages of their infants; cultural mores did not dictate use of Western
calendars for recording of family events. Consequently, reliability of data on which these endpoints
are based is compromised. A resulting uncertainty in the Iraqi data (because of the comparatively
short-term exposures) is classification bias secondary to whether or not methylmercury exposure
occurred during a particular gestational period.
Development of a quantitative estimate of human non-cancer risk for methylmercury has
proved to be a complex undertaking. Difficulty arises from attempts to quantify daily doses of human
exposure. The conventional approach for methylmercury is to use hair concentrations and back-
calculate to blood concentrations and then to a daily intake level. (Methods and assumptions for this
calculation are found in Volume IV, Section 5.3.1.1.) There is variation in the hair-to-blood ratios and
other physiological parameters, such as biologic half-lives.
At the present time, there is limited agreement in the scientific community concerning the
optimal neurological endpoints to use for assessment of mercury toxicity. It is generally agreed that
methylmercury exposure adversely affects cellular processes in broad areas of the nervous system.
Sensory and motor functions appear to be particular adversely affected. A wide range of endpoints
have been used to assess nervous system function in studies of mercury toxicity. Individual scores on
developmental tests were used for the New Zealand study (Kjellstrom, 1989); however, these data are
limited because of cultural differences between the subjects and the populations on which the tests
were standardized. Because of the different cultural practices, the neurological deficits of delayed
onset of walking and talking among children exposed prenatally in the Iraqi population may not be
appropriate measures for risk estimates for Western cultures. Extensive data from laboratory studies
June 1996 2-2 SAB REVIEW DRAFT
-------
with research animals are available. These data clearly support neurological changes as the critical
adverse effect for methylmercury.
A number of additional studies evaluating the association between neurological endpoints and
exposure to methylmercury from fish are underway in the mid-1990s. These ongoing studies evaluate
far more subtle endpoints of neurotoxicity than were assessed in the epidemics in Minamata and
Niigata. These studies also use far more sophisticated neurobehavioral and neuromotor assessments
than were feasible under conditions of the Iraqi studies. Neurobehavioral and neuromotor development
assessments are being carried out on more than 1,600 maternal-infant pairs from fish-consuming
populations in the Seychelles Islands and the Faroe Islands. These studies differ from the epidemics
that occurred hi Iraq, hi that exposures to methylmercury have extended for many years. Steady-state
conditions were clearly established before testing for the adverse effects was performed. In addition,
the Agency for Toxic Substances and Disease Registry of the United States Public Health Service is
sponsoring a group of studies conducted hi the United States that assess neurological end-points
among infants of mothers consuming substantial quantities of fish. An example of these studies is the
neuromotor/neurobehavioral evaluations of infants of high-fish-consuming mothers located in the
vicinity of Oswego, New York and monitored by the Department of Psychology of the State
University of New York. As results from these investigations become available, some of the issues of
variability and uncertainty hi understanding the threshold for adverse neuro-developmental effects of
methylmercury may be clarified. In particular, this evaluation should contribute greatly to an
assessment of the relationship between dose and response hi which fish is the vehicle of exposure to
methylmercury.
2.2 Dose-Response to Methylmercury
2.2.1 Calculation of Methvlmercurv RfD
U.S. EPA has on two occasions published RfDs for methylmercury which have represented the
Agency consensus for that time. These are described hi the sections below. At the time of the
generation of the Mercury Study Report to Congress, it became apparent that considerable new data on
the health effect of methylmercury hi humans were emerging. Among these are large studies of fish
or fish and marine mammal consuming populations hi the Seychelles and Faroes Islands. Smaller
scale studies are hi progress which describe effects hi population's around the U.S. Great Lakes. In
addition, there are new evaluations of published work described m section 3.3.1.1 of Volume IV,
including novel statistical approaches and application of physiologically based pharmacokinetic
models.
As the majority of these new data are either not yet published or have not yet been subject to
rigorous review, it was decided that it was premature for U.S. EPA to make a change in the
methylmercury RfD at this time. An inter agency process, with external involvement, will be
undertaken for the purpose of review of these new data evaluations and evaluations of existing data.
An outcome of this process will be assessment by U.S.EPA of its RfD for methylmercury to determine
if change is warranted.
Human and animal data on elemental, inorganic and methylmercury indicate that systemic
toxic effects (rather than carcinogenicity or germ cell mutagenicity) are most likely to be observed in
humans as a consequence of environmental exposures. The exposure assessment for environmental
mercury from anthropogenic sources appears in Volume III and is summarized in Chapter 3 of Volume
VI. This assessment points to the necessity of considering ingestion of inorganic mercury in water and
June 1996 2-3 SAB REVIEW DRAFT
-------
in food as a component of any site-specific or scenario-specific risk assessment. The modeled
exposure assessment indicates, however, that for the majority of people in the United States,
methylmercury exposure via contaminated fish is the major pathway. It is clear that in the segments
of the population that consume fish or seafood, the majority of mercury exposure will be to
methylmercury. Because methylmercury is the form to which humans are most exposed, the
remainder of the risk characterization will deal with only that form of mercury.
2.2.1.1 Neurotoxicity of Methylmercury
Neurotoxicity of methylmercury has been determined as the critical effect for the RfD; that is,
the adverse effect that is expected to occur at the lowest level of exposure. The RfD was based on
statistical analysis of data from human subjects in Iraq in the 1970s. For a period of approximately
three months this population consumed bread made from seed-grain treated with methylmercury
fungicide. In 1985 an RfD was determined to be SxlO"4 mg/kg-day, based on observation of
paresthesia in adults (Amin-Zaki et al., 1981). The LOAEL was determined to be 3xlO~3 mg/kg-day
(corresponding to 200 ug/L blood concentration), and an uncertainty factor of 10 was applied for use
of a LOAEL hi the absence of a NOAEL. A further uncertainty factor of 10 for sensitive individuals
for chronic exposure was not deemed necessary at the time, because the adverse effects were seen in
what was regarded as a sensitive group of individuals.
Since 1985, there have been questions raised as to the validity of this RfD and, specifically,
whether or not this RfD is applicable to developmental effects. This resulted hi the re-opening of
discussion of the methylmercury RfD by the U.S. EPA RfD/RfC Work Group hi 1992 and 1994.
Consensus on a new RfD was reached hi January of 1995. A detailed description of the derivation of
the RfD can be found in Section 6.3.1.1 of Volume IV, and summary information appears on IRIS.
A study of Iraqi populations by Marsh et al. (1987) was chosen as the most appropriate study
for determination of an RfD protective of a putative sensitive subpopulation, namely infants born to
mothers exposed to methylmercury during gestation. This report described neurologic abnormalities
observed in progeny of women who consumed bread prepared from methylmercury-treated seed grain
while pregnant. Among the signs noted in the infants exposed during fetal development were cerebral
palsy, altered muscle tone and deep tendon reflexes, as well as delayed developmental milestones (i.e.,
walking by 18 months and talking by 24 months). The data collected by Marsh et al. (1987)
summarize clinical neurologic signs of 81 mother and child pairs. From x-ray fluorescent
spectrometric analysis of selected regions of maternal scalp hair, concentrations ranging from 1 to 674
parts per million (ppm) mercury were determined, then correlated with clinical signs observed hi the
affected members of the mother-child pairs. Among the exposed population there were affected and
unaffected individuals throughout the exposure range.
2.2.1.2 Estimation of Mercury Ingestion
In order to quantify an average daily ingestion rate for the mothers, hair concentrations were
determined for periods during gestation when actual methylmercury exposure had occurred. A ratio of
250:1 (jug mercury/mg in hair:ug mercury/L of blood) was used to derive the RfD critical dose. A
complete discussion for the choice of this ratio is provided in Volume IV, Section 6.3.1.1. Conversion
of the hair mercury level to a blood mercury level was done according the following equation:
11 mg/kg hair / 250 = 44 ug/L blood
June 1996 2-4 SAB REVIEW DRAFT
-------
To obtain a daily dietary intake value of methylmercury corresponding to a specific blood
concentration, factors of absorption rate, elimination rate constant, total blood volume and percentage
of total mercury that is present in circulating blood must be taken into account. Calculation was by
use of the following equation based on the assumptions that steady state conditions exist and that first-
order kinetics for mercury are being followed.
d = C x b' * V
A xf
where:
d = daily dietary intake (ug of methylmercury/day)
C = concentration in blood (44 ug/L)
b = elimination constant (0.014 days"1)
V = volume of blood in the body (5 liters)
A = absorption factor (expressed as a unitless decimal fraction of 0.95)
f = fraction of daily intake taken up by blood (unitless, 0.05)
The rationales for use of specific values for equation parameters are in Volume IV, Section 6.3.1.1.
Solving for d provides the daily dietary intake of mercury that results in a blood mercury
concentration of 44 ug/L. To estimate a daily dose (ug/kg-day) the assumed body weight (bw) of 60
kg is included in the equation denominator. While the critical endpoint for the RfD is developmental
effects in offspring, the critical dose is calculated using parameters specific to the mothers who
ingested the mercury-contaminated grain. Data on body weights of the subjects were not available. A
default value of 60 kg (rounded from 58) for an adult female was used.
J CxbxV
A xfx bw
'1
44 fjg/L x 0.014 days' x 5L
0.95 x 0.05 x 60 kg
d - 1.1
Thus 1.1 ug/kg-day is the total daily quantity of methylmercury that is ingested by a 60 kg individual
to maintain a blood concentration of 44 ug/L or a hair mercury concentration of 11 ppm, the
benchmark dose derived below.
2.2.1.3 Grouping of the Response Data
Data on neurotoxic effects in children exposed to methylmercury in utero were used to
determine a benchmark dose used in the calculation of the RfD. Data used in the benchmark dose
calculation were excerpted from the publication Seafood Safety (NRC/NAS, 1991). Because the tables
of incidence of various clinical effects hi children that were provided in this document readily lent
June 1996' 2-5 SAB REVIEW DRAFT
-------
themselves to the benchmark dose modeling approach. The continuous data for the Iraqi population
that were reported by Marsh et al. (1987) were placed in five dose groups, and incidence rates were
provided for delayed onset of walking, delayed onset of talking, mental symptoms, seizures,
neurological scores above 3, and neurological scores above 4 for affected children. Neurologic scores
were determined by clinical evaluation for cranial nerve signs, speech, involuntary movement, limb
tone strength, deep tendon reflexes, plantar responses, coordination, dexterity, primitive reflexes,
sensation, posture, and ability to sit, stand and run. The effects of late walking, late talking, and
neurologic scores greater than 3 were also combined for calculation of a benchmark on all effects in
children. Alternative dose groupings are described in section 2.2.2.6.
2.2.1.4 Derivation of a Benchmark Dose
Benchmark dose estimates were made by calculating the 95 percent lower confidence limits on
doses corresponding to the 1 percent, 5 percent and 10 percent extra risk levels using a quanta!
Weibull model (K.S. Crump Division of Clement International). The Weibull model was chosen for
the benchmark dose calculations for the methylmercury data as recent research suggests it may be the
best model for developmental toxicity data (Faustman et al., 1994). The form of the quantal Weibull
that was used is:
- AO
where d is dose, AO is the background rate, Al is the slope, and A2 is a shape parameter. For each
endpoint and for the combined endpoints, the incidence of response was regressed on the dose. A
Chi-squared test of goodness-of-fit was used to test the null hypothesis (H^ that the predicted
incidence was equal to the observed incidence, so that H0 would be rejected for p-values less than
0.05.
•
2.2.1.5 Adjustments for Background Incidence
As an adjustment for background rates of effects, the benchmark dose estimates for
methylmercury were calculated to estimate the dose associated with "extra risk." Another choice
would have been to calculate based on "additional risk." Additional risk (AR) is defined as the added
incidence of observing an effect above the background rate relative to the entire population of interest:
AR = [P(d)-P(0)]/l. In the additional risk calculation, the background rate is subtracted, but still
applied to the entire population, including those exhibiting the background effect. Thus, background
effects are in a sense "double counted". Extra risk (ER) is always mathematically greater than or
equal to additional risk, and is thus a more conservative measure of risk whenever the background rate
is not equal to zero. Conceptually, extra risk is the added incidence of observing an effect above the
background rate relative to the proportion of the population of interest that is not expected to exhibit
such an effect Extra risk is more easily interpreted than additional risk, because it applies the
additional risk only to the proportion of the population that is not represented by the background rate.
Extra risk has been traditionally used in U.S. EPA's cancer risk assessments (Anderson et al., 1983)
and is discussed in detail hi a report on the benchmark dose by U.S. EPA's Risk Assessment Forum
(U.S. EPA, 1995).
June 1996 2-6 SAB REVIEW DRAFT
-------
The RfD/RfC Work Group chose the benchmark (95% lower bound on the dose for 10 percent
effect level) based on modeling of all effects in children. Recent research (Allen et ah, 1994a, b)
suggests that the 10 percent level for the benchmark dose roughly correlates with a NOAEL for
developmental toxicity data. Note that this conclusion was based on controlled animal studies and on
calculation of additional risk. Both the polynomial and Weibull models place a lower 95 percent
confidence limit on the dose corresponding to a 10 percent risk level at 11 ppm hair concentration for
methylmercury. The benchmark dose rounded to 11 ppm was used in the calculation of the RfD.
2.2.1.6 Calculation of the Methylmercury RfD
A composite uncertainty factor of 10 was used. This uncertainty factor was applied for
variability in the human population, in particular the wide variation in biological half-life of
methylmercury and the variation that occurs in the hair to blood ratio for mercury. In addition, the
factor accounts for lack of a two-generation reproductive study and lack of data for possible chronic
manifestations of the adult paresthesia that was observed during gestation. The default value of one
was used for the modifying factor.
The RfD for methylmercury was calculated using the following equation:
Benchmark Dose
BfD
UF xMF
1.1 ug/kg-day
10
1 x 10~ mg/kg-day
where
UF is the uncertainty factor and MF is the default of 1.
Confidence in the supporting database and confidence in the RfD were considered medium by the U.S.
EPA RfD/RfC Work Group.
2.2.2 Human Dose-Response Issues
The RfD is characterized by variability and uncertainty; an assessment of which is presented in
Appendix D to Volume IV, and in Section 2.3 of Volume VI. Fetal effects of methylmercury
exposure were based on hair mercury analyses of 83 women hi Iraq. The dose-response data derived
from this data set are a best estimate from a relatively small number of human subjects. The size of
the data set becomes a limitation for identifying adverse effects that may occur in a small fraction of
subjects due to factors such as individual variability. The duration of the exposure to methylmercury
(approximately three months in the Iraqi outbreak) was long enough to identify the effects of
methylmercury exposure on the fetus.
June 1996 2-7 SAB REVIEW DRAFT
-------
2.2.2.1 Sensitivity of Human Subpopulations
Neurotoxicity of methylmercury to the developing nervous system is well documented among
several populations of human subjects. Dose-response data have been most extensively analyzed for
the Iraqi population identified in the 1970s epidemic. Additional analyses of methylmercury poisoning
data have been published in 1995. Kinfto et al (1995) estimated threshold doses for adults following
consumption of methylmercury from fisn in Niigata, Japan, and Harada (1995) published an extensive
review of the epidemiology of Minamata disease.
An important issue is the extent to which results from the Iraqi and Japanese populations can
be generalized to other human populations. The task of identifying the nature and extent of exposures
that represent thresholds of dose-response to methylmercury is more complex. Do the Japanese and
Iraqi populations represent particularly sensitive subpopulations among the general population of
human subjects who can respond to methylmercury exposure with developmental neurotoxicity? Or
are there unique characteristics of these populations and patterns of methylmercury exposure that
resulted in them being unusually susceptible to the adverse effects of methylmercury exposure?
It is useful to clarify that there can be at least three broad areas that can render a population
particularly sensitive to methylmercury: responsiveness of the organism to the adverse effect,
differences in dose-response curves, and differences in exposure to the agent
The first basis for sensitivity is that the subpopulation of concern is physiologically susceptible
to the effect. The neurological effect in adults that occurs at the lowest dose is sensory disturbance or
paresthesia. These changes have been reported in both male and female adults regardless of age
(Tsubaki and Irukayama, 1977; Harada, 1995). By contrast methylmercury toxicity that occurs
following fetal exposure to methylmercury is secondary to maternal consumption of fish or grain
products contaminated with methylmercury. For this effect the sensitive subpopulation is the maternal-
fetal pair. Because an estimated 9.5 percent of women of reproductive age in the United States is
pregnant hi a given year, and because the half-life of methylmercury is estimated to range from 35 to
more than 189 days, all women of reproductive capacity can be considered as a sensitive
subpopulation for the developmental effects of methylmercury. A major uncertainty hi identification
of dose-response in susceptible populations is the lack of data to generate separate RfDs for in utero,
childhood and adult exposures.
The second basis for sensitivity is differences hi dose-response to methylmercury. For
example, individual differences exist in the biological half-life of mercury in the body. Persons with
longer body retention of mercury can be anticipated to be more sensitive to the adverse effects of
methylmercury if all other factors are equivalent. It has been reported by Kershaw et al. (1980) and
Sherlock et al. (1984) that the half-lives for methylmercury in blood were 52 (39 to 67) and 50 (42 to
70) days, respectively. Generally, the average biological half-life for methylmercury in humans is
considered to be approximately 70 days (Harada, 1995). However, reported individual values of
biological half-lives range from 33 to 270 days (Birke et al., 1972). The data from the study of Iraqi
methylmercury poisonings indicated a bimodal distribution of biological half-lives; one group
accounting for 89 of the samples had a mean value of 65 days, and the remaining group had a mean
value of 119 days (Al-Shahristani and Shinab, 1974). Lactating women have shorter biological half-
lives for methylmercury (average value 42 days), compared with nonlactating women (average value
79 days) (Greenwood et al., 1978). This is presumably a reflection of excretion of mercury into milk.
These differences can form the basis for individual and subpopulation sensitivity to methylmercury.
June 1996 2-8 SAB REVIEW DRAFT
-------
The third basis for sensitivity to methylmercury is the magnitude of exposure. Because
methylmercury exposure for humans is almost entirely through fish and shellfish, sensitivity of a
subpopulation will be determined by the extent that they consume fish and shellfish. Analyses of data
for the general United States population indicate that based on dietary surveys conducted during 1989/
1991 only 30.9 percent of the general population reported eating fish at least once during a three-day
period. Subpopulations comprised chiefly of anglers, subsistence fishers, and some Native American
populations report fish consumption rates far in excess of the general population. High fish
consumption is another basis for sensitivity of a subpopulation to methylmercury.
2.2.2.2 Modification of Dose
Critical elements of the dose-response relationship reflect the uncertainty and variability that
are an intrinsic part of this assessment Separation of these identifiable bases for differences may help
establish group variability by contrast to individual variability.
\
As with other toxic chemicals response to methylmercury exposure is influenced by
physiological characteristics of the human subpopulation, as well as by individual characteristics of
members of that subpopulation. Typical factors considered to modify dose-response include these:
presence of concurrent disease; concurrent exposure to other toxic agents; altered nutritional status;
genetic differences in the way the agent is metabolized; and differences in biokinetics, or metabolic
response that depend on physiological statues such as pregnancy or lactation.
Gestation may be the time period in which the adverse effects occur at lowest doses of
methylmercury. In the Japanese epidemic in Minamata it became clear that a considerably higher
number of children than usual were bom with cerebral palsy (Harada, 1995). Many of the mothers of
these infants were themselves either initially asymptomatic or had only mild symptoms of
methylmercury neurotoxicity. Records of the number of inhabitants in the region and onset of disease
are detailed for the Japanese epidemics; however, the exposures were chronic, extending over decades.
The initial cases were of severe methylmercury poisoning and resulted in fatalities (Tsubaki and
Irukayama, 1977). Milder cases, atypical cases and incomplete cases were essentially overlooked in
earlier years (Harada, 1995). Many of the cases showed increasingly severe signs and symptoms over
the years, producing a group labelled as "chronic" Minamata disease patients (Harada, 1995). The
basis for progressive cases is not entirely established; however, manifestation of symptoms by
accumulation of methylmercury caused by a relatively low-level exposure over long periods is one of
the possible mechanisms (Harada, 1995). Generally the thresholds for chronic Minamata disease are
for a lower level of methylmercury than is associated with acute onset of Minimata disease.
The data from Iraq obtained during the epidemic of methylmercury poisoning that occurred in
the early 1970s form another basis for dose-response analyses. Because the epidemic occurred in a
region where maintenance of medical surveillance systems was comparatively undeveloped, and many
of the affected people were from very rural villages or were members of nomadic tribes, there is not a
reliable estimate of the size of the potentially exposed population; that is, in terms of incidence there
are no denominator data. It is uncertain why some subjects who consumed methylmercury-treated
seed-grain responded with adverse effects, whereas other persons with presumably comparable
exposures did not experience toxicity.
Among the Iraqi population reporting methylmercury toxicity, there are reports of the presence
of concurrent disease in the form of parasitism and renal and/or urinary tract disease. Whether or not
these conditions modify the dose-response relationship between methylmercury concentrations in hair
June 1996 2-9 SAB REVIEW DRAFT
-------
and/or blood and prevalence of neuromotor deficits associated with methylmercury remains an
uncertainty.
2.2.2.3 Media Factors that Affect Dose-Response
An additional source of uncertainty and variability in the dose-response assessment is the bio-
toxicity of methylmercury in the food vehicle that was the source of methylmercury. Or, stated
another way, is methylmercury from various biological sources bioavailable? Methylmercury toxicity
has been observed following ingestion of fish, pork, and grain contaminated with methylmercury. The
methylmercury exposure in Iraq occurred from seed-grain treated with methylmercury fungicide,
whereas the methylmercury exposures in Minamata, Niigata, New Zealand, and Canada (Kjellstrom,
1989; McKeown, 1983) occurred from methylmercury incorporated into the protein of fish tissue.
Both of the Japanese epidemics wherein methylmercury exposure was from contaminated fish
and the Iraqi epidemic in which grain contaminated with methylmercury was the vehtele for methyl-
mercury exposure have been extensively reported in the biomedical literature. Although the dose at
which these effects occur more frequently than background incidence is uncertain and variable, it is
clear that clinically significant neurological deficits occur following methylmercury ingestion from
several foods.
2.2.2.4 Time-Course of Dose-Response Assessment: Comparison of Short-Term and Long-
Term Exposures in Human Epidemics
The duration of exposure is also a source of uncertainty. It is unclear whether or not it is
physiologically appropriate to generalize conditions associated with paresthesias developed after a
three-month exposure to methylmercury to a lifetime exposure, as the RfD implies. Analyses of the
Iraqi data and additional analyses of the Niigata data published in 1995 (Kinjo et al., 1995; Harada,
1995) provide useful insights on duration of methylmercury exposure. These epidemics differ in two
major ways. The Japanese dose-response data were obtained from chronic exposures to
methylmercury-contaminated fish and shellfish that occurred over several decades. The methylmercury
was bioaccumulated through the aquatic food chain producing an exposure pathway that is highly
similar to that currently under consideration in this Report to Congress. The Iraqi data were obtained
from a population that experienced short-term exposure (approximately three months) to high levels of
methylmercury ingested as organomercurial- fungicide-contaminated seed grain. The extent to which
differences in exposure vehicle (fish contrasted with grain) and duration of exposure (years contrasted
with months) influence time-course and dose-response to methylmercury among human subjects is not
fully known.
Groups of endpoints from the Iraqi data have served as the bases for RfDs — paresthesia
among adults and neurological deficits among infants of women ingesting methylmercury during or
just preceding gestation. In the Japanese epidemics, signs and symptoms of methylmercury poisoning
included sensory disturbances, constriction of visual field, ataxia, impairment of speech and
impairment of hearing. Sensory disturbances and constriction of visual field were present in 100
percent of Minamata disease cases described in 1968 by Tokuomi, ataxia hi 93.5 percent of cases,
impairment of speech in 88.2 percent of cases and impairment of hearing in 85.3 percent of cases
[Tsubaki et al. (1977) in Tsubaki and Ireheuta, 1977]. Among chronic Minamata disease patients
described by Harada (1995) sensory disturbances (glove and stocking type and generalized type) were
present hi 72 percent (1724/2383) of patients. In both the Minamata disease cases described in 1968
and in the chronic Minamata disease cases, sensory disturbance was the neurological change that
June 1996 2-10 SAB REVIEW DRAFT
-------
occurred first. The sensory disturbances initially were described as "glove and stocking" paresthesia
with about 10 percent of cases having perioral sensory disturbances (Harada, 1995). When exposures
continued and the disease progressed, the clinical course of the disease progressed from sensory
disturbances of the extremities, followed by perioral hypesthesia, ataxia and constriction of the visual
field, with a time lag of several months to several years (Tsubaki and Irukayama, 1977).
In Iraq an outbreak of methylmercury poisoning occurred in 1960 and affected an estimated
1,000 patients resulting in 370 hospital admissions (Jalili and Abbasi, 1961; Kantarjina, 1961). These
early outbreaks alerted clinicians and public health officials to the etiology of the most catastrophic
epidemic of methylmercury poisoning ever recorded. A total of over 6,500 poisoning cases were
admitted to hospitals in provinces, and 459 hospital deaths were attributed to methyhnercury poisoning
(Bakir et al., 1973). Unlike the chronic methyhnercury poisoning from contaminated fish that
occurred in Minamata and Niigata, Japan, the Iraqi epidemic was acute in onset. Distribution of grain
treated with methylmercurial fungicide began hi September, 1971. The rate of admissions of cases to
hospitals throughout the country increased hi early January, 1972 to several hundred cases per day.
No new hospital admissions were recorded after March, 1972. Thus this epidemic occurred following
acute, high-dose exposure to methyhnercury.
Data used hi the quantitative analysis of uncertainty and variability hi the U.S. EPA RfD are
based on the Iraqi data reported by Bakir et al. (1973) as further analyzed by Marsh et al., (1987). As
noted above there are no records of the size of the population who consumed grain treated with
methyhnercury fungicide. Likewise, there are no reliable estimates of the numbers of people who
consumed methylmercury-treated grain and developed signs and symptoms of mercury toxicity, but did
not obtain medical attention or become identified as part of the epidemic. Similar signs and symptoms
of methyhnercury poisoning were noted for the short-term exposure hi Iraq and the chronic exposure
hi Japan. The symptoms progressed hi severity as hi Japan with increased exposure. The frequency
of effects is not directly comparable between the two populations as the size of the exposed Iraqi
population is not known because communication and record-keeping were less than optimal, and at
least part of the population of concern consisted of nomads. Whether or not those who obtained
medical care represented a more sensitive subpopulation is not known. What is known, however, is
estimates of body burden of mercury based on analysis of hair and/or blood mercury concentrations
and the occurrence of a constellation of signs/symptoms of methylmercury toxicity.
•
2.2.2.5 Delivered Dose Estimation
Data obtained during the Japanese epidemic included analyses of hair mercury concentrations.
In the Iraq epidemic analyses of mercury concentration hi hair and blood were carried out. Both sets
of data have been used to estimate dose of methylmercury to affected subjects. An analysis of the
threshold dose for adults exposed to methyhnercury in Niigata was published by Kinjo et al. (1995).
To be included as subjects the individuals had been classified as having Minimata Disease. This
definition is presented hi multiple publications including that of Tamashiro et al. (1985). The sign
common to the syndrome of Minamata disease is the bilateral sensory disturbance which is more
severe hi the distal parts of the extremities and which also occurs sometimes hi the perioral area
(Takashiro et al., 1985). The raw data on hah- mercury concentrations did not take hah- length or hair
growth rate into account Consequently the actual mercury measurements can be considered to
represent average values over the period of exposure to pollution derived from hair length and hair
growth rate. Kinjo et al. (1995) include thresholds based on raw data; however, these investigators
considered the maximum hah- mercury concentration to be the more appropriate measure for dose-
response analysis. Maximum hair mercury concentrations were estimated using actual mercury
June 1996 2-11 SAB REVIEW DRAFT
-------
concentrations and estimates of hair growth rate and biological half-lives for methylmercury. The
biological half-life primarily used in their model was 70 days with a hair length of 10 cm for males
and 20 cm for females and a hair growth rate of 1.5 cm/month. Additional biological half-lives (35
and t20 days) and different hair lengths (5 cm for males, 15 cm and 25 cm for females) were
evaluated by changing these variables in the equations used to predict thresholds. The threshold dose
of hair mercury concentration was estimated to be between 40 and 70 ppm by hockey-stick regression
analysis. A wider range of threshold doses was observed when raw hair mercury data were used.
Based on raw data from female subjects a threshold of 21 ppm mercury in hair was identified. Using
a 70-day biological half-life and a hair length of 5 cm, a threshold of 67 ppm was observed.
Data from the Iraqi epidemic were used in development of U.S. EPA's RfD which was
developed hi 1994. These data were input parameters to a physiologically based dose conversion
model for mercury. This model served as the mathematical basis for estimating exposure to mercury
per kilogram body weight per day. Although this model has been extensively used (among other
applications, the National Research Council/National Academy of Sciences' committee report entitled
Seafood Safety, the World Health Organization's Criteria Document on Methylmercury) any
differences between model parameters and actual "values will determine the predictions made. This
model relies on fundamentals such as the hair-to-blood ratio and the half-life of methylmercury in the
blood.
Variability hi biological half-life of mercury has been cited above. Generally a value of 70
days has been used. However, individual values as long as 250 days have been reported by Birke et
al. (1972). Al-Shahristani and Shihab (1974) reported biological half-lives of methylmercury to vary
between 35 and 189 days with an average of 72 days based on data from 48 patients. It is known that
at least one subpopulation has a different value for the half-life of mercury that differs from the
general adult population; lactating women had a shorter half-life for mercury than did nonlactating
adults (Greenwood, 1978).
Extrapolation of dose-response conversions across a wider range than the range of the actual
data results in uncertainty; this occurs when modeled data are used to predict beyond the range of
observed data. Significant departures from non-linearity or differences between the shape of the
modeled dose-response curve and the observed data may occur at extreme in the distribution. This is
an intrinsic issue when modeled data are utilized. Whether or not intermittent exposures resulting
from occasional consumption of highly contaminated media results in similar biokinetics of
methylmercury remains an uncertainty.
2.2.2.6 Grouping of Data
Dose groupings other than those used hi Seafood Safety were also done and benchmark doses
run as above for comparison. Both density-based grouping and uniform concentration intervals were
used.
The local density of observations relative to the mercury level in hair was analyzed using a
density estimation algorithm (ksmooth function in S-PLUS for Windows, Ver. 3.1; S-PLUS Guide to
Statistical and Mathematical Analysis). The function estimates a probability density for the
distribution of a variable by calculating a locally-weighted density of the observations. That is, the
function estimates the probability that an observation will be near a specific value based on how the
actual values are clustered. In this case, the function was used to estimate the probability density for
an observation in the neighborhood of any given maternal hair mercury concentration. The density
June 1996 2-12 SAB REVIEW DRAFT
-------
plot is shown in Figure 2-1. The peaks represent relatively greater numbers of data points than the
troughs in the vicinity of the associated hair mercury concentrations.
Figure 2-1
Density of Data Points Relative to Hg Concentration
in Hair for Iraqi Cohort Data
ppm Hg in hair
The density distribution is characterized by four distinct peaks. Exposure dose groups were defined as
trough-to-trough intervals with the peak values taken as the nominal value for each interval. The
nominal dose-group value, concentration ranges and incidence of combined developmental effects are
given in Table 2-1. A benchmark dose was calculated from the incidence of all effects as grouped in
Table 2-1. The lower 95% confidence interval on the benchmark dose for 10% response is 13 ppm
compared to the 11 ppm value used as the basis for the RfD.
Table 2-1
Density-Based Dose Groupings
Nominal Dose (ppm)
1.18
10.6
78.8
381
Dose Range (ppm)
1 -4
5 -28
29 - 156
157 - 674
Incidence
5/27
3/16
10/17
18/21
The other alternative dose grouping approach was to divide the entire exposure range into four
equal log-dose intervals. The geometric midpoint of each interval was taken as the nominal value for
the interval. The nominal dose-group value, concentration ranges and incidence of combined
developmental effects are given in Table 2-2. The benchmark calculated as the lower bound on the
10% incidence for all effects is 10.3 ppm, compared to the 11 ppm used for the RfD.
June 1996
2-13
SAB REVIEW DRAFT
-------
Table 2-2
Uniform Dose Groupings
Nominal Dose (ppm)
2.25
11.5
58.6
299
i
Dose Range (ppm)
1 -5
6-25
26 - 132 .
133 - 674
Incidence
5/28
3/14
9/17
19/22
2.2.2.7 Paresthesias as a Reliable Endpoint
The former RfD of 3xlO"4-mg/kg-day was based on paresthesia in adults. A re-evaluation of
the data set and exposure calculation was done with subsequent determination of a benchmark dose for
paresthesia in adults of 3.6 ug/kg body weight/day (RfD Work Group Notes of 13 October 1994).
Among the uncertainty and variability issues in use of transient paresthesias as an adverse health effect
is the subjectivity of the condition. Transient paresthesias refers to tingling and numbness of
extremities or the mouth area for a temporary period and is a clinically defined endpoint. These
temporary paresthesias are fully reversible and occur in a number of benign (e.g., position of a limb
during sleep) or serious conditions (e.g., osteoarthritis or diabetes). The duration of a temporary
paresthesia is an important consideration and can range from a few minutes to hours or days.
In the epidemics of methylmercury poisoning in Minamata and Niigata, the development of
paresthesias was extensively described (among others see Tsubaki et al., Neurological Aspects of
Methylmercury Poisoning in Tsubaki and Irukayama, 1977). Sensory abnormalities were identified
and considered an early indication of methylmercury poisoning hi the Iraqi epidemic (Bakir et al.,
1973). It is unclear from the published materials what duration of effect was needed to be classified
as paresthesia. Reporting of paresthesia may reflect subject or examiner recall bias in either a negative
or positive direction. Consequently this endpoint is quite subject to classification bias; however,
personal communication from one of the investigators (Dr. Thomas Clarkson, University of Rochester,
July, 1995) indicated that the clinicians who conducted the initial Iraqi investigation were familiar with
the paresthesias produced by methylmercury exposure because they had evaluated Iraqi patients in the
earlier epidemic in 1960. Although a standardized definition of paresthesia was very likely not
developed, the investigators were familiar with the clinical picture of methylmercury-induced sensory
disturbance.
A second issue for analyses of data on paresthesias is the background prevalence of temporary
paresthesias in the subpopulation of interest. If temporary paresthesias were narrowly defined as
caused only by methylmercury exposure, one interpretation of an appropriate background rate would
be zero. Temporary paresthesias occur, however, hi a number of benign and disease conditions. In
the uncertainty analysis (Appendix D to Volume IV) carried out in support of this risk
characterization, determination of a background rate was based on Bakir et al. (1972). The response
data for exposed individuals do not show any background response, and so there does not appear to be
an appreciable background rate of paresthesia in the general population. An estimate of 7.2 percent
was developed from the data of Bakir et al. (1972, 1973) representing 40 hospitalized subjects. The
June 1996
2-14
SAB REVIEW DRAFT
-------
benchmark dose modeling for paresthesias used the prevalence of paresthesias among 35 female
subjects whose hair mercury concentrations were under 10 parts per million.
The calculated dose for subjects with paresthesia used a 70-day half-life as the measure of
central tendency. Duration of exposure is also a major concern in calculation of dose of
methylmercury exposure that produces paresthesias. Methylmercury is retained in tissues. In the
methylmercury poisoning epidemic in Iraq, the duration of exposure to methylmercury was estimated
to be three months duration, although exposures as long as six months could have occurred (using
September, 1971 the date when methylmercurial seed-grains were introduced and March, 1972 as the
date of last hospitalization of cases). If exposure is prolonged, the dose estimated to produce
paresthesias may differ based on laboratory data identifying the mechanisms of action by which
methylmercury produces nerve damage. A detailed discussion of exposure duration, short vs. long
exposure to methylmercury in production of paresthesias is shown in Appendix D of Volume IV.
2.2.2.8 Neuro-Developmental Effects
As with other health-based endpoints, the general issues of representativeness of the population
who sought medical attention and became subjects in the study is a concern. In the Japanese
epidemics extensive medical surveys were done during the 1960s hi Minamata and Niigata (1965 and
1967) (Tsubaki and frukayama, 1977; also reviewed by Harada, 1995). Identification of severe
developmental disturbances were among the earlier changes identified among patients born from 1955
and later hi the Minamata area of Kyushu, Japan (Harada, 1977, 1995). Under the conditions present
hi Minamata area during 1955-1957, Harada identified an overall morbidity of 6.9 percent, which was
much higher than the rate of usual congenital cerebral palsy present in Japan (Harada, 1977). Harada
noted (Harada, 1995) that for congenital Minamata disease, as with other cases of infantile cerebral
palsy, the diagnosis occurs only after an extended time has elapsed since birth. In small fishing
villages of Yudo, Tsukinowa, and Modo, Japan between 1955 and 1958 there were 188 births with a
9.0 percent incidence of cerebral palsy (Harada, 1995). During this period the overall national
incidence of cerebral palsy was approximately 0.2 percent (Harada, 1995).
In the Iraqi epidemic, the first reports of infant-mother pairs exposed to methylmercury did not
indicate an unusual sensitivity of the fetus compared to the exposed adult (Amin-Zaki et al., 1976).
Follow-up at five years, however, indicated developmental delays in motor skills and unpaired
intelligence in one-sixth of the young children (Amin-Zaki et al., 1981). Delayed motor development
was defined as inability of the infant to sit without support by the age of 12 months, to pull
himself/herself to standing position by 18 months, or to walk two steps without support by 2 years of
age. Language development was considered to be delayed when, at the age of 2 years, a child with
good hearing failed to respond to simple verbal communication. There are no standardized
intelligence quotient ranges for Iraqi children. The child's mental development was judged based on a
combination of the mother's impressions of the child's development and the judgment of two
physicians.
The background prevalence of late talking/late walking among the Iraqi population not
exposed to methylmercury is an uncertainty. The major part of the variance in the developmental
effects threshold distribution arises from uncertainty in the estimate of the threshold based on ppm
mercury in hair, which accounts for 84 percent of the variance. These data show a very broad range
of susceptibilities in this exposed population, up to a 10,000-fold span between the 5th and 95th
percentiles when projected to the general population (data of Marsh et al., 1987, as analyzed by Hattis
and Silver, 1994). A primary factor is that hair methylmercury concentrations imprecisely predict
June 1996 2-15 SAB REVIEW DRAFT
-------
toxicity, either because some important data are missing or because significant nonlinear processes are
involved. For example, in the Marsh et al. (1987) data, it is noted than an individual with the highest
estimated methylmercury exposure is a non-responder when the endpoint is developmental effects on
the nervous system. This could reflect individual susceptibility to methylmercury toxicity.
Alternatively, this observation may be a consequence of misclassification — the individual may have
been exposed during a period of time which was not a critical developmental window. There is
potential for misclassification as calculation of exposure time was dependent on subject recall of the
gestational period and birth date.
Recall of birth data* for the infant is of major importance in assessing the prevalence of
developmental delays such as late walking or late talking. This uncertainty is particularly an issue
with the Iraqi data set because of cultural differences. Published information and personal
communication with the study authors suggest that within the Iraqi nomadic culture no particular
significance is attached to the age at which walking and talking first occur. The database used to
assess the distribution of ages in which late walking and late talking are assessed is a European
database. It is known that ethnicity and race are factors that influence age at which motor skills are
acquired.
2.3 Uncertainty in the Human Health RfD for Methylmercury
2.3.1 Qualitative Discussion of Uncertainties in the RfD for Methylmercurv Alternate Analyses
Two additional human epidemiologic studies of separate populations (Kjellstrom et al.,
1986a,b, 1989; McKeown-Eyssen et al., 1983) generally support the dose range of the benchmark dose
level for perinatal effects. Both of these studies are described hi section 3.3.1.1 of Volume IV. A
recent analysis of the Kjellstrom data was published by Gearhart et al. (1995). In this analysis the
authors used a PBPK model which Incorporated a fetal compartment They calculated a benchmark
dose on all 28 tests included in the initial study design by Kjellstrom; this was done assuming values
of 1 and 5% for background deficiency hi test scores. The range of benchmark doses calculated was
10 to 31 ppm maternal hair mercury. The authors' preferred benchmark was 17 ppm, for an estimated
background incidence of 5% and the lower bound on the 10% risk level.
Chronic rodent (Bornhausen et al., 1980) and nonhuman primate studies (Burbacher et al.,
1984; Gunderson et al., 1986; Rice et al., 1989a,b) provide data to support LOAELs for other
developmental end points.
The principal study (Marsh et al., 1987) is a detailed report of human exposures with
quantitation of methylmercury by analysis of specimens from affected mother-child pairs. A strength
of this study is that the quantitative data are reported on the affected population, and quantitation is
based upon biological specimens obtained from affected individuals. A threshold or presumed no
effect level was not easily defined; application of modeling techniques were needed to define the lower
end of the dose-response curve. This may indicate high variability of response to methylmercury in
the human mother-child pairs or misclassification of assigning pairs to the cohort. Concerns have been
raised as to the applicability of a risk assessment based upon data from grain-consuming population
when the application of this risk assessment is for segments of the U.S. population consuming fish. It
is thought that a diet rich in animal protein (such as fish) also delivers selenium. Selenium appears to
interact with mercury in some experimental systems and has been suggested to increase the latency
period for onset of symptoms of neurotoxicity which has been observed in exposed humans. It is not
June 1996 2-16 SAB REVIEW DRAFT
-------
thought that the exposed Iraqi population was selenium-deficient or significantly malnourished:
however, the effect of additional dietary selenium on the dose-response curve is uncertain.
The most appropriate basis for calculation of an RfD for methylmercury has been the subject
of much scientific discussion; several plausible alternatives to the U.S. EPA assessment have been
proposed. ATSDR used the analysis reported by Cox et al. (1989, see discussion below) of the Iraqi
developmental data in the derivation of an intermediate MRL (minimal risk level). Using delayed
onset of walking as the critical effect, a LOAEL of 14 ppm mercury in hair was determined. A dose
conversion from ppm hair to daily intake to maintain blood mercury levels hi pregnant women was
done in a very similar manner to that employed by U.S. EPA. Values for parameters in the equation
were consistent between the two agencies with one exception; namely the use of a blood volume of
4.1 L by ATSDR compared to 5 L by U.S. EPA. The methylmercury intake level calculated by
ATSDR to maintain a hair level of 14 ppm is 1.2 ug/kg-day compared to 1.1 ug/kg-day to maintain a
hair level of 11 ppm (used by U.S._EPA), this is not a significant difference.
The state of New Jersey currently uses an RfD of 0.7xlO~4 mg/kg-day (described in Stern,
1993) compared to U.S. EPA's RfD of IxlO"4 mg/kg-day. The critical effect chosen was
developmental endpoints in the Iraqi children exposed in utero including delayed onset of walking.
The LOAEL chosen was the mercury hair level equivalent to a mercury blood level of 44 ug/L. To
determine the intake level, the equation in Section 2.2.1.2 of this volume was used, but with different
values for two parameters, namely, b and f.
Crump et al. (1995) reanalyzed data from the Iraqi methylmercury poisoning episode presented
by Marsh et al. (1981). Using a hockey stick parametric dose-response analysis of these data, Cox et
al. (1989) concluded that the "best statistical estimate" of the threshold for health effects was 10 ppm
mercury in hair with a 95 percent range of uncertainty between 0 and 13.6. In their analysis, Crump
et al. (1994) reported that the statistical upper limit of the threshold could be as high as 255 ppm.
Furthermore, their maximum likelihood estimate of the threshold using a different parametric model
was said by the authors to be virtually zero. These and other analyses demonstrated that threshold
estimates based on parametric models exhibit high statistical variability and model dependency, and
are sensitive to the precise definition of an abnormal response.
Using a statistical analysis for trend that does not require grouping of the data, Crump et al.
(1994) demonstrated that the association between health effects and methylmercury concentrations in
hair is statistically significant at mercury concentrations in excess of about 80 ppm. In addition,
Crump et al. (1994) calculated benchmark doses by applying dose-response models to each of the
three endpoints: late walking, late talking and neurological score. Their calculation of the 95 percent
lower bounds on the hair concentration corresponding to an additional risk of 10 percent ranged from
54 ppm to 274 ppm mercury in hah*. Crump et al. (1994) concluded that the trend analyses and
benchmark analyses provided a sounder basis for determining RfDs than the type of hockey stick
analysis presented by Cox et al. (1989). They felt that the acute nature of the exposures, as well as
other difficulties with the Iraqi data, present limitations in the use of these data for a chronic RfD for
methylmercury.
Cox et al. (1995) have published a recent analysis of the data on late walking in Iraqi children
exposed in utero to methylmercury. The authors indicate that dose-response analyses based on the
"late walking" endpoint are unreliable because of four influential observations in the data set from
Marsh et al. (1978). The data points hi question are the only responders below 150 ppm (Hg in hair).
In particular Cox et al. (1995) state that the four observations are isolated from the remainder of the
June 1996 2-17 SAB REVIEW DRAFT
-------
responders and would be expected to have considerable influence on threshold estimate. This
conclusion is based on a visual interpretation of a plot of the data (Figure 2 in Cox et al., 1995).
Based on visual inspection of the same figure, an argument could be made that the separation is not
that marked considering the first eight responders. No quantitative sensitivity analysis was performed
to investigate the effect of removing one or more of these data points. Cox et al. (1995) point out that
if the four points are assumed to represent background, then the threshold for late walking would be
greater than 100 ppm. It would seem unlikely, however, that these observations represent background
given that no responses were observed in the 37 individuals with lower levels of exposure. It should
be noted that the U.S. EPA benchmark dose was done on incidence of all effects, rather than on late
walking only.
The Cox et al. (1995) and Crump et al. (1995) analyses deal primarily with one endpoint;
namely, late walking. This appears to be the most sensitive of the endpoints described in March et al.
(1978). Both Cox et al. and Crump et al., as well as the U.S. EPA analysis in Appendix D of Volume
IV, show considerable uncertainty in thresholds estimated from the data on late walking.
Late walking, as assessed in the exposed Iraqi population (March et al., 1978) is almost
certainly a valid indicator of methylmercury toxicity but may well be unreliable as the sole basis for
detailed dose-response analysis. The primary reason for this may be the uncertainty in maternal recall
for both birth date and date of first walking. The uncertainty, in this particular case could be quite
large, given the lack of recorded information. The'primary impact of this kind of uncertainty would
be on the response classification of individuals at the upper bound of normal (18 months for first
walking) and at the lower bound of abnormal. The lowest abnormal first walking times presented in
March et al. (1978) 20 months. The impact of assuming uncertainty in the classification of the
observations in these two groups is large given the large number of observations in the two groups (19
data points at 18 months and 8 data points at 20 months). The analysis in Appendix D to Volume IV
of the Report to Congress shows that thresholds estimated for late walking are unstable when
classification uncertainty is considered. The same kind of subjective uncertainty is applicable to the
late walking endpoint, as well. The thresholds for late walking, however, are much more stable,
statistically, as there are fewer observations that are near the normal/abnormal threshold value of 24
months.
Marsh et al. (1995) have published results of a study conducted between 1981 and 1984 in
residents of coastal communities of Peru. The prospective study was of 131 child-mother pairs; testing
for potential effects of fetal methylmercury exposure ^s*patterned after the study of children exposed
in utero in Iraq. Peak maternal hair methylmercury ranged between 1.2 to 30 ppm with a geometric
mean of 8.3 ppm. These authors showed no effects of methylmercury on measures similar to those
performed on the Iraqi children (including time of first walking and talking). A NOAEL (in the
absence of a LOAEL) from this study would be 30 ppm maternal hair mercury. This is consistent
with the U.S. EPA benchmark dose of 11 ppm.
Fetal effects of methylmercury exposure were based on hair mercury analyses from 83 women
in Iraq. Recommendations based on this data set are a best estimate based on a relatively small
number of human subjects. The size of the data set becomes a limitation for identifying adverse
effects that may occur in a small fraction of subjects due to factors such as individual variability. A
limitation of these data is the relatively small number of maternal-infant pairs (81) whose exposures
fell within the range of interest for this assessment. Efforts to interpret these data have considered the
issue of threshold modeling (among other references see the NIEHS Report to Congress on
Methylmercury, 1993). The duration of the exposure to methylmercury (approximately three months
June 1996 2-18 SAB REVIEW DRAFT
-------
in the Iraqi outbreak) was long enough to identify the effects of methylmercury exposure on the
outcome of pregnancy.
Concern has been raised by various scientists as to the impact that as yet unpublished studies
will have on the risk assessment for methylmercury. Reports have delivered at scientific meetings
results of studies of populations in the Faroes and Seychelles Islands known to consume large amounts
of seafood. Data on parts of the Seychelles Study have recently been published. The interpretation by
some risk assessors is that the effects noted in the Iraqi population exposed to contaminated grain are
itot being seen at similar doses of methylmercury delivered in utero via contaminated seafood.
As the majority of these new data are either not yet published or have not yet been subject to
rigorous review, it was decided that it was premature for U.S. EPA to make a change hi the
methylmercury RfD at this time. An interagency process, with external involvement, will be
undertaken for the purpose of review of these new data, evaluations of these data and evaluations of
existing data. An outcome of this process will-be assessment by U.S. EPA of its RfD for
methylmercury to determine if change is warranted.
It has been suggested that a developmental toxicity RfD is needed for methylmercury. This
may not be necessary, however, if the critical effect is developmental toxicity and the uncertainty
factors used to estimate the lifetime RfD do not involve an adjustment for less than lifetime exposure
nor lack of complete database.
2.3.2 Quantitative Analysis of Uncertainty in the Methvlmercury RfD
2.3.2.1 Introduction
This section summarizes the methylmercury RfD uncertainty analysis presented in Appendix D
to Volume IV of this Report Details of the methods applied and the results obtained can be found in
Appendix D. The purpose of this analysis is two-fold: first, to determine plausible bounds on
uncertainty associated with the data and dose conversions used to derive the methylmercury RfD;
second, to compare the RfD to estimated distributions of human population thresholds for adverse
effects. This analysis is a modeled estimate of the human threshold for specific health effects
attributable to methylmercury exposure. The basis for the analysis and the RfD is the data from the
1971 Iraqi methylmercury poisoning incident, specifically the data from the Marsh et al. (1987)
population referred to as the Iraqi cohort An adult paresthesia benchmark dose was also based on
data presented in Bakir et al., (1973). The analysis also includes studies pertinent to the conversion of
mercury concentrations in hair to estimated ingestion levels.
For purposes of this analysis, the human population threshold was defined as the threshold for
the most sensitive individual of an identified sensitive subpopulation. The definition of sensitive
subpopulations excludes hypersensitive individuals whose susceptibilities fall far outside the normal
range. A threshold is defined as the level of exposure to an agent or substance below which a specific
effect is not expected to occur. The definition of threshold does not include concurrent exposure to
other agents eliciting the same effect by the same mechanism of action. In other words, there is an
assumption that the induced response is entirely a result of exposure to a single agent. The 81
pregnant female/offspring pairs comprising the Iraqi cohort were taken as a surrogate for the most
sensitive subpopulation expected in the general U.S. population consuming fish. The sensitive
subpopulation was identified for the uncertainty analysis as humans exposed to methylmercury. in
utero.
June 1996 2-19 SAB REVIEW DRAFT
-------
•The uncertainty analysis examined the major sources of uncertainty explicitly and implicitly
inherent to the methylmercury RfD and attempted to bound them quantitatively. The principal
uncertainties arise from the following sources: the variability of susceptibilities within the Iraqi
cohort; population variability in the pharmacokinetic processes reflected in the dose conversion; and
response classification error.
The response classification is the assignment of an individual observation to one of two
categories — responder or nonresponder. The response classification for each of the developmental
endpoints reported by Marsh et al. (1987) was based on a fixed value (response decision point) that,
when exceeded, constitutes a response. It is possible that some responses were misclassified,
particularly those for responses in the immediate vicinity of the response decision point; a responder
may have been classified as a nonresponder or vice versa. The response classifications for late
walking and late talking are particularly susceptible to this type of error. The response estimates were
based on subject recall hi members of a population that does not traditionally record these events.
Other areas of uncertainty are those directly related to the RfD methodology. Specifically, it
was concluded by an Agency Work Group that there were no adequate chronic or reproductive studies.
An uncertainty factor of 10 is generally applied when chronic studies are not available. This
uncertainty factor is based on an assumption inherent to the RfD methodology that increased exposure
duration will lower the dose required for observation of the effect. Support for this assumption has
been published (Weil arid McCollister, 1963; Dourson and Stara, 1989) and is discussed in Section
D.2.2.2 of Appendix D to Volume IV. An uncertainty factor of 3 is generally applied if reproductive
studies are not available. NOAELs for reproductive studies are generally two-fold to three-fold higher
than NOAELs for chronic studies and are not expected to be the basis for the RfD more than 5 percent
of the time (Dourson et al., 1992).
2.3.2.2 Methods
Thresholds were estimated in a two-stage process. The first stage was the estimation of
threshold distributions based on hair mercury concentrations, which was accomplished by applying a
regression model to successive bootstrap samples of the observations in Marsh et al. (1987). This
process is detailed in Section D.2.1 of Appendix D to Volume IV. The second stage was the
conversion of the thresholds expressed as ppm mercury hi hair to mg methylmercury per kg body
weight per day (mg/kg-day); this involved a Monte Carlo analysis of the variability of the underlying
biological processes. For details of methods, see Appendix D to Volume IV.
Because the Iraqi cohort is considered to be a sensitive subgroup, as defined in the RfD
methodology, the output distributions of the uncertainty analysis are meant to reflect the uncertainty
around an estimate of the thresholds for effects in humans including sensitive individuals. The results
for each endpoint should be interpreted as the distribution of the uncertainty around the human
population threshold. The results should not be interpreted as the distributions of individual thresholds
within the population. Estimates of risk above the threshold cannot be obtained from this analysis.
The uncertainty analysis was limited to only those data and equations directly related to the
derivation of the methylmercury RfD. Other data sets or models were not considered. A few sources
of uncertainty in the data used to derive the methylmercury RfD have not been included in this
analysis. Exposure classification error arising from uncertainty as to the correspondence of actual
exposure and critical exposure period cannot be estimated from the data as published by Marsh et al.
(1987). This source of uncertainty could be a major contributor to the apparent extreme variability of
June 1996 2-20 SAB REVIEW DRAFT
-------
susceptibilities in the Iraqi cohort. Variability in the interpretation of the definition of a response was
not estimated in this analysis. That is, there would be expected differences in individual interpretation
of first walking or first talking (probably for the latter). The classification errors assumed for this
analysis only accounted for uncertainty in the timing of the event given an unequivocal positive
response. Also, the response decision points defining an adverse effect were accepted uncritically.
For example, changing the definition of late walking to either greater than 16 months or greater than
20 months would have a significant effect on the analysis. Measurement error for hair mercury
concentrations has not been estimated for this analysis; the necessary data are unavailable in the
published reports (Marsh et al., 1987; Cox et al, 1989).
The results of this analysis are conditional on a specific representation of population variability
in the parameters of the dose conversion variables. That is, the choice of the form and parameters for
the distributions assigned to each of the variables is largely a matter of judgment; the particular set of
parameters chosen for each distribution is only one option of a number of possible choices; and
uncertainty as to the value of the parameters is not included in the analysis. For example, the choice
of the (log-triangular) distribution for half life of methylmercury was made on the basis of best fit
with respect to the 5th, 50th and 95th percentiles of the combined_data from several studies. This
particular distribution does not allow for values less than 28 days or greater than 125 days, but could
be easily modified to do so. Such a modification would, however, have only a small effect on the
Monte Carlo-generated distribution for the dose conversion factor.
The threshold analysis shows that adult paresthesia was the most sensitive individual effect
observed for the Iraqi cohort, particularly when adjusted for the effects of continuing exposure. That
is, in this analysis, paresthesia in adults was estimated to be observable at a lower exposure than the
developmental endpoints. The adult paresthesia bootstrap thresholds were also the most unstable as
measured by the frequency of nonsignificant slopes. The RfD fell between the 39th and 91st
percentiles of the duration-adjusted adult paresthesia threshold distribution, a considerably larger range
than that for any of the developmental effects. On the average, the RfD fell below the 1st percentile
for all developmental effects, with only a 5 percent chance that it was as high as the 16th percentile.
A discussion of factors affecting reliability of paresthesia as an endpoint is provided in Section 5.1.3.1
of this volume.
The results of the response-classification uncertainty analysis suggest that the late walking
endpoint and adult paresthesia were unreliable as measures of methylmercury toxicity for the Iraqi
cohort. The exclusion of late walking from the combined developmental effects would not have a
very large impact on the threshold distribution, increasing the thresholds by about 50 percent.
Although the response-classification uncertainty analysis was based on hypothetical classification error
rates, a two-month uncertainty in recall of these events was not unlikely in this particular situation.
These results suggest that strong conclusions should not be based on the late walking and adult
paresthesia endpoints.
2.3.2.3 Conclusions of Analysis of Uncertainty Around Human Health Effects of
Methylmercury
A major source of the variability was in the estimation of bootstrap thresholds from the Iraqi
cohort data as evidenced by the 12- to 20-fold difference in the 5th and 95th percentiles of the
bootstrap threshold distributions. The uncertainty arising from limited exposure duration contributed
almost as much, with a 12.5-fold difference in the 5th and 95th percentiles. The corresponding spreads
hi the dose conversion distributions were 2.4-4.2 fold. Correlations between variables were important
June 1996 2-21 SAB REVIEW DRAFT
-------
with respect to the variance of the Monte Carlo, simulations but were not well-defined by empirical
data. Additional areas of uncertainty remain to be modeled.
Of the developmental endpoints, the neurological effects, which are determined by a battery of
tests and do not depend on subject recall, would seem to be the most objective measure of
methylmercury toxicity. Late walking was not a reliable endpoint because of sensitivity to
classification error.
The RfD of IxlO"4 mg/kg-day is very likely below the threshold for developmental effects but
may be above the threshold for exposure duration-adjusted adult paresthesia. Strong conclusions based
on the latter result are not.warranted because of the sensitivity of the adult paresthesia threshold to
classification error and the general lack of data addressing the effects of exposure duration.
June 1996 2-22 SAB REVIEW DRAFT
-------
3. WILDLIFE HEALTH EFFECTS: HAZARD IDENTIFICATION AND DOSE-
RESPONSE
3.1 The Framework for Ecological Risk
U.S. EPA defines ecological risk assessment as "a process that evaluates the likelihood that
adverse ecological effects may occur or are occurring as a result of exposure to one or more stressors"
(U.S. EPA, 1992a). Although ecological risk assessment follows the same basic risk paradigm as
human health risk assessment, there are three key differences.
• Ecological risk assessment can consider effects on populations, communities and
ecosystems hi addition to effects on individuals of a single species.
• No single set of eGological values to be protected is applicable in all cases; instead,
they must be selected for each assessment based on both scientific'and societal merit
• Nonchemical stressors (e.g., physical disturbances) often need to be evaluated as well
as chemical stressors.
The problem formulation phase of an environmental risk assessment consists of three main
components: (1) characterizing the stressors, potential exposure pathways, ecosystems potentially at
risk, and ecological effects; (2) selecting endpoints (the ecological values to be protected); and (3)
developing a conceptual model of the problem (U.S. EPA, 1992a).
In this Report only terrestrial and freshwater aquatic ecosystems were considered for
evaluation. It is recognized that mercury deposited hi coastal areas can be translocated to estuarine
environments, and biota that inhabit these and nearby marine systems have the potential to be
adversely impacted. Presently, however, uncertainties regarding mercury deposition, cycling and
effects hi marine environments are so great as to preclude even a qualitative risk assessment
Of the pathways by which ecosystems and components of ecosystems might be exposed to
atmospheric mercury, exposure of high trophic level wildlife to mercury in food is particularly
important. The trophic level and feeding habits of an annual influence the degree to which that
species is exposed to mercury. Mercury biomagnifies in aquatic food chains with the result that tissue
concentrations of mercury increase as trophic levels increase. Predatory animals primarily associated
with aquatic food chains accumulate more mercury than those associated with terrestrial food chains.
Thus, piscivores and carnivores that prey on piscivores generally have the highest exposure to
mercury. In a study of fur-bearing mammals hi Wisconsin, the species with the highest tissue levels
of mercury were otter and mink, which are top mammalian predators on aquatic food chains (Sheffy
and St. Amant, 1982). Top avian predators of aquatic-based food chains include raptors such as the
osprey and bald eagle. Smaller birds feeding at lower levels in aquatic food chains also may be
exposed to substantial amounts of mercury because of their high food consumption rate
(consumption/body weight/day) relative to larger birds. In the methylmercury poisoning epidemics in
Minamata, Japan, extensive poisoning of marine life, birds that fed on the marine life, and domestic
animals were reported hi years preceding the severe human poisonings that occurred (Irukayama et al.,
1977). The general picture was one of severe neurological damage to multiple species in the aquatic
food chain, including death of fish; severe incoordination and death in birds; ataxia and convulsions in
domestic cats; and finally gross neurological defects and death in humans (Tsubaki and Irukayama,
1977).
June 1996 3-1 SAB REVIEW DRAFT
-------
Although clear causal links have not been established, mercury originating from airborne
deposition may be a contributing factor to population effects on bald eagles, river otters and mink.
Evidence is available to support the possibility of toxic effects on the common loon and the Florida
panther. Effects of mercury originating from point sources on restricted wildlife populations have
been conclusively demonstrated and provide a tissue residue basis for evaluation of risk to other
populations.
Effects data were insufficient for evaluation of any intact ecosystem, community of species or
population of animal or plants. This limitation of data necessitated the choice of individual wildlife
effects for selected species as the ecological value to be protected.
3.2 „ Health Hazards of Methylmercury Exposure to Wildlife
Methylmercury present in terrestrial and aquatic food chains creates risks to the health and
reproductive success of wildlife. The epidemics of human disease from grain treated with mercurial
fungicides and from fish contaminated with methylmercury were preceded by major epizootic
outbreaks of death, neurological disease, and reproductive failure among wildlife and domestic
animals. During the period 1940 through the 1970s, treatment with seed grains with organomercurial
fungicides resulted in poisonings of massive numbers of seed-eating birds and their predators (Borg et
al., 1970). In Minamata death and serious neurological disturbances among birds and cats were
reported in the years preceding the first cases of human Minamata disease in 1956 (Tsubaki and
Irukayama, 1977). As is the situation for humans, the risk characterization issues for wildlife and
domestic animal effects of methylmercury from fish center on the quantities of methylmercury that
produce disease and on the likelihood of exposure to these quantities of methylmercury.
A principal weakness identified in the current risk characterization for methylmercury is the
limitations of the toxicity database for the wildlife species. Problems identified during risk
characterization included these: limited information in case reports of small numbers of animals, and
either no, or limited, exposure data; uncertainty about the agent that caused the toxic response in
wildlife potentially exposed to pesticides, PCB, etc.; no monitoring for subtle indicators of effect (e.g.,
paresthesias) even in experimental situations; questions on whether or not sensory deficits were of
concern for wildlife; shorter than chronic studies; dose-groups in experimental studies that did not span
the range of the NOAEL to development of frank effects; limited data on tissue mercury
concentrations; and limited histopathological data.
3.2.1 Mammalian Species
Methylmercury toxicity hi non-human mammals appears to follow the pattern observed in
humans: neurotoxicity is the endpoint of concern. Methylmercury adversely affects the central
nervous system of multiple species of wildlife producing sensory, visual, auditory and motor
impairment. In chronic studies by O'Connor and Nielse (1981) using river otters (Lutra canadensis)
fed methylmercury, 2 ppm (0.09 mg/kg bw/day) caused anorexia and ataxia during the six-month test
period. In mink, 27 ppm of dietary phenylmercuric chloride caused lethality hi 40 percent of the
males and 31 percent of the females within six weeks of exposure (Borst and Lieshout, 1977).
Wobeser et al. (1976a,b) reported neurotoxicity hi adult mink and kits fed methylmercury hi fish or
chow. Methylmercury has been implicated hi Florida panther deaths and population decline, and has
been reported to produce neurotoxic effects in domestic cat.
June 1996 3-2 SAB REVIEW DRAFT
-------
3.2.2 Avian Species
The most complete dose-response information for avian wildlife species describe the adverse
effect of methylmercury on reproduction. These data exist for multiple avian wildlife species. For
example, Fimreite (1971) identified a LOAEL for reproductive effects (reduced survival, reduced egg.
production, defective shells) of 0.18 mg/kg-day in ring-necked pheasants (Phasianus colchicus) fed
seed treated with methylmercury dicyandiamide. Scott (1977) identified a LOAEL for reproductive
effects (reduced fertility, reduced egg number, reduced survival, defective shells) of 4.9 mg/kg/day in
domestic chickens. Heinz observed reproductive and behavioral effects in a three observation study in
mallards; the LOAEL of 0.064 mg/kg-day from this study served as the basis for the avian wildlife
criterion.
3.3 Dose-Response to Methylmercury for Wildlife Species
Despite major gaps in information on methylmercury toxicity to wildlife, adverse effects of
mercury exposure have been evaluated for selected species. The species identified for risk
characterization were wildlife chosen because they are predicted to be the most highly exposed to
methylmercury, not because of any known inherent sensitivity to methylmercury toxicity. Lacking
additional toxicity information, little guidance is available concerning which wildlife species are likely
to be the most sensitive to methylmercury.
The adverse health outcomes considered in risk characterization for wildlife species are
neurotoxicity and reproductive and/or developmental effects. The databases for wildlife species are
extremely limited. A major uncertainty in this risk characterization is cross-species extrapolation.
Compared with endpoints for humans, the wildlife effects are far more severe. Wildlife effects are
either lethality or gross clinical poisoning.
3.3.1 Mammalian Species
Dose-response to methylmercury has been estimated in greater detail for mink than for other
mammalian wildlife species. The data of Wobesser (1973) and Wobeser et al. (1976a,b, 1979) on
mink (Mustela vison) have been used to estimate risk for mammalian wildlife species. The LOAEL
for mink based on nervous system lesions identified through histopathology was 1.1 ppm
methylmercury in diet (0.24 mg/kg bw/day in males and 0.16 mg/kg bw/day in females).
The most complete evaluation of quantities of methylmercury in fish fed to wildlife species are
those of Wobeser et al. (1976a). Wobeser et al. (1976a) studied the effects of dietary consumption of
mercury in the form of contaminated fish, caught locally and mixed with the normal mink chow to
produce an average dose of 0.33 ppm methylmercury in the diet. Materials and methods are described
in the original publications and in Volume V. Wobeser et al. (1976b) fed minks diets containing 1.1,
1.8, 4.8, 8.3 and 15 ppm total mercury from contaminated fish in the mink chow. A LOAEL was
identified for mink as 1.1 ppm (0.24 mg/kg body weight/day in males and 0.16 mg/kg bw/day in
females) based on a finding of nervous system histopathology without clinical signs. Higher doses of
methylmercury caused anorexia, ataxia, and death within 60 to 80 days at 1.8 ppm; death occurred
within 26 to 36 days at 4.8 ppm, and death resulted within 19 to 26 days at 8.3 ppm (Wobeser,
1976b). From these studies (Wobeser, 1976a and 1976b) a NOAEL of 0.05 mg/kg bw/day and a
LOAEL for nervous tissue lesions of 0.16 mg/kg bw/day based on male mink is identified.
June 1996 3-3 SAB REVIEW DRAFT
-------
Under the study conditions (Wobeser, 1973; Wobeser et al., 1976a,b), exposure to 1.1 ppm
methylmercury in the diet for 93 days produced histopathological changes in the central nervous
system (specifically neuronal necrosis) but no clinical signs and symptoms of methylmercury
poisoning. When the dietary level of methylmercury was increased by 0.7 ug/g (from 1.1 to 1.8 ug/g
diet) the mink became anorexic, developed ataxia (severe incoordination) and died with clinical signs
and symptoms of methylmercury poisoning (Wobesser, 1973). These changes occurred after 50 to 60
days of exposure (for anorexia) and 60 to 78 days of exposure (for ataxia) to the 1.8 ppm diet. The
increase of 0.7 ug/g dietary methylmercury resulted in appearance of signs and symptoms of
methylmercury poisoning and death. Ingestion of diet with higher concentrations of methylmercury
(4.8 and 8.3 ug/g diet) decreased the survival time of mink to 26-36 and 19-26 days, respectively
(Wobeser, 1973). All mink dying after receiving higher concentrations of dietary methylmercury
(Wobeser, 1973) had gross clinical symptoms of methyhnercury poisoning including anorexia and
ataxia. Overall, the dose-response curve for methylmercury was very steep as reflected by the onset of
mercury-related signs and symptoms of severe poisoning if methyhnercury were increased from 1.1 to
1.8 ug/g diet Higher concentrations of dietary methylmercury decreased the time that mink survived
on these diets.
In the studies of Wobeser et al., methyhnercury had been incorporated into the diet on a
weight/weight basis yielding study groups for which methyhnercury concentrations were expressed on
a parts per million (ppm or ug/g) basis. Assumptions needed to interpret these data were body weight
of the mink and food consumption by mink; these were obtained from the Exposure Factors Handbook
(U.S. EPA, 1994).
An additional uncertainty is whether or not the dose-response pattern for mink will apply to
river otters. Application of mink dose-response data to otters is supported by the observations of
O'Connor and Nielson (1981). These investigators identified dietary levels of 2.0 ug/g methylmercury
as acutely toxic to otter.
3.3.2 Avian Species
The most comprehensive studies of adverse reproductive effects of methylmercury exposure to
avian species are based on the research of Heinz (1974, 1975, 1976a,b, 1979). Initially, Heinz (1974)
identified a NOAEL of 0.5 ppm based on reproductive effects in a 21-week duration study. In a later
study, reproduction hi the first and second generation ducks was evaluated (Heinz 1976a,b) and the
NOAEL for the first generation was again determined to be 0.5 ppm. The second generation,
however, suffered adverse reproductive effects that included the following: eggs laid outside nest box
at the 0.5 ppm dose. Consequently, the LOAEL for reproductive effects for the second generation was
0.5 ppm with no NOAEL identified. A third generation of mallards also demonstrated adverse
reproductive effects at 0.5 ppm mercury in the diet. Effects observed included reduced numbers of
sound eggs laid per day and thinner egg shells.
Heinz (1975, 1976a,b, 1979) also examined behavioral effects of mercury exposure in the
approach response of chicks to maternal calls and avoidance of frightening stimuli. In third-generation
ducklings there was a reduction hi the response rate and speed of response to maternal calls. When
data were pooled from all studies and subjected to analysis of variance with multiple comparisons,
alterations of behavior were observed in the lowest dose groups in all generations. These alterations
included the following: reduction in the number of ducklings that approached maternal calls, and an
increase in the distance traveled to avoid a threatening stimulus. In summary, no NOAEL could be
June 1996 3-4 SAB REVIEW DRAFT
-------
demonstrated for behavioral effects, and the NOAEL for reproductive effects could only be
demonstrated for the first generation.
For the determination of the appropriate selection of the LOAEL, consideration was given to
using 3 ppm in the first generation with 0.5 ppm NOAEL. It was concluded, however, that effects
observed in subsequent generations at 0.5 ppm should not be discounted. It seems likely that the
effects observed in the second and third generations were a result of the earlier onset of dosing (adult
onset vs. onset as ducklings). For this reason, 0.5 ppm was selected as a LOAEL for mallard ducks.
Assuming a feeding rate of 128 g/kg/day for adult mallards, the LOAEL for reproduction and behavior
is 0.064 mg/kg/bw day. Dose-response data for the mallard has not been estimated beyond the
LOAEL. An additional uncertainty in this assessment is the extent that the dose-response data for
mallards apply to osprey, kingfishers, bald eagles and other piscivorous birds.
3.4 Wildlife Criteria \
3.4.1 Wildlife Criteria Methodology
Calculation of wildlife criteria (WC) for mercury was based upon the use of a wildlife
reference dose approach, combined with knowledge of the extent to which mercury becomes
concentrated in aquatic food chains. The methods used to calculate these values were based on those
described in the Proposed Great Lakes Water Quality Guidance for the Great Lakes Water Quality
Initiative (henceforth referred to as the "Proposed Guidance," U.S. EPA, 1995b). This approach yields
a measurement endpoint, which is the total mercury concentration in water that is believed to be
protective of piscivorous wildlife. A similar equation was first used by the State of Wisconsin to set
Wild and Domestic Animal Criteria (State of Wisconsin, 1989). The entire approach, including both
the equation and data requirements for its parameterization, was later modified by U.S. EPA for
incorporation into the Proposed Guidance (U.S. EPA, 1993c) and Final Guidance (U.S. EPA, 1995b).
Description of the calculation of the wildlife criteria, description of terms of the equation and support
for the values used are found in Chapter 4 of Volume V.
The WC for mercury is defined as the concentration of total mercury in surface water that, if
not exceeded, protects both avian and mammalian wildlife that use the water as a drinking or foraging
source. Thus, the WC is the highest aqueous concentration of mercury that causes no significant
reduction in growth, reproduction, viability or usefulness (in a commercial or recreational sense) of a
population of animals exposed over multiple generations. For the purpose of this analysis, the term
"aqueous concentration" refers to the total concentration of all mercury species in filtered water,
including both freely dissolved forms and mercury that is associated with dissolved organic material.
It is recognized that methylmercury is the form of mercury that bioaccumulates in fish.
The method, in its current form, was reviewed in 1992 at a workshop entitled the National
Wildlife Criteria Methodologies Meeting, sponsored by U.S. EPA (1994). Subsequently, it was used
to develop interim Tier I WC for four compounds (PCBs, DDT, dieldrin and mercury) in the Great
Lakes Basin (U.S. EPA, 1993b). These criteria have received public comment. The method has been
reviewed by the Science Advisory Board (SAB) on two occasions, most recently in June of 1994.
Detailed descriptions of the method, including comparisons with other proposed methods for setting
wildlife criterion values, are presented in U.S. EPA (1994, 1995b).
The equation used in Volume V to calculate wildlife criteria for mercury is that described in
the Proposed Guidance to the GLWQI (U.S. EPA, 1995b):
June 1996 . 3-5 SAB REVIEW DRAFT
-------
(TDx [1IUF]) X Wt
"A
WA H- [(FDJ(FA x BAFJ
where:
WC = wildlife criterion value (pg/L; after converting from ug/L)
TD = tested dose (ug/g bw/day)
UF = uncertainty factor
WtA = average species weight (g)
WA = average dally volume of water consumed (L/d)
FD3 = fraction of the diet derived from trophic level 3
FA = average daily amount of food consumed (g/d)
FD4 = fraction of the diet derived from trophic level 4
BAF3 = aquatic life bioaccumulation factor for trophic level 3 (L/g; methylmercury
concentration in fish/total mercury in water)
BAF4 = aquatic life bioaccumulation factor for trophic level 4 (L/g; methylmercury
concentration in fish/total mercury hi water)
In me equation used in this Report the term F (defined in the GLWQI as the food ingestion
rate of prey for a tropic level) was broken into the terms FA and FD above. The UF considers
uncertainty in three areas described below.
This equation encompasses both a hazard and an exposure component. The GLWQI equation
includes a term TD for "tested dose". In this Report, data were reviewed to ascertain an appropriate
NOAEL, which was used for the TD. In the absence of a NOAEL, a LOAEL was used with the
addition of an appropriate uncertainty factor (UFL) to indicate uncertainty around the toxic threshold.
An uncertainty factor (UFA) may be used to provide a margin of safety when applying data from a
species other than the species of concern. A third uncertainty factor (UFS) may be used to extrapolate
from subchronic to chronic exposures. Collectively, the application of the UF to the TD results in the
estimation of a "reference dose" for subsequent calculation of WC.
3.4.2 Bioaccumulation Factors
Bioaccumulation factors (BAFs) for trophic levels 3 and 4 (forage fish and larger, piscivorous
fish, respectively) were estimated hi Appendix A to Volume V. The BAF for any given trophic level
is defined as the ratio of the total mercury concentration hi fish flesh divided by the concentration of
total dissolved mercury hi the water column. The BAF represents the accumulation of mercury hi fish
of a specific trophic level from both water intake and predation on contaminated organisms.
In Volume V BAFs were estimated for trophic level 3 (foraging fish) and trophic level 4
(piscivorous fish) designated as BAF3 and BAF4, respectively. BAF3 was estimated by two different
methods: the method of the GLWQI which combined use of bioconcentration factors for lower
trophic levels with predator-prey factors; and calculation of a BAF from field measurement data.
June 1996 3-6 SAB REVIEW DRAFT
-------
BAF4 was calculated by three methods: the GLWQI method: calculations based on field measurement
data; and multiplication of the field-data-based BAF3 by a predator-prey factor for trophic level 4.
Results of Monte Carlo simulations for each of the methods are given in Table 3-1. The equations
used in the Monte Carlo analyses are given in Section 5.4.1.2 of Volume V.
Table 3-1
Summary of Bioaccumulation Factors for Trophic Levels 3 and 4
(mean, 5 Percent, and 95 Percent values)
Recommended
\lethod
Geometric Mean
5th pctl
95th pctl
GSD*
BAF3
66,200
Field-Derived
66,200
6,400
684,000
4.14
GLWQI
25,200
2,310
308,000 .
4.41
BAF4
BAF3x
PPF4b
335,000
22,700
4,700,000
5.05
335,000
Field-Derived
400,000
23,600
6,780,000
5.59
GLWQI
136,000
8,760
2,070,000
5.25
* Geometric Standard Deviation
b Predator Prey Factor
The selection of the BAFjX^PF^ as the recommended approach was based on several
considerations. Although the mean values of all three BAF4 simulations agree within a factor of three,
the GLWQI results stand somewhat apart. The mean value of 136,000 for the GLWQI method falls at
the 30th and 35th percentile of the BAF4 distributions for the field-derived and BAF x PPF methods,
respectively. The GLWQI method is also more complex with more variables and assumptions than the
other two approaches. The BAF x PPF and field-derived methods represent a consolidation of earlier
stages of the GLWQI method and should give more accurate results than the GLWQI method provided
that the data defining the distributions are at least as good as the data defining the GLWQI' variables.
Five studies are available for defining BAF3; however, three of the critical variables in the GLWQI
method are based on only one or two studies. In addition the field measurements for BAF3 and PPF4
apply directly to variables, while the bioconcentration factor (BCFs) in the GLWQI approach do not.
That is, the measurements are taken directly from fish at the appropriate trophic levels for BAF3 and
PPF4; BCFmHg (the BCF for methylmercury) and BCF^g (the BCF for inorganic mercury) apply to
phytoplankton (trophic level 1) but are estimated from measurements in trophic level 3 fish. The
BAF3 x PPF4 approach is also less variable than either of the other two methods, as indicated by the
geometric standard deviation. Uncertainty and variability in the BAF are described in Sections 3.4.6.3,
3.4.6.8 and 3.4.6.12.
3.4.3 Other Exposure Parameters
Exposure parameters for the present analysis are shown in Table 3-2. The scientific basis for
these parameters is reviewed elsewhere (U.S. EPA, 1993a, 1995a,b). For this analysis, it was assumed
that prey not attributed to trophic levels 3 and 4 were derived from non-aquatic origins and do not
June 1996
3-7
SAB REVIEW DRAFT
-------
contain mercury. Were these prey to contain mercury, WC values calculated for the relevant species
would decrease.
Table 3-2
Exposure Parameters for Mink, Otter, Kingfisher, Osprey, and Eagle
Species
Mink
Otter
Kingfisher
Osprey
Eagle
Body WL
(WtA)
kg
0.80
7.40
0.15
1.50
4.60
Ingestion Rate
(FA)
kg/day
0.178
1.220
0.075
0.300
0.500
Drinking Rate
(WA)
L/day
0.081
0.600
0.017
0.077
0.160
Trophic Level
of Wildlife
Food Source
3
3,4
3
3
3,4
Percent
Diet at
Each
Trophic
Level
90
80,20
100
100
74,18
3.4.4 Health Endooint (TD)
Based on the information hi Sections 3.3.1 and 3.3.2, the TDs used for calculation of a WC
for mercury were these:
For avian wildlife - A LOAEL of 64 ug/kg bw/day
For mammalian wildlife - A NOAEL of 55 ng/kg bw/day
3.4.5 Calculation of Wildlife Criterion Values
WC values were calculated for each of the wildlife species of concern using exposure
parameters values recommended hi previous sections. UFAs were employed as recommended in the
GLWQI to extrapolate from test species to the species of interest. Because the mammalian TD
(NOAEL) was derived from studies with mink, the UFA for species extrapolation of the mink WC was
set equal to 1.0. Otter were considered sufficiently similar to mink so that a UFA of 1 was also
considered appropriate. A UFS of 10 (for a subchronic study) was applied. UFA of 3 was used for
extrapolation of mallard data to the kingfisher, eagle and osprey. A UFL of 3 was employed for use
of a LOAEL hi the absence of a NOAEL. Calculations of WC values for each of the selected species
follow.
June 1996
3-8
SAB REVIEW DRAFT
-------
For the mink:
(TD x [1/(UFA x UFS x UFj)]) x WtA
WCS = —
WA^[(0.9)(FAxBAF3)]
wc _ (0.055 mglkgld x [1/(1 x 10 x 1)] x 0.8 Jig
5 ~ 0.081 Lid + [(0.9) (0.178 kg/d x 66,200)]
WCS
For the otter:
(TD x [1J(UFA x UFS x UFj)] x WtA
YfCs "
WA + [(0.8) (FA x BAFJ + (0.2) (FA x RAFj ]
wc . (0.055 mglkgld x [11(1 xlOx 1)] x 1.4kg
5 0.60 Lfd + [(0.8) (1.22 J^/d x 66^00) + (0.2) (1.22 Ag/J x 335,000)]
- 278
For the kingfisher:
WC
s
(W x [1I(UFA x UFS x UFj)] x WtA
WA * [(1.0) (FAxBAF3)]
WC = (0.064 mglkgfd x [l/(3 x 1 x 3)] x 0.15
5 ~ 0.017 + [(1.0) (0.075 x 66,200)]
WCS = 193 pg/L
June 1996 3-9 SAB REVIEW DRAFT
-------
For the osprey:
we = (TD x [II(I/FA x UFs x UFl)] x Wt*
5 WA + [(1.0) (FA x BAFJ]
wc m (0.064 mgfkg/d x [11(3 x 1 x 3)] x 1J kg
5 ~ 0.077 Lid + [(1.0) (0.3 kg/d x 66,200)]
WCS = 483 pg/L
For the bald eagle:
WCS
(TD x [lf(UFA x UFS x UFjj] x WtA
WA + [(0.74) (FA x BAFJ + (0.18) (FA x BAF4]
wc _ (0.064 mglkgld x (11(3 x 1 x 3)] x 4.6 kg
5 0.16 Lid + [ (0.74) (0.5 kgfd x 66,200) + (0.18) (0.5 *g/J x 335,000)]
WCS - 538
The geometric mean of the two WCS values calculated for mammals is 346 pg/L. The
geometric mean of the three avian values is 405 pg/L. The lowest of these is the WC calculated for
avian species; therefore, the WC for mercury is 346 pg/L.
The evaluation of data and calculation of WC in this Report was done in accordance with the
methods and assessments published in the Final Water Quality Guidance for the Great Lakes System:
Final Rule (U.S. EPA, 1995). Availability of additional data led to differences in calculated values of
the WC in this Report and those published in the final rule. Differences were the result of three
factors. The Report uses more recent data to derive BAF. Second, the final rule appropriately used
some region-specific assumptions that were not used in the nationwide assessment in the Report; for
example, consumption of herring gulls by eagles. Finally different endpoints were used for the
evaluation of mammals because the purposes of the assessments in the Report and final rule were
different. In the final rule, a risk-management decision was made to base the wildlife criterion on
endpoints likely to influence whole populations (mortality, growth). In this Report a more sensitive
endpoint was selected with the goal of assessing the full range of effects of mercury. The difference
in the results reflects the amount of discretion allowed under Agency Risk Assessment Guidelines.
June 1996 3-10 SAB REVIEW DRAFT
-------
3.5 Uncertainty Around the Dose-Response Assessments for Methylmercury
3.5.1 Uncertainty in the Wildlife Criteria
The species selected for the present analysis were chosen because they are the most exposed to
methylmercury, not because of any known inherent sensitivity to methylmercury. Lacking toxicity
information, little guidance is available concerning which wildlife species are likely to be the most
sensitive to mercury.
A formal analysis of uncertainty around the wildlife criteria estimate was not attempted. Such
an analysis would require specification of numeric distributions for each of the parameters in the
equation. While theoretically possible, this approach is of questionable value because the overall
analysis is intended to be protective of that subset of each species that feeds extensively at the top of
aquatic food chains. Thus, incorporation of data reflecting the range of dietary items upon which the
bald eagle feeds would tend to generate an extremely broad range of wildlife criteria values for this
species. In addition, data for several of the parameters in the equation, in particular the NOAEL and
UF estimates, are presently sufficient only to generate point estimates.
A restricted uncertainty analysis involving only incorporation of numeric distributions for each
of the B AF estimates could be accomplished using existing data, but would probably not be useful.
B AF distributions generated by Monte Carlo analysis of field data are thought to reflect real, naturally-
derived variation in mercury bioaccumulation and biomagnification. Despite the relative abundance of
such data, BAFs expressed on a total mercury basis remain difficult to interpret. Because
methylmercury is the form of mercury accumulating in fish, wildlife criteria distributions based on the
distribution of methylmercury BAFs are more likely to yield information of value to risk assessors.
3.5.2 Sensitivity Analysis
In a sensitivity analysis, an attempt is made to characterize the extent to which a calculated
value changes with the parameters upon which it depends. Examination of the equation for calculation
of wildlife criteria values suggests that a proportional relationship exists between the wildlife criterion
and the NOAEL, UF or average species weight (WtA). The relationships between the wildlife criteria
and parameters that appear hi the denominator are not as apparent and must be explored by varying
these parameters one-by-one in systematic fashion. The analysis is also complicated by the variable
relationship that exists between fraction of diet from trophic level'3 (FD3) and fraction of diet from
trophic level 4 (FD^). In the otter and eagle, FD3 and FD4 tend to be reciprocal (although in the eagle
these values do not add up to 1.0). In the mink, however, FD3 is assigned a value of less than 1.0 and
the remainder of the diet is assumed to consist of prey that are not aquatic in origin and are not
contaminated with mercury.
General conclusions can be reached regarding the sensitivity of wildlife criteria estimates to
changes hi these parameters. These can be described as follows:
(1) A decrease in any parameter that appears in the denominator will have a larger effect
on the wildlife criterion than an equivalent percentage-wise increase.
(2) When BAF3 appears alone hi the denominator, a percentage-wise increase in BAF3 or
FD3 will cause a less than proportional decrease hi the wildlife criterion; conversely a
June 1996 3-11 SAB REVIEW DRAFT
-------
decrease in BAF3 or FD3 will cause a greater than proportional increase in the wildlife
criterion.
(3) When both BAF3 and BAF4 appear in the denominator, an equivalent percentage-wise
change in BAF4 (and by extension PPF^ has a greater impact on the wildlife criterion
than a change in BAF3, but in either case the effect is less than proportional.
(4) If BAF3 and BAF4 are both allowed to change (holding PPF4 constant), a percentage-
wise increase in BAF3 (and by extension BAF^ will have a less than proportional
effect on wildlife criterion, while a decrease in BAF3 will have a greater than
proportional impact.
(5) Under all circumstances, a percentage-wise increase in FA will cause a less than
proportional decrease in wildlife criterion, while a decrease in FA will cause a greater
than proportional increase in wildlife criterion.
(6) - Owing to its small contribution to the analysis as a whole, large changes in WA have a
very small impact on the wildlife criterion.
With the exception of FA, it is not possible to conclude that for all species the wildlife
criterion is most sensitive to one or the other of the parameters in the denominator of the equation.
For species that feed at one trophic level, all parameters other than FA have the potential to change
wildlife criterion in a proportional or greater than proportional manner. For species that feed at two
trophic levels, the BAF at the lower trophic level becomes relatively less important, but it may still
have a large impact on wildlife criterion if the percentage of the diet represented by this lower trophic
level is large (e.g., in the mink).
3.5.3 Uncertainties Associated with the GLWOI Methodology
Efforts to develop wildlife criteria for the protection of piscivorous wildlife are relatively
recent in origin, and the methods employed for this purpose continue to undergo modification and
refinement Owing to the complexity of natural systems, uncertainties associated with the development
of wildlife criteria are to be expected. Additional uncertainties derive from the relative scarcity of
wildlife toxicity information and the necessity of extrapolating individual-based effects to higher levels
of biological organization (e.g., populations).
Uncertainties associated with the GLWQI methodology have been reviewed elsewhere (U.S.
EPA, 1994). Those areas that are especially pertinent to the development of a wildlife criterion for
mercury are described below.
3.5.3.1 Limitations of the Toxicity Database
Substantial uncertainties underlie most of the toxicity data for mercury in wildlife.
Comparison of NOAELs and LOAELs between species requires adoption of unproved assumptions
about the uptake, distribution, elimination and toxic effects of mercury. Additional uncertainties derive
from the necessity of extrapolating from LOAELs to NOAELs, and extrapolating from subchronic
endpoints to chronic endpoints. In some instances there may also be a need to account for the
possibility that test results do not adequately protect the most sensitive individuals. This may be
June 1996 3-12 SAB REVIEW DRAFT
-------
particularly germane to the case of the Florida panther, wherein there is concern for individual
animals.
Existing data on wildlife populations are complicated by the possibility that "naturally
incorporated" mercury is accompanied by other contaminants that are exerting some or all of the
observed effect Ideally, it is desirable to compare the effects of mercury that has been incorporated
naturally with effects due to mercury that has been added to a prepared diet. By adding mercury into
the diet, the researcher can better control the dose to the animal. The bioavailability of mercury in
such a formulation may be very different from that which exists naturally. Charbonneau et al. (1976),
however, have demonstrated that the bioavailability and toxicity of methylmercury to cats is equivalent
whether given in contaminated fish or added in the diet
Despite a lack of toxicity information and problems concerning its interpretation, NOAELs
estimated for piscivorous birds and mammals are very similar (55 ug/kg for mammals versus 21 ug/kg
for birds). Moreover, the existence of toxicity information for the mink eliminates the need to
incorporate additional uncertainty factors into the analysis. Unfortunately, similar data for piscivorous
birds do not exist.
All wildlife species of interest cannot be tested. The use of uncertainty factors for species
extrapolation is likely, therefore, to continue. Existing information can be used, however, to suggest
which species should be singled out for testing. Properly applied, species sensitivity factors probably
represent a relatively small source of error in the calculation of wildlife criteria values.
Finally, comparisons between wildlife and human NOAELs are complicated by differences in
the ability of a given study to reveal an adverse effect when it occurs. For wildlife, most of the
endpoints selected can be considered severely adverse or frank effects. Very few studies to date have
been designed to study subtle adverse effects or precursors to adverse effects in wildlife.
Developmental neurotoxicity endpoints are of particular interest because of their demonstrated
sensitivity hi humans. The question arises, therefore: what would the LOAEL or NOAEL for a given
wildlife species have been had the researcher been looking for (or was able to detect) these more
subtle effects? One possible approach to this question is to examine the results of studies in which
both frank and more subtle effects were observed and determine the corresponding difference between
dosage levels.
The available data suggest that uncertainty factors presently employed to derive chronic
reference doses for birds and mammals are unlikely to be greatly underprotective or overprotective,
and are, therefore, reasonable.
3.5.3.2 LOAEL-to-NOAEL Uncertainty Factor UFL
In determining the wildlife criteria for mercury exposure hi wildlife, a NOAEL is preferred as
the value to be used as the term TD. The wildlife criterion can be considered a wildlife WC with an
adjustment for exposure assumptions. As is the case for human health RfDs, the wildlife WC is an
attempt to estimate a threshold dose for adverse effects and then to determine a level below that
threshold dose. It is assumed that daily consumption of an amount of material below the threshold for
adverse effects should be without ill effect.
June 1996 3-13 SAB REVIEW DRAFT
-------
In cases in which studies do not identify a NOAEL, the data are examined to identify a
LOAEL to be used in estimating the WC. A'UFL of 3 or 10 (based on U.S. EPA Reference Dose
methodology) is typically applied when a LOAEL is used in the absence of a NOAEL.
In determining the RfD for human exposure to methylmercury, a large number of laboratory
animal studies on methylmercury toxicity were summarized as supporting data. Results from many of
those studies permitted estimation of both a LOAEL and a NOAEL. Those studies were examined in
an effort to determine the most appropriate UFL for wildlife exposure to mercury. Details of this
analysis can be found in Section 4.5.2 of Volume V.
The ratios of LOAEL to NOAELs for laboratory animal studies were plotted versus frequency.
These ratios can be thought of as the reduction in the LOAEL necessary to estimate the corresponding
NOAEL. The majority of ratios lie between one and two (n=6) and between four and five (n=9).
Only one ratio of the 19 was greater than 10. A ratio of five indicates that the NOAEL observed
following exposure to methylmercury is five-fold less than the corresponding LOAEL. These data
imply that most ratios between LOAELs and their corresponding NOAELs will be less than 10.
A similar analysis of animal toxicity data (Weil and McCollister, 1963) was'provided by
Dourson and Stara (1983). None of the LOAEL-to-NOAEL ratios from studies of 52 chemical
substances exceeded 10. Only two of the 52 ratios exceeded five. The Dourson and Stara (1983)
analysis has been cited in support of the use of a variable UFL of as much as 10 in deriving reference
doses for humans. Dourson and Stara (1983) recommended the application of a relatively large UFL
when estimating a NOAEL from a LOAEL for a severe or frank toxicological effect. Conversely, a
low UF could be applied when the toxicological effect was considered to be relatively mild.
The UFL of three was selected by the authors of this Report for use with the avian LOAEL
from the same data (Heinz, 1975, 1976, 1979) as a reasonable compromise between UF ratios of two
and five.
3.5.3.3 Validity of BCF/BAF Paradigm
The wildlife criteria for mercury calculated in the GLWQI relied on bioconcentration factors
(BCF) values determined from laboratory data. This methodology is based on a bioaccumulation
paradigm (steady-state BCF x FCM) that was developed for neutral hydrophobic organic compounds
and that may be inappropriate for application to mercury.
Field studies indicate that many fish, if not most, accumulate mercury throughout their lives,
often in a nearly linear fashion with age. Moreover, most of the mercury accumulated by fish at
trophic levels 3 and 4 is thought to be taken up from dietary sources. Thus, particularly for long-lived
piscivorous fish, a relatively short (one year or less) waterborne exposure cannot duplicate the extent
of accumulation that takes place in nature. In addition, the relationship between a concentration of an
applied mercury species hi the laboratory and the concentrations of multiple species present hi the
environment (some of which may not be bioavailable) is completely unknown.
The apparent progress to "steady-state" observed in several chronic laboratory studies (see
McKim et al., 1976) should not be misinterpreted as an actual steady-state condition, but instead
probably reflects growth dilution with rapidly growing fish. Such growth dilution will tend to depress
BCF values during the period of rapid growth, but as the growth rate slows continued accumulation of
mercury would result hi an increase hi whole-body concentration with age.
June 1996 3-14 SAB REVIEW DRAFT
-------
BAF values for trophic levels 3 and 4 were estimated using field residue data. Uncertainties
remain, however, with respect to the naturally-derived variability that exists around these estimates. A
growing body of information suggests that much of this variability can be attributed to site-specific
faaors that control the net rate of mercury methylation, but numerous additional factors may influence
the extent to which mercury accumulates and biomagnifies in aquatic food webs.
3.5.3.4 Selection of Species of Concern
The species identified for the present analysis were selected because they were considered
likely to be exposed not because of their inherent sensitivity to mercury. Lacking toxicity information,
little guidance is available concerning which wildlife species are most sensitive to mercury. In
addition, there are problems associated with any comparison of laboratory and field data. Such
comparisons, however, are complicated by the presence/absence of additional stressors such as
confinement, handling, and weather, differences between natural and prepared diets, and the interplay
between "inherited" (egg) residues and that which the chick consumes. Toxicity can be difficult to
observe in a field study, even when it is occurring, due to any number of factors.
Exposure and sensitivity are related. If a species was 10 times more sensitive than the eagle
on a delivered dose basis, but because of its dietary habits received less than 10 percent of the dose, it
would not be expected to show adverse effects at water concentrations protective of the eagle.
Pharmacokinetic considerations may also be important. Thus, it has been suggested that birds
eliminate a substantial amount of mercury through incorporation into plumage. The frequency and
extent to which birds molt may, therefore, impact their apparent sensitivity in an environmental setting.
It has also been suggested that some birds and mammals demethylate, or otherwise eliminate mercury
by some route other than in hair or plumage (Wren et al., 1976). Enhanced elimination would be
particularly important if it represented an adaptive strategy for piscivorous species. The need for
toxicity information has already been noted. As such information becomes available it may be
necessary to revise the wildlife criteria for mercury.
There is a need also to consider animals other than birds and mammals. In particular, there is
a need to characterize the exposure of carnivorous reptiles such as the alligator, which is known to
consume considerable quantities of fish and which also feeds on piscivorous animals (e.g., raccoon)
that are known to accumulate mercury (Roelke et al., 1991).
3.5.3.5 Trophic Levels at Which Wildlife Feed
It can be expected that representatives of the same species will be exposed to different levels
of mercury due to different feeding habits and/or differences in the availability of specific prey items.
For example, bald eagles living on the shores of the Great Lakes may consume significant numbers of
herring gulls (Kozie and Anderson, 1991). Since the gulls themselves are piscivores, feeding primarily
at trophic level 3, it has been argued that when an eagle consumes a gull it is feeding at trophic level
4 or higher. Eagles living in other parts of the country, or migrating into an area during a particular
time of year, may consume relatively few fish, feeding instead on carrion, including rabbits, squirrels,
and dead domestic livestock such as pigs and chickens (Harper et al., 1988). Other populations,
however, are critically dependent upon the seasonal availability of fish, particularly spawning
salmonids.
0
For some species, such as the kingfisher and river otter, it can be reasonably assumed that
fish always comprise a high percentage of the diet. For others, such as the eagle and mink,
June 1996 3-15 SAB REVIEW DRAFT
-------
considerable variations in diet are likely to exist. Still others, such as the Florida panther, consume
prey (e.g., the raccoon) that, as a species, consume variable amounts of aquatic biota, but are thought
to represent a close link to the aquatic food chain in South Florida.
3.5.3.6 Variability in BAFs at Each Trophic Level
A concern related to the issue of feeding preference is the possibility that trophic levels
presently assigned to the wildlife species in this analysis overestimate the actual extent to which they
are exposed to mercury. This is because BAFs are developed to represent the average value for a
trophic level, when in fact piscivorous birds and mammals are more likely to target prey at the lower
end of the size (age) distribution. Thus, eagles are more likely to consume a 1 kg northern pike than a
10 kg individual, yet both are represented in the BAF for trophic level 4. Similarly, kingfishers are
probably limited to smaller representatives of trophic level 3 than would be true of an osprey. The
reason that these differences are important is that mercury tends to accumulate throughout the life of
an individual fish with the result that concentrations in an older individual at a given trophic level may
far exceed those hi a younger individual.
3.5.3.7 Natural History Considerations
Natural exposures are likely to vary in both spatial and temporal domains. This is particularly
true of species that migrate, including the bald eagle, osprey and belted kingfisher. The necessity of
incorporating this type of information and the means by which this can be accomplished are open
questions.
3.5.3.8 Individuals Versus Populations
The methods used to develop wildlife criteria for mercury are based on effects data from
individual organisms. The stated assessment endpoint for this analysis, however, is the health of
wildlife populations. The relationship between individuals and populations is likely to vary with the
species and a large number of environmental factors (e.g., availability of food in a given breeding
season). For a given population, the loss of a significant number of individuals may have little effect,
particularly if environmental factors (like carrying capacity) limit population size. For other
populations, in particular those with low fecundity, loss of a relatively few individuals could have a
large impact. Clearly, there is a need to be able to extrapolate toxic effects on individuals to effects
on populations. Unfortunately, this type of analysis is complicated by numerous factors (such as
relationship of one population to other populations) and is essentially impossible to apply on a national
scale.
Finally, a focus on populations may not always be appropriate, particularly when endangered
species are Involved. The same may also be true when various factors contribute to the possibility of
regional effects. For example, 95 percent of eagles nationwide might be protected by a wildlife
criterion for avian species, but in a given region mortality could approach 100 percent if low pH of
surface waters contribute to higher than average accumulation of mercury in the aquatic food chain.
3.5.3.9 Species Versus Taxa
The wildlife criterion developed for mercury in birds was calculated as the geometric mean of
values for three species. Similarly, the geometric mean of values for two species was used to
represent all mammals. This approach is reasonable if the wildlife criteria calculated for each species
June 1996 3-16 SAB REVIEW DRAFT
-------
within a taxa are similar, but would fail to protect species for which the wildlife criterion value is
much lower than the others with which it was averaged. In the latter case, averaging would effectively
lead to protection to < 100 percent of all species.
In the present analysis, wildlife criteria values calculated for eagles, osprey and kingfisher
were within a factor of three of one another. Wildlife criteria values for mink and otter agreed within
a factor of about two. As additional data are gathered, there is a need to identify species that, by
virtue of sensitivity and/or exposure, are particularly vulnerable to mercury. Decisions could then be
made concerning the advisability of special measures to ensure their protection.
3.5.3.10 Discussion of Uncertainties Associated with the GLWQI Methodology
The existing limited data suggest that BAF values represent the most important source of
uncertainty in present efforts to calculate water-based wildlife criteria values, although a lack of
toxicity information and incomplete knowledge of what wildlife eat also contribute substantially to
uncertainty. Considerable progress has been made in understanding and predicting how lake water
characteristics (e.g., pH, temperature, dissolved organic carbon) affect methylation rates, and in time it
may be possible to adjust BAF predictions as needed to represent surface waters of concern. The
prospect for continuing uncertainty surrounding these estimates argues, however, for adoption of a
residue-based approach, that is, the use of measured mercury residues in fish and wildlife to identify
populations at risk.
It is important to recognize that BAF values are calculated as the ratio of a tissue
concentration and a water concentration. Emphasis has been placed on problems associated with
obtaining the numerator in this equation. Considerable uncertainty, however, also exists with respect
to the denominator. In several instances it has been shown that with improved analytical methods,
mercury levels in a given water body tend to come "down", resulting in an increase in the apparent
BAF. This "decline" is usually not thought to be real but instead reflects improvements in sampling
technique and analytical methods.
It is also ^unclear which of the mercury species are bioaccumulative and should, therefore,
appear in the denominator. Presently, the denominator in most studies consists of total amount of
mercury in filtered water. It is more likely that there may be multiple "pools" of mercury, each of
which is bioavailable to varying degrees. In this regard it is important to realize that even in highly
polluted systems >99 percent of all methylmercury is complexed, either hi biomass, or with dissolved
organic material, paniculate material and sediments.
An effort was made to treat the uncertainty in BAF estimates by using a Monte Carlo
simulation approach. The advantage of this approach is that it explicitly treats known variation in
these parameters thereby providing for the statistical possibility of a high or low end result In
addition, the distributions themselves follow from the processes at work. As more information about
mercury is obtained, the distributions can be unproved. One example of this relationship has already
been discussed, namely, the fact that a skewed BAF distribution for trophic level 4 would tend to
follow from random sampling of a fish population due to the relative scarcity of the oldest individuals.
June 1996 3-17 SAB REVIEW DRAFT
-------
4. CHARACTERIZATION OF MERCURY EXPOSURE OF SELECTED HUMAN AND
WILDLIFE POPULATIONS
Exposure of human or wildlife populations to methylmercury may be estimated from the
results of modeling or from interpretation of survey and monitoring data. In this Report both
approaches have been utilized. Modeling permitted estimation of environmental mercury
concentrations and exposures that resulted from anthropogenic sources. Use of dietary survey data
combined with measured mercury concentrations in fish were used to estimate exposures to
methylmercury to the general United States population and selected subpopulations. For piscivorous
wildlife, data on mercury concentrations in fish tissues have been used to estimate the magnitude of
mercury exposures to these species.
4.1 The Modeling Analysis
The exposure assessment conducted for this Report examined mercury exposure through two
different approaches. The first approach addressed, through the use of models, the atmospheric fate
and transport and the exposures that result from atmospheric mercury emissions of selected stationary
anthropogenic combustion and manufacturing sources: municipal waste combustors (MWC), medical
waste incinerators (MWI), coal- and oil-fired utility boilers, chlor-alkali plants (CAP), primary lead
smelters and primary copper smelters. The second approach estimated the current exposures to the
general U.S. population that may result from methylmercury concentrations in freshwater and marine
fish. This was done through the use of fish consumption surveys and central tendency estimates of
chemically-measured methylmercury concentrations in fish.
4.1.1 Study Design of the Modeling Analysis
The fate, transport and exposure modeling of mercury was presented in Chapters 4, 5 and 6 of
Volume HI, an Assessment of Exposure from Anthropogenic Mercury emissions hi the United States.
The primary differences in these modeling exercises are the sources of the mercury concentrations in
the atmosphere and soil and the mercury deposition rate. In each instance the fate and transport of
deposited mercury was modeled for the same two hypothetical sites; that is a Western and an Eastern
site. Three different settings were overlayed on each site; rural (agricultural), lacustrine (or water
body) and an urban setting. These were selected because of the variety they provide (and to mimic
exposure situations likely to be found hi the U.S.). Three different hypothetical humans were assumed
to reside in each setting (total number was nine). Five hypothetical piscivorous wildlife species
(described in the preceding chapter) were also assumed to inhabit the lacustrine setting.
Chapter 4 of Volume HI utilized measured atmospheric and soil concentrations and a measured
deposition rate as inputs to the aquatic and terrestrial fate, transport, and exposure models (IEM2) at
the hypothetical Western and Eastern U.S. sites. See Figure 4-1 for an overview of this modeling
effort.
In Chapter 5 of Volume HI, modeled estimates were used as inputs. These were the 50th and
90th percentiles of the atmospheric mercury concentrations and the deposition rates that were predicted
by the RELMAP model for the Eastern and Western halves of the U.S. These estimates served as
inputs to the aquatic and terrestrial fate, transport, and exposure models (IEM2) at the hypothetical
Western and Eastern U.S. sites. Additionally, the environmental fate of mercury at a hypothetical site
hi the Eastern U.S. that is distant from anthropogenic emissions sources was also modeled. See Figure
4-2 for an overview of this modeling effort.
June 1996 4-1 SAB REVIEW DRAFT
-------
Figure 4-1
Fate and Transport Models Used and Exposure Routes Considered to
Examine Exposure Predictions Using Measured Environmental Concentrations
Deposition Rate of Mercury= 10/Xg/m/yr
June 1996
4-2
SAB REVIEW DRAFT
-------
Figure 4-2
Fate, Transport and Exposure Modeling Conducted in the Long Range Transport Analysis
June 1996
4-3
SAB REVIEW DRAFT
-------
Two separate modeling analyses were conducted for Chapter 6 of Volume III. In the first
analysis (Figure 4-3), hypothetical emission sources, or model plants, were developed to represent
major anthropogenic combustion and manufacturing sources: municipal waste combustors (MWC),
medical waste incinerators (MWI), coal- and oil-fired utility boilers, chlor-alkali plants (CAP), primary
lead smelters and primary copper smelters. The atmospheric fate and transport of the mercury
emissions from these representative model plants were modeled to determine local mercury deposition
using the COMPDEP model. The predicted mercury air concentrations and deposition rates that
resulted from individual model plants at 2.5, 10, and 25 kilometers from the source were used as
inputs to the aquatic and terrestrial fate, transport, and exposure models (IEM2) at the hypothetical
Western and Eastern U.S. sites.
The second modeling analysis in Chapter 6 of Volume IE (Figure 4-4) added the predictions
of Ihe COMPDEP model for the area around the individual model plants to the hypothetical Western
and Eastern locations to either the 50th or 90th percentile predictions of the RELMAP model for the
Western or Eastern sitfes, respectively. These combined model predictions were used as inputs to the
aquatic and terrestrial fate, transport* and exposure models (IEM2) at the hypothetical Western and
Eastern U.S. sites.
Because of the hypothetical nature of both the individual humans and the sites that were
considered, estimates of exposures to mercury resulting from the consumption of nonlocal fish, from
occupation or from background sources were not added to the exposure estimates developed in Chapter
6 of Volume ffl. These sources of mercury exposure may be significant, and for a site-specific
assessment it may be appropriate to consider these for members of an exposed subpopulation.
For each modeling analysis listed above, the same two hypothetical sites were used. One site
was located in the Eastern U.S. and the other in the Western U.S. The primary differences between
the two hypothetical locations were the assumed erosion characteristics for the watershed and the
amount of dilution flow from the water body. The eastern site was defined to have steeper terrain in
the watershed than the western site. Both sites were assumed to have flat terrain for purposes of the
air modeling.
•
For each modeling analysis at both hypothetical sites, the fate of deposited mercury was
examined in three different settings: rural (agricultural), lacustrine (or around a water body), and
urban. The primary differences between the urban and rural settings were the three hypothetical
humans assumed to inhabit each. In addition to three different hypothetical human inhabitants, the
lacustrine setting included the modeling of a circular drainage lake with a diameter of 1.68 Km,
average depth of 5 m, and a 2 cm benthic sediment depth. The ratio for the watershed area to surface
water area was 15 to 1; the watershed area was 33 Km. Piscivorous wildlife species were also
assumed to inhabit the lacustrine setting. The wildlife species considered were: mink, otter, bald
eagle, osprey and kingfisher; all were assumed to consume fish from the hypothetical lake in this
setting.
As noted previously, hypothetical humans were developed to represent several specific
subpopulations expected to have both typical and higher exposure levels. These individuals were
assumed to inhabit each setting. The high-end rural scenario consisted of a subsistence farmer and
child who consumed elevated levels of locally grown food products. The subsistence fanner was
assumed to raise livestock and to consume home-grown animal tissue and animal products, including
chickens and eggs as well as beef and dairy cattle. It was also assumed that the subsistence fanner
collected rainwater hi cisterns for drinking. The hypothetical individual used in the average rural
scenario was assumed to derive some of his food from a small garden.
June 1996 4-4 SAB REVIEW DRAFT
-------
Figure 4-3
Fate, Transport and Exposure Modeling Conducted in the Local Impact Analysis
Local Ha Source
June 1996
4-5
SAB REVIEW DRAIT
-------
Figure 4-4
Fate, Transport and Exposure Modeling Conducted in the Combined COMPDEP and RELMAP Local Impact Analysis
Local Ha Source
June 1996
4-6
SAB REVIEW DRAFT
-------
In the urban high-end scenario, an adult was assumed to derive some food from a small garden
similar in size to that of the average rural scenario. To address the fact that home-grown fruits and
vegetables generally make up a smaller portion of the diet in urban areas, the contact fractions were
based on weight ratios of home-grown to total fruits and vegetables consumed for city households.
The high-end urban scenario included a pica child. The average urban scenario consisted of an adult
who worked outside of the local area. The exposure duration for inhalation of the average adult,
therefore, was only 16 hours a day compared to the 24 hours a day for the rural scenario and high-end
urban scenario. The only other pathway (i.e., noninhalation) considered for this scenario was ingestion
of average levels of soil.
Three fish-consumption scenarios for humans were considered for the lacustrine setting. For
the adult high-end fish consumer scenario (or subsistence fisher), an individual was assumed to ingest
large amounts of locally-caught fish, to eat home-grown garden produce (plant ingestion parameters
identical to the rural home gardener scenario), to consume drinking water from the affected water body
and to inhale the air on a 24-hour basis. A child of a high-end local fish consumer was assumed to
ingest local fish, local garden produce, and soil as well as to inhale the affected air. The exposure
pathways considered fof recreational angler scenario, evaluated only fish ingestion, inhalation, and soil
ingestion. These consumption scenarios were thought to represent identified fish-consuming
subpopulations in the United States.
Piscivorous birds and mammals were also assumed to inhabit areas adjacent to the hypothetical
lakes considered. The piscivorous animals were exposed to be mercury only through the consumption
of fish from the lake. The five wildlife species were not selected because they were more sensitive to
methylmercury exposure than other wildlife, but rather on the basis of exposure; Fish-consuming
species were, thus, the only groups considered in this assessment. All five wildlife species were
assumed to consume fish from trophic levels 3 and/or 4 and to inhabit the aquatic environment
modeled for a lifetime. Mercury concentrations in food sources other than fish and migratory
behaviors were not considered.
4.1.2 Uncertainties and Defaults Used in Exposure Modeling
The exposure analysis relied heavily on the modeling of the fate and transport of emitted
mercury because no monitoring data have been identified that conclusively demonstrate or refute a
relationship between any of the individual anthropogenic sources in the emissions inventory and
increased mercury concentrations in environmental media or biota. To determine if there is a
connection between the above sources and increased environmental mercury concentrations, three
models were utilized to address many major scientific uncertainties.
(1) The RELMAP model was used to predict average annual atmospheric mercury
concentrations as well as wet and dry deposition flux for 40 Km2 grids across the
continental U.S. The model predictions were based on anthropogenic emissions from
the sources described in Volume n, Inventory of Anthropogenic Mercury Emissions in
the United States.
(2) The COMPDEP model was used to predict average annual atmospheric mercury
concentrations as well as the wet and dry deposition resulting from emissions within
50 Km of a single source.
June 1996 4-7 SAB REVIEW DRAFT
-------
(3) The Indirect Exposure Models (EEM2) was used to predict environmental
concentrations and the exposures that result from atmospheric mercury concentrations
and deposition.
Volume III and the appendices describe at length the justification for choices of values for
model parameters; for example the size of combustors, the stack heights, amount of precipitation, adult
body weight and the amount and types of foods consumed. In this section of the Risk
Characterization, several of the major areas of uncertainty are highlighted without reiteration of the
entire list of parameter justifications generated in Volume III. Obviously, when models are utilized
there is an associated uncertainty.
4.1.2.1 Emissions Uncertainties
The degree of uncertainty in the emission rates varies for each hypothetical emission source
(model plant). This uncertainty is reflected in the qualitative characterization of the potential human
health and ecological risks.
Physical characteristics of anthropogenic emission sources vary. There is general
understanding of how these variations of physical characteristics affect dispersion of emitted mercury.
The following characteristics affect mercury emission rates, mercury speciation and mercury
transport/deposition rates: stack height, stack diameter, exit gas velocity, stack gas temperature, plant
capacity factor (relative average operating hours per year), stack mercury concentration, and mercury
speciation. In the exposure analysis the physical characteristics that were predicted to have the
greatest impact on the modeling of atmospheric transport of mercury were chemical species of mercury
emitted, exit gas velocity and stack height
There is substantial variation hi the mercury content of the feed mixes that enter combustors.
Emissions of mercury (including the divalent mercury species, and elemental mercury in various
speciation percentages) are influenced by the type of fuel used (e.g., coal, oil, municipal waste), flue
gas cleaning and operating temperatures. To the extent that these factors vary in a facility, chemical
characteristics of mercury emissions will vary. Consequently, the exit stream can vary from nearly all
elemental mercury to nearly all divalent mercury, contributing to the variability in fate and transport of
mercury.
The chemical species released from anthropogenic sources is expected to determine the
atmospheric fate and transport characteristics of the emissions. Modeling of the exact chemical species
(e.g., HgCl2, Hg(OH)2) was not attempted. It is possible to break the divalent mercury species down
further as, for example, reactive, non-reactive, or particle-bound. This was infrequently measured for
the sources considered, which contributes to both variability and uncertainty hi the results of the
atmospheric modeling. Determining the concentration and speciation of mercury in stack emissions is
also complicated by sampling difficulties related to identification of the chemical species in the emitted
gas. Sampling procedures may alter the physical characteristics of the emitted mercury. To the extent
that the chemical species are uncertain and variable, the predictions of atmospheric fate and transport
are uncertain and variable.
4.1.2.2 Atmospheric Reactions of Emitted Mercury
Atmospheric chemistry data for mercury are incomplete. Some atmospheric reactions of
mercury, such as the oxidation of elemental mercury to divalent mercury in cloud water droplets have
been reported and have been incorporated into the modeling. Other chemical reactions in the
June 1996 4-8 SAB REVIEW DRAFT
-------
atmosphere such as those which may reduce divalent species to elemental mercury or processes by
which mercury attaches to atmospheric particulates have not been adequately reported or modeled.
There is inadequate information on the atmospheric processes which affect wet and dry
deposition of mercury. Atmospheric paniculate forms and divalent species of mercury afe thought to
wet and dry deposit more rapidly than elemental mercury; however, the relative rates of deposition are
uncertain.
4.1.2.3 Deposition of Atmospheric Mercury
Based on experimental data, divalent mercury and particulate-bound mercury will deposit on
land. The deposition velocity of mercury may differ with chemical species and conditions of land use
patterns. The deposition velocity for atmospheric mercury over soil and over water is very poorly
defined.^ The following gaps in information result in uncertainties in this risk characterization.
• There is a lack of adequate emission data for various sources, including natural
sources. This includes emissions data on the amounts of various forms of mercury
that may be emitted from stacks.
• Emissions of particulates from various combustion sources depend on these factors:
Type of furnace and design of combustion chamber,
Composition of feed/fuel;
Particular matter removal efficiency and design of air pollution control
Equipment; and
Amount of air in excess of stoichiometric amount that is used to sustain
temperature of combustion.
These conditions are highly variable in actual operation of specific incinerators. Consequently,
emission of mercury and particulates is highly variable.
• There is a lack of information on the effect of atmospheric transformation processes on
wet and dry deposition; for example, how deposition is affected by the transformation
of elemental mercury to divalent mercury, or vice versa.
• There is no validated air pollution model that estimates local wet and dry deposition of
an emitted gas (such as elemental mercury).
• There is a lack of data on methylation processes in water bodies; thus, assumptions
related to this transformation of mercury must be made in modeling.
• There is a lack of data on the transfer of mercury between environmental and
biological compartments (e.g., uptake of mercury into aquatic organisms).
The parameters exerting the most influence on the exposure assessment are these:
• Total mercury emission rate (grams/second);
• Assumption regarding speciation of the total mercury;
• Vapor/particle phase partition estimate;
• Stack height for the plant; and
June 1996 . 4-9 SAB REVIEW DRAFT
-------
• Exit gas velocity.
4.1.2.4 Mercury Concentrations in Water and Aquatic Biota
Available measured mercury concentrations in water, soil and fish'near anthropogenic mercury
emissions sources do not consistently indicate local deposition of mercury around a point source.
These observations raise questions with respect to uncertainty in the models used to estimate indirect
exposures to mercury.
A bioaccumulation factor (BAF) was used hi the estimation of mercury concentrations in fish
as a consequence of mercury hi the water body. Discussion of the BAF derivation is in Section
2.7.2.3 of Volume V. Determination of an appropriate BAF relies on analyses of mercury hi water
and mercury hi fish. There is a major limitation on the analytical quality (detection limits,
contamination control, chemical speciation) for mercury hi both water and hi fish. This makes
estimation of the bioaccumulation factor uncertain.
Transport to water bodies from watersheds varies with local soil conditions including extent of
erosion, run-off and the soil-water partition of mercury. Likewise transport of various species of
mercury within water bodies is highly variable. These chemical species of mercury may be bound to
suspended soil/humus or attached to dissolved organic carbon. These local conditions are highly
variable and poorly documented. Removal of divalent mercury and methylmercury from the
groundwater into upper soil layers is poorly characterized. There are likely to be concentration-
dependent differences hi the distribution of mercury between the donating and receiving water bodies
and between water and soil. Within the water bodies, the species of mercury is expected to be
concentration-dependent
Data indicate that 25-60 percent of divalent mercury and methylmercury organic complexes in
the water column are particle-bound. This variation hi the fraction of mercury that is bound to
particles influences the fate and transport of environmental mercury.
•
The bioaccumulation of methylmercury from the environment differs for plants and animals in
the terrestrial food chain and fish/fish-eating mammals in the aquatic food chain. For example, the
total mercury concentrations in fish are often 100 to 1,000 times greater than hi terrestrial mammalian
tissue (meat). Most of the mercury in meat is not likely to be methylmercury, hi contrast to fish in
which nearly 100 percent of the mercury is methylmercury. Depending on the type of diet chosen
(e.g., fraction of total caloric intake from fish or fish-eating mammals), the methylmercury content of
the diet will vary greatly.
4.1.2.5 Mercury Accumulation hi Fish
Mercury accumulation hi fish was modeled using bioaccumulation factors (BAF) and predator-
prey factors (PPF) at the higher trophic levels. The BAF paradigm was modified from methodology
developed hi the Great Lakes Water Quality Initiative. BAFs were calculated for trophic levels
3 and 4; that is, for small fish and for larger predatory fish.
Field studies indicate that many, if not most fish, accumulate mercury throughout their lives,
often hi a nearly linear fashion with age (see for example: Scott and Armstrong, 1972; MacCrimmon
et al., 1983; Wren et al., 1983; Mathers and Johnansen, 1985: Skurdal et al., 1984; Wren and
MacCrimmon, 1986; Sorenson et al., 1990; Jackson, 1991; Gutenmann et al., 1992; Glass et al., 1993;
Suchanek, 1993; Lange et al., 1993). The bioaccumulation factor depends on the age of the fish.
June 1996 4-10 SAB REVIEW DRAFT
-------
Differences in the age (and consequently the size) of fish consumed by either wildlife or humans are a
major source of variability in the bioaccumulation factor.
Methylmercury as a percentage of total mercury present in the bodies of aquatic organisms
appears to decrease with decreasing trophic level. Organisms at trophic levels 1 and 2 are, in general,
short-lived and do not have the opportunity to accumulate mercury for periods lasting many years. As
mercury is transferred up the food chain there appears to be a progressive enrichment of the most
bioaccumulative form, methylmercury.
Most of the mercury accumulated by fish at trophic level 4 is thought to be taken up from
dietary sources. Thus, particularly for long-lived piscivorous fish, a relatively short (one year or less)
waterborne exposure will not duplicate the extent of accumulation that takes place over a longer period
of time.
A probabilistic Monte Carlo simulation approach was used for the estimation of BAFs; this is
described in Appendix A to Volume V. This approach was taken to allow explicit quantitative
expression of the overall variability surrounding the various estimates of the BAFs. This analysis was
also done to determine the relative sensitivity of the estimates to specific individual variables. The
distributions were based on data from field studies which measured mercury concentrations in fish at
trophic levels 3 and 4.
Because of the large variance in the BAF distributions, it is most appropriate to apply BAFs
derived from valid data collected at the site of concern. The geometric mean values of the BAF
distributions were used for this exposure assessment rather than the upper or lower percentiles. The
values based on the probabilistic simulations were 66,200 and 335,000 for the BAF for trophic level
three (BAF3) and the BAF for trophic level four (BAF^, respectively.
There is uncertainty as to whether a single BAF value is appropriate for derivation of water
concentrations when the fish size range of the fish-consuming populations is known. For example,
kingfishers feed on smaller fish while human recreational anglers primarily consume large fish. This
is handled in the exposure assessment by assigning a portion of the total fish consumed to either
trophic level 3 or 4. For example, only BAF3 is used in estimating king fisher mercury exposure.
4.2 Estimates of Methylmercury. Exposure Based on Monitoring Data, Dietary Surveys and
Mercury Residue Data
4.2.1 Non-Human Mammalian Species Exposures to Methvlmercurv
Mink (Mustela visori) and otter (Lutra canadensis) occupy top trophic positions in the aquatic
foodweb and bioaccumulate mercury from food. The diet of mink varies with location, time of year,
and available prey. Mink consume fish, small animals, crayfish, birds, and amphibians (Linscombe et
al., 1982). Otters, by contrast, are much more consistently fish eaters whose diet consists of at least
95 percent fish (Toweill and Tabor, 1982). For both otter and mink, the mercury concentrations in
these animals' tissues have been positively associated with mercury levels in prey (for example; fish,
shellfish, crayfish) (Wren and Stokes, 1986; Foley et al., 1988; Langlois and Langis, 1995). Mink and
otter accumulated about 10 times more mercury on a concentration basis than did predatory fishes
from the same drainage areas (Kucera, 1983). These correlations were statistically significant (Foley
et al., 1988) on the basis of mercury in the watershed because of the importance of fish, shellfish and
crayfish to the minks' and otters' diets.
June 1996 4-11 SAB REVIEW DRAFT
-------
Case reports of clinical mercury poisoning exist for wild mink (Wobeser and Swift, 1976) and
otter (Wren, 1985). Such reports are rare, but this would be expected given the rapid onset of
symptomatology of methylmercury poisoning, and assuming that the wild mink exhibit the same
progression of signs and symptoms observed in a laboratory setting. Under the experimental situation
established by Wobeser (1973), the minks' conditions deteriorated from anorexia and ataxia to death
within two or three days at exposures producing liver mercury concentrations in excess of
approximately 20 ug/g. The short time-period between onset of gross signs and symptoms of
methylmercury intoxication and death decreases the likelihood of observing in the wild clinically ill
mink prior to death. Consequently assessment of mercury exposure to wildlife has been based on
mercury concentrations in body organs such as liver, kidney and brain rather than an observation of
gross clinical symptamology. The magnitude of the concentration in one organ for both mink and
otter (for example, liver) is highly correlated with other organs (for example, kidney or brain); see
reports of Wobeser (1973), Kucera (1983), Wren and Stokes (1986). Usually mercury concentrations
in liver are used for comparison across studies.
Liver mercury concentrations hi the range of 20 to 25 ug/gram fresh weight were associated
with severe, clinically evident mercury poisoning in mink fed 1.8 ug/gram methylmercury in diet
(Wobeser, 1973). Among animals that died during the experimental period, liver mercury
concentrations averaged above approximately 25 ug/gram fresh weight (Wobeser, 1973). Using mink
and otter trapped by fur traders or trappers, mercury concentrations have been reported for Quebec
{Langlois and Langis, 1995), Ontario (Wren et al., 1986), Manitoba (Kucera, 1983), New York State
(Foley et al., 1988); and Georgia (Halbrook et al., 1994). The range of concentrations reported in
different geographic locations is substantial. Wild mink with liver mercury concentration as high as
20 ug/g were identified hi northern Quebec (Langlois and Langis, 1995).
There are substantial region-to-region differences in mercury concentrations in tissues of mink
and otters. There are also differences among individual animals trapped in a particular location.
Consequently broad generalizations are difficult regarding how close liver mercury concentrations of
wildlife are to liver mercury concentrations of experimentally poisoned mink. However, the upper
range of liver mercury concentrations of mink from northern Quebec (Langlois and Langis, 1995),
otters from Georgia (Hallbrook et al., 1994) and otters from Ontario (Wren et al., 1986) approximate
those of clinically poisoned animals. Based on these reports, methylmercury poisoning sufficiently
severe to be fatal to mink and otters can be projected at current mercury exposures in some geographic
locations.
Sublethal effects on mink and otters can be projected to be more wide-spread with additional
reports showing average liver mercury concentrations approximately one-third of those in moribund
mink with experimental methylmercury poisoning. For example, hi some geographic areas, average
concentrations are about one-third those of mink with clinical mercury poisoning in a laboratory
situation. Liver mercury concentrations of river otters from the lower coastal plain in Geogria
averaged 7.5 ug/g (Hallbrook et al., 1994); this is approximately 33 percent of the concentrations
associated with severe intoxication and/or death in a closely related specie, the mink (Wobeser et al.,
1976a,b). In many geographic regions [e.g. Georgia (Halbrook et al., 1994), New York State (Foley et
al., 1988)], mercury concentrations in mink and otters' tissues are 10 to 30 percent of concentrations
associated with severe, clinically evident methylmercury poisoning in mink.
Average tissue mercury concentrations for mink and otter from multiple regions of North
American are within an order of magnitude of tissue mercury concentrations of mink severely
poisoned experimentally. For example, data showing mink liver mercury concentrations averaging 2
ug/g or higher were reported in several regions of New York State (Foley et al., 1988), Ontario (Wren
June 1996 4-12 SAB REVIEW DRAFT
-------
et al., 1986), and Manitoba (Kucera, 1983). Concentrations in excess of 20 \ig/g occurred in mink
dying of methylmercury poisoning (Wobeser, 1973; Wobeser et al., 1976a,b, 1979).
There may be other factors in addition to methylmercury concentration in the food supply of
the mink and otter that are responsible for the association. Liver mercury concentrations in wild mink
were not always predictably associated with proximity to long-term mercury contamination. For
example, Wren et al., (1986) found that wild mink trapped in the English River system, severely
contaminated by mercury discharge from a chloralkali plant 15 to 22 years earlier than the dates of
mink trapping, had a range of 0.6 to 6.9 ug/g liver. By contrast mink trapped in the Turkey Lakes
watershed, a region considered relatively pristine, had liver mercury concentrations ranging between
1.1 and 7.5 ug/gram fresh tissue (Wren et al., 1986). Another region of Ontario was substantially
lower hi mercury contamination; wild mink from Cambridge had average liver concentrations of 0.14
ug/g (fresh weight) (Wren et al., 1986).
%
4.2.2 Avian Species Exposure to Methylmercurv
During the decades when seed-grams were treated with organo-mercurial fungicides huge
numbers of wild birds were poisoned fatally with mercury. In the 1970s declining use of organo-
mercurial fungicides greatly reduced the severity of mercury exposure. However, mercury residues
either through natural or anthropogenic mercury- sources remain. During the period 1990 through mid-
1995 several reports of mercury concentrations hi avian species have been published hi the peer-
reviewed literature (among others see Bowerman et al., 1994; Burger et al., 1993, 1994; Custer and
Hohman, 1994; Spalding et al., 1994; • Sundlof et al., 1994; Langlois and Langis, 1995; Lonzarich et
al., 1992; Thompson et al., 1992). Considered with earlier information on mercury concentrations in
tissues from avian wildlife, mercury is a common contaminant of avian tissues from diverse
geographic locations. Mercury concentration in tissues have been reported for the following birds:
seabirds from colonies hi the Northeast Atlantic (Thompson et al., 1992); the common tern hi
Buzzards Bay, Massachusetts (Burger et al., 1994); the California clapper rail from the salt marshes of
central and northern California (Lonzarich et al., 1992); canvasback ducks hi Louisiana (Custer and
Hohman, 1994); wading birds of Southern Florida (Sundlof et al., 1994; Spalding et al., 1994; Burger
et al., 1993); loons in the Great Lakes regions and Ontario (Barr, 1986); and the Bald Eagle in the
Great Lakes Region (Bowerman et al., 1994).
Feeding habits of particular avian species are a major predictor of risk of mercury toxicity in
the 1990s. When seed-grains were treated with organo-mercurial fungicides herbivorous and
omnivorous species were at risk of mercury toxicity, as were carnivorous birds. Because of the
biomagnification of methylmercury hi the aquatic foodweb, birds which feed on fish, crayfish or
shellfish now have higher exposures to methylmercury than do non-fish eating birds. Birds, such as
the heron, which consume large fish as their prey are predicted to be at greater risk of methylmercury
poisoning than are birds consuming smaller fish (Spalding et al., 1994; Sundlof et al., 1994). When
the quantity of fish consumed on a body weight basis is also considered smaller birds, such as the
kingfisher, have been judged to be at elevated risk of methylmercury poisoning.
Several estimates exist in the published literature on quantities of mercury in soft tissues (liver,
kidney, brain) associated with mercury poisoning in avian species. Experimental studies of survival .
and reproductive success of black ducks (Anas rubripes) by Findley and Stendell (1978) indicated that
adult black ducks would tolerate liver mercury concentrations of 23 ppm and appear in good health.
They found, however, that even though the black ducks fed methylmercury in diet appeared in good
health mat they had impaired reproductive success indicated by reduced hatchability of eggs and high
duckling mortality. Findley et al. (1979) concluded that concentrations of mercury in excess of 20
June 1996 4-13 SAB REVIEW DRAFT
-------
ug/g fresh weight in soft tissues should be considered extremely hazardous to avian species.
Scheuhammer (1991) in a review indicated that the major effects of methylmercury in avian species
were neurological, developmental and reproductive. The neurological changes included weakness,
walking or flying difficulties and inco-ordination which were associated with brain mercury
concentrations of 15 ug/g (fresh weight), or liver or kidney mercury concentration of 30 ug/g (fresh
weight). Schuehammer observed that generally significant reproductive impairment due to
methylmercury occurs at about one-fifth the tissue concentrations required to produce overt
neurotoxicity. Scheuhammer indicated that liver mercury concentrations of 2 to 12 ug/g (fresh weight)
in adult breeding pheasants and mallard ducks were linked to decreased hatchability of eggs. Barr
(1986) reported adult loons with total mercury concentrations in the brain as low as 2 ppm (fresh
weight) showed aberrations in reproductive behavior, resulting in lowered incubation success and
abandonment of territories. Both methylmercury and total mercury concentrations in liver and brain of
the loon have correlation coefficients of 0.58 and 0.46, respectively. Barr (1986) noted that clinical
signs of mercury poisoning, such as impaired vision and ataxia, had been found in several avian
species (as reported by Evans and Kostyniak, 1972; Hays and Risebrough, 1972) at mercury
concentrations lower than those present in the loons from one of the sites of Barr's investigations.
Barr (1986) notes that impairment of vision or ataxia in a visual hunter such as loon would be likely
to reduce its chances of procuring adequate food and defending a territory.
Mercury concentrations in livers of wading birds in Southern Florida (Sundlof et al., 1994;
Spalding et al., 1994) and the merganser in northern Quebec (Lanlois and Langis, 1995) are in the
range associated with adverse reproductive and neurological effects in other species of birds. Sundlof
et al. (1994) reported that four great blue herons (Ardea herodias) collected from the central
Everglades contained liver mercury at concentrations typically associated with overt neurological signs
(>30 ug mercury/g fresh weight). Furthermore, these investigators found between 30 percent and 80
percent of the potential breeding-age birds collected in an area encompassing the central Everglades
contained liver mercury at concentrations associated with reproductive impairment in ducks and
pheasants. In a parallel study Spalding et al. (1994) determined the magnitude of mercury
contamination associated with death of great white herons (Ardea herodias occidentalism. Birds that
died of acute causes (e.g., trauma from collision with powerlines or vehicles) had much lower liver
mercury concentrations (geometric mean 1.8 ug/g fresh weight, range 0.6 to 4.0 ug/g fresh weight)
than did birds that died of chronic diseases (geometric mean 9.8 ug/g fresh weight, range 2.9 to 59.4
ug/g fresh weight).
The common merganser (Mergus merganser) and red-breasted merganser (Mergus serrator)
were among wildlife species sampled in the Great Whale and Nottaway-Broadback-Rupert (NBR)
hydroelectro projects in northern Quebec (Langlois and Langis, 1995). Liver mercury concentrations
for these species were reported as mean ± standard deviation (SD) (shown in Table 4-1). Using
standard statistical procedures it is estimated that 33.3 percent of the liver mercury concentrations for
the respective species would be greater than the mean+one standard deviation. If the liver
concentrations associated with neurological, reproductive and developmental effects in other avian
species are applicable to the common and red-breasted merganser, adverse health and reproductive
effects are associated with mercury exposures experienced by these avian species.
June 1996 4-14 SAB REVIEW DRAFT
-------
Table 4-1
Liver Mercury Concentration in Common Merganser, Red-Breasted Merganser
and Herring Gulls from Northern Quebec (Langlois and Langis, 1995)
Mercury Concentration, liver (|ig/g fresh weight)
Species
Common
Red-breasted
merganser
Herring
gull
Location
Great Whale
Mean±SD
17.5 ±12.0
12.4±18.8
2.9±2.4
Mean + 1 SD
66.7th Percentile
29.5
33.2
5.3
Mean + 2 SD
95th Percentile
41.5
50.0
7.7
NBR
Mean± SD
10.5*7.5
No values
reported
3.6±2.5
Mean + 1 SD
66.7th Percentile
17.5
-
6.1
Mean + 2 SD
95th Percentile
25.0
—
9.7
Tissue mercury concentrations and population dynamics of the common loon (Gavia immef) in
an area with mercury-contaminated waters in northwestern Ontario were reported by (Barr, 1986).
Mercury concentrations for total and methylmercury for adults and chicks for liver, muscle, and brain
are shown in Table 4-2. The concentration of total mercury residue in loon tissues decreased hi the
Table 4-2
Mercury and Methylmercury Concentrations in Tissues (jig per Gram Fresh Weight)
from the Common Loon in Northwestern Ontario (Barr, 1986)
Adults
Total
mercury
Methyl-
mercury
Chicks
Total
mercury
Methyl-
mercury
Liver
Mean
12.95
11.67
0.91
0.80
SD
11.67
2.40
0.33
0.29
Range
1.64-
47.71
0.00-
10.20
0.35-1.47
0.32-1.36
Muscle
Mean
2.33
1.65
0.44
0.37
SD
2.07
1.60
0.22
0.20
Range
0.16-6.87
0.15-6.59
0.14-0.89
0.09-0.80
Mean
0.86
0.65
0.37
0.37
Brain
SD
0.89
0.79
0.18
0.17
Range
0.31-
4.61
0.22-
4.27
0.14-
0.78
0.14-
0.75
sequence liver > muscle >brain, but the percentage of methylmercury increased from liver < muscle <
brain. Barr (1986) found that almost 100 percent of the mercury transferred from adult loons through
eggs to chicks was organic mercury with no net loss of methylmercury in chick tissue. Levels of
June 1996
4-15
SAB REVIEW DRAFT
-------
methylmercury in eggs and in the brain of newly hatched chicks frequently exceeded levels in the
female parent's brains. There was a statistically significant correlation between total mercury levels in
the brain of nesting females and their eggs (p=0.005).
*
The upper portion of the range of liver mercury concentrations for the loon was greater than
mercury concentrations associated with overt clinical toxicity in other avian species. Barr (1986)
reported finding loons that were emaciated, expected to accompany either anorexia or reduced ability
to obtain prey. Bair's conclusions were than there was a strong negative correlation between
successful use of territories by breeding loons and mercury contamination (Barr, 1986). Liver mercury
concentrations (mean approximately 13 ppm, range approximately 2 to 48 ppm mercury) were higher
than the range identified by Schuehammer as being associated with reproductive failure in other avian
species: 2 to 12 ppm mercury (Scheuhammer, 1991). Scheuhammer concluded that results suggest a
reduction in egg laying, and in nest and territorial fidelity at mercury concentration ranging from 0.3
to 0.4 ppm in prey and 2 to 3 ppm in adult loon brain and loon eggs. These data confirm earlier
reports by Fimreite and Reynolds (1973) that the common loon may be particularly adversely affected
by high levels of methylmercury accumulated from their diet
4.2.3 Human Intake of Methylmercury Estimated Through Dietary Surveys and Mercury Residue
Data
Food intake primarily from the ingestion of contaminated fish is the only significant source of
methylmercury exposure to the general human population (Stern, 1993; Swedish EPA, 1991; WHO,
1990). Total mercury concentrations in meats and cereals often measure hundreds of times less than
in fish (Swedish EPA, 1991). In most non-fish foodstuffs mercury concentrations are typically near
detection limits and are comprised mainly of inorganic species (WHO, 1990). In contrast, most of the
mercury in fish is methylated and fish methylmercury concentrations are typically above analytic
detection limits.
Available techniques to estimate fish consumption include long-term dietary histories and
questionnaires to identify typical food intake or short-term dietary recall techniques. Temporal
variation in dietary patterns is an issue to consider in evaluation of short-term recall/record data. For
epidemiological studies that seek to understand the relationship of long-term dietary patterns to chronic
disease, typical food intake is the relevant measure to evaluate (Willett, 1990). Because
methylmercury is a developmental toxin that may produce adverse effects following a comparatively
brief exposure period (i.e., a few months rather than decades), comparatively short-term dietary
patterns can have importance.
Estimates of fish consumption rely, on dietary survey data that can be obtained using a variety
of dietary survey techniques. Critical elements in any survey aimed at determining intake of
methylmercury from fish are these:
• Species of fish or shellfish consumed;
• Concentration of methylmercury in the fish;
• Quantity of fish consumed.
The duration of fish consumption is also of importance; however, the time period of
consumption that is relevant when conducting an assessment of risk depends on the health endpoint of
concern. To illustrate, acute effects of certain fish contaminants (such as paralytic shell fish toxin or
cigutara toxin) may result from eating as little as one meal of contaminated fish. By contrast, if one is
interested in the benefits of consuming unsaturated fatty acids (e.g., omega .3 fatty acid) on prevalence
June 1996 4-16 SAB REVIEW DRAFT
-------
of cardiovascular disease, decades of exposure for a group of persons is typically required to establish
whether or not an effect would occur. For a health endpoint such as developmental deficits associated
with a particular period during gestation (e.g., adverse effects of maternal consumption of
methylmercury from fish on the developing fetal nervous system), short-term consumption patterns
during the critical weeks or months of gestation are considered the relevant period for the health
endpoint.
Survey methods can broadly be classified into longitudinal methods or cross-sectional surveys.
Typically long-term or longitudinal estimates of intake can be used to reflect patterns for individuals
(e.g., dietary histories); or longitudinal estimates of moderate duration (e.g., month-long periods) for
individuals or groups. Cross-sectional data are used to give a "snap shot" in time and are typically
used to provide information on the distribution of intakes for groups within the population of interest.
Cross-sectional data typically are for 24-hour or 3-day sampling periods and may rely on recall of
foods consumed following questioning by a trained interviewer, or may rely on written records of
foods consumed. Additional discussion of these issues are found on Appendix H to Volume HI.
During the past decade reviewers of dietary survey methodology .(for example, the Food and
Nutrition Board of the National Research Council/National Academy of Sciences; the Life Sciences
Research Office of the Federation of American Societies of Experimental Biology) have evaluated
various dietary survey techniques with regard to their suitability for estimating exposure to
contaminants and intake of nutrients. The Food and Nutrition Board of the National Research
Council/National Academy of Sciences hi their 1986 publication, Nutrient Adequacy Assessment Using
Food Consumption Surveys, noted that dietary intake of an individual is not constant from day to day,
but varies both hi amount and in type of foods eaten (intraindividual variation). Variations between
persons in their usual food intake averaged over time is referred to as interindividual variation.
Among North American populations, the intraindividual (within person day-to-day) variation is usually
regarded to be as large as or greater than the interindividual (person to person) variations. Having
evaluated a number of data sets the Academy's Subcommittee concluded that 3 days of observation
may be more than is required for the derivation of the distribution of usual intakes.
Major sources of data on dietary intake of fish used in preparing this report to Congress are
the cross-sectional data from the USDA Continuing Surveys of Food Intake by Individuals conducted
hi the years 1989 through 1991 (CSFII 89/91) and the longer-term data on fish consumption based on
recorded fish consumption for variable numbers of periods of one-month duration during the years
1973/1974. The SE data are from the National Purchase Diary (NPD 73/74) conducted by the Market
Research Corporation.
The CSFII 89/91 data are cross-sectional data based on a 3-day sampling period. When
appropriately weighted they can be used to estimate the food consumption patterns for the general
United States population for the period 1989/1991 there were 11,706 respondents. The survey is
designed to represent all seasons of the year and all days of the week. Because the food consumption
records rely on standard coding of food intake and records of types of fish represented by a particular
dietary record, it is possible to estimate quantities of particular types of fish that were consumed for
the population as a whole and for subpopulations of interest. The portion size consumed by
individuals is recorded, as is the person's self-reported body weight.
The CSFII 89/91 data on fish consumption have been used to estimate fish and methylmercury
intake by various population subgroups. These calculations rely on values for the methylmercury
concentrations in food supplied to the US EPA by the National Marine Fisheries Service. These data
are presented in detail in Appendix H to Volume III.
June 1996 4-17 SAB REVIEW DRAFT
-------
4.2.3.1 Estimates of Human Intake of Methylmercury Based on Longitudinal Data
Estimates of fish consumption in the 1970s were determined by the NPD Research Inc., a
market research and consulting firm that specializes in the analysis of consumer purchasing behavior
as recorded in monthly diaries. That survey was funded by the Tuna Research Institute (TRI) as part
of a study of tuna consumption. Later, the National Marine Fisheries Service (NMFS) received
permission from TRI to obtain the data (SRI International Contract Report to U.S. EPA, 1980).
The NPD 73/74 data are based on a sample of 7,662 families (25,165 individuals) out of 9,590
families sampled between September 1973 and August 1974. Data recorded in the survey reflect the
marketing nature of the survey design and have limitations with regard to quantities of fish consumed
on a body weight basis. To illustrate, the fish consumption was based on questionnaires completed by
the female head of the household in which she recorded the date of any meal containing fish, the type
of fish (species), the packaging of the fish (cartned, frozen, fresh, dried, or smoked, or eaten out), \
whether fresh fish was recreationally caught or commercially purchased, the amount of fish prepared
for the meal, the number of servings consumed by each family member and any guests, and the
amount of fish not consumed during the meal. Meals eaten both at home and away from home were
recorded.
Use of these data, to estimate intake of fish or mercury on a body weight basis are limited by
the following data gaps.
1. This survey did not include data on the quantity of fish represented by a serving and
information to calculate actual fish consumption from entries described as breaded fish
or fish mixed with other ingredients. Portion size was estimated by using average
portion size for seafood from USDA Handbook #11. Table 10, page 40-41. The
average serving sizes from this USDA source are shown in Table 4-3.
Table 4-3
Average Serving Size (gins) for Seafood from
USDA Handbook # 11 Used to Calculate Fish Intake by FDA (1978)
Age Group
(years)
0-1
1-5
6-11
12-17
18-54
55-75
Over 75
Male
Subjects
(gins)
20
66
95
131
158
159
180
Female
Subjects
(gms)
20
66
95
100
125
130
139
2. There may have been systematic under-recording of fish intake; Crispin-Smith et al.
noted that typical intakes declined 30% between the first survey period and the last
June 1996
4-18
SAB REVIEW DRAFT
-------
survey period among persons who completed four survey diaries (Crispin-Smith et al.,
1985).
3. There have been changes in the quantities and types of fish consumed between
1973/1974 and present To illustrate, the United States Department of Agriculture
indicated (Putnam, 1991) that, on average, fish consumption increased 27% between
1970 and 1990. Whether or not this increase applies to the highest percentiles of fish
consumption (e.g., 95th or 99th percentile) was not described in the publication by
USDA.
4. An analyses of these data using the sample weights to project these data for the
general United States population was prepared by SRI International under US EPA
Contract 68-01-3887 in 1980. U.S. EPA was subsequently informed that the sample
weights were not longer available. Consequently additional analyses with these data in
a manner that can be projected to the general population appears no longer to be
possible.
5. Body weights of the individuals surveyed do not appear in published materials. If
body weights of the individuals participating in this survey were recorded these data
do not appear to have been used in subsequent analyses.
Data on fish consumption from the NPD 73/74 survey have been published by Rupp et al.,
1980 and analyzed by US EPA's contractor SRI International (1980). These data indicate that when a
month-long survey period is used 94% of the surveyed population consumed fish. The species of fish
most commonly consumed are shown in Table 4-4.
June 1996 4-19 SAB REVIEW DRAFT
-------
Table 4-4
Fish Species and Number of Persons Using the Species of Fish.
Adapted from Rupp et al., 1980
Category
Tuna, light
Shrimp
Flounders
Not repotted (or identified)
Perch (Marine)
Salmon
Clams
Cod
Pollock
Haddock
Herring
Oysters
Crab, other than King
Trout (freshwater)
Lobster, northern
Halibut
Scallops-
Mackerel, other than jack
Whitefish
Snapper
Hake
Catfish (freshwater)
Lobster, spiny
Smelt
Bass
Perch (freshwater)
Bluegills
Crappie
Trout (marine)
Benito
Number of Individuals Consuming Fish
Based on 24,652 Replies*
16,817
5,808
3,327
3,117
2419
2,454
2,242
1,492
1,466
1,441
1,351
1,239
1,168
970
575
574
526
515
492
490
372
376
350
328
326
268
265
228
220
148
* More than one species of fish may be eaten by an individual.
Rupp et al. also estimated quantities of fish and shellfish consumed by 12-18 year-old
teenagers and, by adults 18 to 98 years of age. These data are shown in Table 4-5. The distribution of
fish consumption for age groups that included women of child-bearing ages are shown hi Table 4-6.
June 1996
4-20
SAB REVIEW DRAFT
-------
Table 4-5
Fish Consumption from the NPD 1973/1974 Survey
(modified from Rupp et ah, 1980)
Age Group
Teenagers aged 12-18
Years
Adults aged 18 to 98
Years
50th Percentile
1.88 kg/year
2.66 kg/year
90th
Percentile
8.66 kg/year
14.53 kg/year
99th
Percentile
25.03 kg/year
or
69 grams/day
40.93 kg/year
or
112 grams/day
Maximum
62.12 kg/year
167.20 kg/year
Table 4-6
Percent of Females By Age* Consuming Fish/Shellfish
from SRI (1980)
Age Group (years)
10-19
20-29
30-39
40-49
47.6-60.0 gins/day
0.2%
0.9%
1.9%
3.4%
60.1-122.5 gms/day
0.4%
0.9%
1.7%
2.1%
over 122.5 gms/day
0.0%
0.0%
0.1%
0.2%
* The percentage of females in an age bracket who consume, on average, a specified amount (grams) of fish per day. The calculations in
this table were based upon the respondents to the NPD survey who consumed fish in the month of the survey. The NPD Research estimates
that these respondents represent, on a weighted basis, 94.0% of the population of U.S. residents (from Table 6, SRI Report, 1980).
4.2.3.2 Human Intake of Methylercury Estimated Through Cross-Sectional Dietary Surveys
General Population, All
Human methylmercury intake from fish for the general U.S. population was estimated in this
Report by combining data on mercury concentrations in fish species (expressed as micrograms of
mercury per gram fresh-weight of fish tissue) with the reported quantities and types of fish species
consumed by fish eaters or "users" in the USDA's Continuing Surveys of Food Intake by Individuals
(CSFH 89/91). When appropriately weighted they can be used to estimate the food consumption
patterns for the general United States population for the period 1989/1991. Because the food
consumption records rely on standard coding of food intake and records of types of fish represented by
a particular dietary record it is possible to estimate how much of particular types of fish were
consumed for the population as a whole and for subpopulations of interest The portion size consumed
by individuals is recorded, as is the person's self-reported body weight.
The dietary survey methodology consisted of an assessment of three consecutive days of food
intake and selection of interviewees from probability samples for non-institutionalized United States
June 1996
4-21
SAB REVIEW DRAFT
-------
households. Survey respondents numbered 11.706 individuals who were surveyed across all four
seasons of the year and all seven days of the week. Respondents also reported their body weights, and
these data were utilized in Volume III to estimate fish consumption on a per body weight basis. Use
of these survey data provides a nationally based estimate of fish intake by the general population of
the United States.
Analysis of CSFH 89/91 data by U.S. EPA's Office of Water
(personal communication from Helen Jacobs) determined confidence intervals around the various
percentiles for total fish and shellfish consumed. The percentile confidence intervals were estimated
using the percentile bootstrap methods with 1,000 bootstrap replications. These estimates are for the
quantity of uncooked fish consumed by all 11,912 individuals who reported consumption data hi the
years 1989, 1990, and 1991 (see Table 4-7). Table 4-8 presents the daily average per capita estimates
of fish consumption based on cooked fish.
Table 4-7
Daily Average Per Capita Estimates of Uncooked Fish Consumption from CSFII 89/91a
Percentile
Mean
90th
95th
99th
Grams/person/day
Estimate
20.08
70.11
102.01
173.18
90% Interval
Lower Bound
18.82
65.37
99.26
162.80
Upper Bound
21.35
74.20
106.67
176.52
a Based on uncooked fish and shellfish. Source of Analysis: OPPTS, U.S. EPA. Means are for total population
of 11,912 individuals surveyed.
Table 4-8
Daily Average Per Capita Estimates of Cooked Fish Consumption
U.S. Population - Finfish and Shellfish
Habitat
Fresh/Estuarine
Grams/Person/Day
Statistic
Mean
50th %
90th %
95th %
99th %
Estimate
5.57
0.00
15.91
39.42
91.67
90% Interval
Lower Bound
4.94
0.00
13.71
37.48
88.92
Upper Bound
6.19
0.00
17.06
41.64
99.61
June 1996
4-22
SAB REVIEW DRAFT
-------
Table 4-8 (continued)
Daily Average Per Capita Estimates of Cooked Fish Consumption
U.S. Population - Finfish and Shellfish
Habitat
Marine
All Fish
Grams/Person/Day
Statistic
Mean
50th %
90th %
95th %
99th %
Mean
50th %
90th %
95th %
99th %
Estimate
12.10
0.00
45.50
69.70
121.33
17.66
0.00
61.47
87.23
152.36
90% Interval
Lower Bound
11.21
0.00
42.43
66.98
113.40
16.58
0.00
58.47
84.00
146.38
Upper Bound
12.99
0.00
49.33
73.91
133.97
18.75
0.00
65.33
89.98
160.65
* Percentile confidence intervals were estimated using the percentile bootstrap method with 1,000 bootstrap replications.
Note: Estimates are projected from a sample of 11,912 individuals to the U.S. population of 242,707,000 using 3-year
combined survey weights.
Source of individual consumption data: combined 1989, 1990, and 1991 USDA Continuing Survey of Food Intakes by
Individuals (CSFI).
The fish component of foods containing fish was calculated using data from the recipe file for release 7 of the USDA's
Nutrient Data Base for individual Food Intake Surveys.
The CSFII 89/91 survey design reflected known sources of variability in estimating dietary
intakes in general. The extent to which comparatively short-term assessments of dietary intake predict
long-term fish consumption patterns remains an uncertainty. Nutritional epidemiologists (among others
see Willet, 1990) have observed that these surveys provide a cross-sectional view of dietary intake that
better predicts central tendency than the extremes of the range of typical fish consumption behavior.
In Volume III comparisons were made between quantities consumed and the upper quartile of the fish-
consuming subpopulation of the general United States population and estimates of quantities of fish
consumed by subpopulations of high fish-consuming Native American Tribes and anglers. Fish
consumption rates reported by several tribes and by high fish-consuming anglers, corroborate the daily
consumption rates of the extreme end of the distribution of the CSFII 89/91 (see also Figure 4-5 in this
volume). Since these individuals are part of the U.S. population, their consumption rates may be
reflected in the CSFII 89/91. It should be noted that the angler and Native American fish consumption
surveys utilize different types of survey methods; this further corroborates the high-end estimates of
CSFII 89/91.
Among nationally representative weighted samples of individuals, 30.9 percent reported
consumption of fish and/or combinations of fish, shellfish, or seafood with starches in a 3-day
sampling period. Of individuals reporting fish consumption, approximately 98 percent consumed fish
only once, and about 2 percent consumed fish in two or more meals during the 3-day survey period.
For foods consumed by only a minority of the population, estimates of per capita consumption .rates
overestimate the consumption rate for the general population, but underestimate the consumption rate
June 1996
4-23
SAB REVIEW DRAFT
-------
among the portion of the population which actually consumes the food item. As a consequence, in
this risk characterization fish consumption estimates are based on a "per consumer" basis.
Respondents in CSFII 89/91 who reported eating fish indicated the fish species and quantity of
fish consumed. Fish consumption has been reported to be recalled with greater accuracy than other
food groups (Karverti and Knuts, 1985). Nevertheless, an uncertainty in these data is the ability of
consumers to identify the species of fish consumed. The species of fish identified by the respondents
were recorded as part of the dietary records of the survey. These fish species were identified and used
to estimate dietary intake of methylmercury. The survey and results are described in Appendix H to
Volume HI.
Selection of a database for mercury residues in fish was determined by several factors.
Because the dietary consumption data were nationally based, preference was given to fish residue data
that included fish from widely diverse geographic areas. This choice aimed at avoiding data more
appropriate for site-specific assessments that would result from the sampling of a very limited
geographic area (e.g., a state-wide survey). In addition, the preferred database should include many
individual types of fish to represent the consumed species; inclusion of many species types is
considered to provide a better approximation of the estimate of central tendency from individual
samples. A third preference was for data collected over a time period that approximated the years of
the dietary survey. The third criteria was judged the least important in selection of the database
because residues of mercury in soils and sediments continue to contribute mercury to the aquatic food-
web and are considered to minimize year-to-year variability in mercury concentrations in fish tissues.
Data describing methylmercury concentrations in marine fish were largely based on the National
Marine Fisheries Service (NMFS) database, the largest publicly available database on mercury
concentrations in marine fish. This NMFS database has been collected over the past two decades.
Comparison of the values for central tendency (e.g., 50th percentile) in mercury concentrations
between the NMFS database and FDA's compliance data on selected species (Carrington et al., 1995)
indicated close agreement in mercury concentrations. Table 4-9 presents data on mercury
concentrations in fish used in calculations for this report, as well as concentrations cited by FDA
(1978) and by Stern et al. (1996.)
«
For freshwater fish, two publications reporting mercury concentrations in multiple species of
fish were chosen. Data reported by Bahnick et al. (1994) and Lowe et al. (1985) were used to
estimate average mercury concentrations in fresh-water finfish from across the United States. Both
Lowe et al. (1985) and Bahnick et al. (1994) used nationwide surveys of freshwater fish. Both
databases suffer from a limited number of samples at any one site; however, a strength of each data set
is that many sites were sampled. These data are described hi detail in Appendix H of Volume HI.
When average mercury concentration in fresh-water fish are compared, the results from Lowe et al.
(1985) and Bahnick et al. (1994) differ by a factor of approximately two. Consequently separate
analyses of methylmercury intake from fish were prepared to assess the impact of the database chosen
for mercury residues.
June 1996 . 4-24 SAB REVIEW DRAFT
-------
Table 4-9
Summary of Mercury Concentrations in Fish Species
Micrograms Mercury per Gram Fresh Weight (\ig Hg/g)
Data Used by USEPAa
Mercury Study
Report to Congress
In Review
Fish Species
Abalone
Anchovies
Bass, Freshwater
Bass, Sea
Bluefish
Bluegills
Bonito
Bonito
Butterfish
Carp,
Common
Catfish
(channeljarge
mouth, rock,
striped, white)
Catfish
(Marine)
dams
Cod
Crab, King
Average
(MS Hg/g)
0.016
0.047
Avgs.= 0.157
(Lowe et aL,
1985) and 0.38
(Bahnick et at.,
1994)
Not Reported
Not Reported
0.033
Not Reported
Not Reported
Not Reported
0.093
0.088
Not Reported
0.023
0.121
0.070;
Calculations
based on 5
species of crab
combined at
0.117
Data Used by US FDAb
Report on the Chance of U.S.
Seafood Consumers Exceeding "The Current
Daily Intake for Mercury and Recommended
Regulatory Controls"
1978
Fish
Species
Abalone
Anchovies
Bass,
Striped
Bass, Sea
Bluefish
Bluegills
Bonito
(below 3197)
Bonito
(above 3197)
Butterfish
Caip
Catfish
(freshwater)
Catfish
(Marine)
Clams
Cod
Crab, King
Average
(MS Hg/g)
0.018
0.039
0.752
* 0.157
0.370
0.259
0.302
0.382
0.021
0.181
0.146
0.475
0.049
0.125
0.070
Maximum
(UgHg/g)
0.120
0.210
2.000
0.575
1.255
1.010
0.470
0.740
0.190
0.540 -
0.380
1.200
0.260
0.590
0.240
Data Used by
State of
New Jersey
Fish
Species
Not
Reported
(NR)
NR
Bass,
freshwater
Sea Bass
Bluefish
NR
NR
NR
Butterfish
Catfish,
freshwater
Cams
Cod/Scrod
See crab.
Crab
NR
Average
(fig Hg/g)
0.41
0.25
0.35
0.05
0.15
0.05
0.15
0.15
June 1996
4-25
SAB REVIEW DRAFT
-------
Table 4-9 (continued)
Summary of Mercury Concentrations in Fish Species
Micrograms Mercury per Gram Fresh Weight (ug Hg/g)
Data Used by USEPA*
Mercury Study
Report to Congress
In Review
Crab
Crappie
(black, white)
Croaker
Dolphin
Drums,
Freshwater
Flounders
Groupers
Haddock
Hake
Halibut
Halibut
Halibut
Halibut
Herring
Kingfish
Lobster
Lobster
Lobster
Spiny
Mackerel
Mackerel
0.117
0.114
0.125
Not Reported
0.117
0.092
0.089
0.145
0.250
0.250
0.250
0.250
0.013
0.100
0.232
0.232
0.232;
Includes spiny
(Pacific)
lobster=0.210
0.081;
Averaged Chub
= 0.081;
Atlantic= 0.025;
Jack=0.138
0.081
Data Used by US FDAb
Reoort on the Chance of U.S.
Seafood Consumers Exceeding "The Current
Daily Intake for Mercury and Recommended
Regulatory Controls"
1978
Crab, other
than HI
Crappie
Croaker
Dolphin
Drums
Flounders
Groupers
Haddock
Hake
Halibut 4
Halibut 3
Halibut 2H
Halibut 25
Herring
Kingfish
Lobster,
Northern 11
Lobster
Northern 10
Lobster,Spiny
Mackerel,
Atlantic
Mackerel,
Jack
0.140
0.262
0.124
0.144
0.150
0.096
0.595
0.109
0.100
0.187
0.284
0.440
0.534
0.023
0.078
0.339
0.509
0.113
0.048
0.267
0.610
1.390
0.810
0.530
0.800
0.880
2.450
0.368
1.100
1.000
1.260
1.460
1.430
0.260
0.330
1.603
2.310
0.370
0.190
0.510
Data Used by
State of
New Jersey
NR
NR
NR
Dolphin
(Mahi-mahi)
NR
Flounder
NR
Haddock
Hake
Halibut
Halibut
Halibut
Halibut
Herring
Kingfish
Lobster
Lobster
Lobster
Mackerel
Mackerel
0.25
0.10
0.05
0.10
0.25
0.25
0.25
0.25
0.05
0.05
0.25
0.25
0.25
0.28
0.28
June 1996
4-26
SAB REVIEW DRAFT
-------
Table 4-9 (continued)
Summary of Mercury Concentrations in Fish Species
Micrograms Mercury per Gram Fresh Weight (jag Hg/g)
Data Used by USEPAa
Mercury Study
Report to Congress
In Review
Mackerel
Mackerel
Mackerel
Mackerel '
Mullet
Oysters
Perch,
White and
Yellow
Perch,
Ocean
Pike,
Northern
Pollock
Pompano
Rockfish
Sablefish
Salmon
Scallops
Scup
Sharks
Shrimp
Smelt
Snapper
Snapper
Snook
Spot
0.081
0.081
0.081
0.081
0.009
0.023
0.110
0.116
0.310
0.127
0.150
0.104
Not Reported
Not Reported
0.035
0.042
Not Reported
1.327
0.047
0.100
0.25
0.25
Not Reported
Not Reported
Data Used by US FDAb
Report on the Chance of US.
Seafood Consumers Exceeding "The Current
Daily Intake for Mercury and Recommended
Regulatory Controls"
1978
Mackerel,
King (Gulf)
Mackerel,
King (other)
Mackerel,
Spanish 16
Mackerel,
Spanish 10
Mullet
Oysters
Perch,
Freshwater
Perch,
Marine
Pike
Pollock
Pompano
Rockfish
Sablefish
Salmon
Scallops
Scup
Sharks
Shrimp
Smelt
Snapper.Red
Snapper,
Other
Snook
Spot
0.823
1.128
0.542
. 0.825
0.016
0.027
0.290
0.133
0.810
0.141
0.104
0.340
0.201
0.040
0.058
0.106
1.244
0.040
0.016
0.454
0.362
0.701
0.041
2.730
2.900
2.470
1.605
0.280
0.460
0.880
0.590
1.710
0.960
8.420
0.930
0.700
0.210
0.220
0.520
4.528
0.440
0.058
2.170
1.840
1.640
0.180
Data Used by
State of
New Jersey
Mackerel
Mackerel
Mackerel
Mackerel
Mullet
NR
Perch
NR
NR
NR
NR
NR
NR
Salmon
NR
NR
Shark
Shrimp
Smelts
Snapper
Snapper
NR
Spotfish
0.28
^
0.28
0.28
0.28
0.05
0.18
0.05
1.11
0.11
0.05
0.31
0.31
0.05
June 1996
4-27
SAB REVIEW DRAFT
-------
Table 4-9 (continued)
Summary of Mercury Concentrations in Fish Species
Micrograms Mercury per Gram Fresh Weight (ug Hg/g)
Data Used by USEPA"
Mercury Study
Report to Congress
In Review
Squid
Octopi
Sunfish
Swordfish
Tillefish
Trout
Trout
Tuna
Tuna
Tuna
Whitefish
Other finfish
Other shellfish
0.026
0.029
Not Reported
0.95
Not Reported
0.149
0.149
0.206;
Averaged:
Tuna, light
skipjack=0.136
Tunajight
yellow=0.218;
Albacore=0.264
0.206
0.206
Not Reported
Not
Reported
Data Used by US FDAb
Report on the Chance of U.S.
Seafood Consumers Exceeding "The Current
Daily Intake for Mercury and Recommended
Regulatory Controls"
1978
Squid and
Octopi
Squid and
Octopi
Sunfish
Swordfish
' Tillefish
Trout,
Freshwater
Trout,
Marine
Tuna,
Light
Skipjack
Tuna,
Light Yellow
Tuna, White
Whitefish
Other finfish
0.031
0.031
0.312
1.218
1.607
0.417
0.212
0.144
0.271
0.350
0.054
0.287
0.400
0.400
1.200
2.720
3.730
1.220
1.190
0.385
0.870
0.904
0.230
1.020
Data Used by
State of
New Jersey
Squid
NR
NR
Swordfish
NR
Trout
Trout
Tuna,
fresh
Tuna,
fresh
Tuna,
fresh
Whitefish
Finfish,
other
Shellfish,
other
0.05
0.93
0.05
0.05
0.17
0.17
0.17
0.04
0.17
0.12
Fish Species (Freshwater) Not Reported by FDA, 1978
Bloater
Smallmouth
Buffalo
Northern
Squawfish
Sauger
0.0.93
0.096
0.33
0.23
June 1996
4-28
SAB REVIEW DRAFT
-------
Table 4-9 (continued)
Summary of Mercury Concentrations in Fish Species
Micrograms Mercury per Gram Fresh Weight (ug Hg/g)
Data Used by USEPAa
Mercury Study
Report to Congress
In Review
Sucker
Walleye
Trout (brown,
lake, rainbow)
0.114 (Lowe et
aL, 1985;
0.167 (Bahnick
et aL, 1994).
0.100 (Lowe et
aL. 1985) and
0.52 (Bahnick
et al., 1994).
0.149 (Lowe et
al., 1985) and
0.14 (Bahnick
et al., 1994 for
brown trout).
Data Used by US FDAb
Report on the Chance of VS.
Seafood Consumers Exceeding "The Current
Daily Intake for Mercury and Recommended
Regulatory Controls"
1978
-
Data Used by
State of
New Jersey
Fish Species Reported by the State of New Jersey
and Not Reported by EPA or FDA
Blowflsh
Orange roughy
Sole
Weakfish
Porgy
Blackfish
Whiting
Turbot
Sardines
Tilapia
0.05
0.5
0.12
0.15
0.55
0.25
0.05
0.10
0.05
0.05
a From Appendix H, Volume UL
b FDA, 1978.
c Stern et al., 1996 - in press.
June 1996
4-29
SAB REVIEW DRAFT
-------
Findings of this evaluation based on dietary patterns for the general U.S. population and
mercury residue data in fish were used to estimate mercury consumption from fish. Data on fish
consumption (in grams per day) are shown for persons who reported consuming fish at least once in
the three-day sampling period. These estimates are based on the total grams of fish or shellfish
consumed within the 72-hour sampling period which is then averaged over the three 24-hour periods.
For example, if a person reported consuming 30 grams of fish during the 72-hour period, this persons
24-hour average would be 10 grams. The mercury concentrations for particular fish and shellfish
species used in these calculations is shown in Table 4-9 (columns under data used by U.S. EPA).
Consumption of methylmercury (expressed per kilogram self-reported body weight) from
marine and freshwater finfish and shellfish for the general U.S. fish-consuming population were
estimated; the data are shown in Tables 4-10 through 4-13. Tables 4-12 and 4-13 give estimated
consumption of fish species designated as freshwater fish. For comparison, methylmercury intake
from all fish and shellfish for the general U.S. population are shown in Tables 4-10 and 4-11. The
most commonly consumed marine finfish for the general population is tuna. Use of the data of
Bahnick et al. (1994) which reported overall higher mercury concentrations than those of Lowe et al.
(1985) provided a range of estimates of methylmercury intake for the general U.S. fish-consuming
population.
Methylmercury intakes calculated and shown in Appendix H of Volume III have been
developed for a nationally based rather than a site-specific assessment. The CSFII 89/91 data from
USDA was designed to be representative of the U.S. population. The concentrations of methylmercury
in marine fish and shellfish were derived from a database that is national in scope and the data on
fresh-water finfish were from two large studies that sampled fish at a number of sites throughout the
United states. The applicability of these data to site-specific or region-specific assessments must be
judged on a case-by-case basis.
Subgroups of General Population
In selection of sensitive subpopulations of humans, sensitivity may reflect inherent
responsiveness of the subpopulation to the hazard (i.e., toxicity based) or reflect elevated exposures to
the agent of concern. With respect to risks posed by methylmercury from fish, two subpopulations of
humans are of particular interest in this risk characterization: women of child-bearing age and
childrea Women of child-bearing age are of concern because developmental effects following in utero
exposures are the basis for the RfD and because the developing nervous system would be expected to
be most vulnerable to methylmercury toxicity. Because 9.5 percent of women ages 15 through 44
years are pregnant in a given year, and the half-life of mercury averages 70 days, the entire population
of women of child-bearing age is judged to be of concern.
Estimated methylmercury intake from finfish and shellfish of both marine and freshwater
origin for women of child-bearing age are presented hi Tables 4-10 and 4-11. The data of Bahnick et
al. (1994) were used to calculate the estimates shown in Table 4-10. The estimated exposures in
Bahnick et al. 1994 are generally higher hi methylmercury concentration than are those of Lowe et
al. (1985) which were used to present the estimates shown in Table 4-11. Because the quantities of
freshwater fish consumed are a small part of the total fish intake for the general population,
differences in average mercury concentrations between the data of Lowe et al. (1985) and Bahnick et
al (1994) do not result in marked differences in average mercury intake from fish. Tables 4-12 and 4-
13 present estimated methylmercury intakes from freshwater fish for U.S. women of child-bearing age.
June 1996 4-30 SAB REVIEW DRAFT
-------
Table 4-10
Consumption of All Fish & Shellfish (gms/day) and Methylmercury per Kg body weight from Fish among Respondents of the
1989-1991 CSFII Survey. Data for "Users" Only. Bahnick et al. estimates for fresh-water fish Methylmercury Concentrations"
Gender
Males
Females
Percentage
Min
5th
25th
50th
75lh
95th
97.5th
99th
Max
Min
5th
25th
50th
75th
95th
97.5
99th
Max
Aged 14 Years of Younger
N
380
340
Fish
gms/day
1.9
3.5
14.2
23.0
45.0
80.6
104.5
132.4
139.4
1.0
4.7
14.0
23.0
37.5
67.4
112.1
113.8
153.5
Methyl- .
mercury
"g/kgbw
<0.01
0.01
0.09
0.16
0.25
0.84
0.94
1.33
1.51
<0.01
0.02
0.07
0.15
0.28
0.81
0.90
1.29
1.69
Aged 15 through 44 Years
N
646
864
Fish
gms/day
1.9
9.3
24.6
45.5
73.3
124.5
153.3
177.9
312.0
1.3
7.0
18.6
28.6
55.8
110.8
132.8
174.6
461.0
Methyl-
mercury
Mgfcgbw
< 0.01
0.01
0.04
0.08
0.16
0.33
0.46
0.60
1.98
<0.01
0.01
0.04
0.08
0.16
0.35
0.48
0.74
2.76
Aged 45 Years or Older
N
556
828
Fish
gms/day
2.5
9.3
22.7
40.3
64.5
127.2
172.8
203.1
388.9
1.9
7.2
18.9
31.8
56.0
107.8
134.1
159.0
250.2
Methyl-
mercury
MgfcEbw
<0.01
0.01
0.03
0.07
0.13
0.38
0.71
0.95
1.57
<0.01
0.01
0.04
0.07
0.14
0.34
0.55
0.63
1.67
Total
N
1582
2032
Fish
gms/day
1.9
7.0
20.4
37.8
65.3
120.8
149.2
177.2
388.9
1.0
7.0
18.6
28.6
52.3
106.7
127.4
161.3
461.0
Methyl-
mercury
Mg/kBbw
<0.01
0.01
0.04
0.09
0.17
0.45
0.64
0.94
1.98
<0.01
0.01
0.04
0.09
0.17
0.44
0.60
0.90
2.76
Data weighted to reflect U.S. population.
June 1996
4-31
SAB REVIEW DRAFT
-------
Table 4-11
Consumption of All Fish & Shellflsh (gms/day) atod Methylmercury per Kg body weight from Fish among Respondents of the 1989-
1991 CSFII Survey. Data for "Users" Only. Lowe et al. estimates for fresh-water fish Methylmercury Concentrations'*
Gender
Males
Females
Percentage
Min
5th
25lh
50
-------
Table 4-12
Consumption of Freshwater Fish (gms/day) and Methylmercury per Kg body weight from Fish among Respondents of
the 1989-1991 CSFII Survey. Data for "Users" Only. Fish methylmercury concentrations based on Bahnick et al., (1994)a
Gender
Males
Females
Percentage
Mm
5th
25th
50th
75th
95th
97.5th
99th
Max
Min
5lh
25th
50th
75th
95th
97.5th
99th
Max
Aged 14 Years of Younger
N
60
46
Fish
gms/day
4.0
8.8
17.2
26.2
37.5
57.0
78.7
86.4
96.5
1.0
2.5
13.8
14.1
37.5
43.7
63.3
71.0
71.0
Methyl-
mercury
HS/kBbw
0.04
0.07
0.15
0.19
0.24
0.55
0.68
0.77
0.77
0.02
0.05
0.12
0.16
0.19
0.64
0.65
0.87
0.87
Aged 15 through 44 Years
N
B
80
109
Fish
gms/day
7.6
19.0
42.9
58.5
77.1
140.6
224.4
224.4
247.8
8.8
18.1
23.4
37.5
56.9
178.3
178.3
213.2
217.8
Methyl-
mercury
Mg^Sbw
0.03
0.06
0.11
0.16
0.24
0.50
0.52
0.55
0.64
0.02
0.05
0.08
0.11
0.19
0.68
0.68
0.71
0.73
Aged 45 Years or Older
N
82
115
Fish
gms/day
3.1
18.1
23.4
43.0
63.7
147.6
172.9
388.9
388.9
2.0
13.8
22.4
34.0
67.5
102.0
172.9
172.9
222.1
Methyl-
mercury
MB/kgbw
0.01
0.04
0.09
0.11
0.17
0.35
0.45
0.71
0.71
0.01
0.04
0.07
0.12
0.2
0.39
0.55
0.55
0.55
Total
N
222
270
Fish
gms/day
3.1
12.7
26.0
44.7
72.4
134.0
172.9
224.4
388.9
1.0
10.7
21.4
31.2
56.2
118.9
172.9
178.3
222.1
Methyl-
mercury
Mg^Shw
0.01
0.05
0.10
0.16
0.24
0.46
0.55
0.71
0.77
0.01
0.04
0.08
0.13
0.19
0.55
0.65
0.68
0.87
Data are weighted to be representative of the U.S. population.
June 1996
4-33
SAB REVIEW DRAFT
-------
Table 4-13
Consumption of Freshwater Fish (gms/day) and Methylmercury per Kg body weight from Fish among Respondents of the 1989-1991
CSFII Survey. Data for "Users" Only. Fish methylmercury concentrations based on Lowe et al., (1994)
Gender
Males
Females
Percentage
Min
5th
25th
50th
75th
95th
97.5th
99th
Max
Min
5th
25th
50th
75th
95th
97.5th
99th
Max
Aged 14 Years of Younger
N
60
46
Fish
gms/day
4.01
8.8
17.2
26.2
37.5
57.0
78.7
86.4
96.5
1.0
2.5
13.8
14.1
37.5
43.7
63.3
71.0
71.0
Methyl-
mercury
HB/kBbw
0.02
0.05
0.08
0.11
0.23
0.30
0.30
0.42
0.53
0.01
0.02
0.06
0.07
0.16
0.33
0.36
1.06
1.06
Aged 15 through 44 Years
N
80
109
Fish
gms/day
7.6
19.0
42.8
58.5
77.1
140.6
224.4
224.4
247.8
8.8
18.1
23.4
37.5
57.9
178.3
178.3
213.2
217.8
Methyl-
mercury
MB^Bbw
0.01
0.03
0.07
0.11
0.17
0.27
0.27
0.33
0.66
0.02
0.03
0.04
0.09
0.13
0.29
0.30
0.40
0.69
Aged 45 Years or Older
N
82
115
Fish
gms/day
3.1
18.1
1
23.4
43.0
63.7
147.6
172.9
388.9
388.9
2.0
13.8
22.4
34.0
67.5
102.0
172.9
172.9
222.1
Methyl-
mercury
HB^Bbw
0.01
0.02
0.04
0.07
0.13
0.26
0,39
0.40
0.44
<0.01
0.02
0.04
0.07
0.13
0.25'
0.30
0.31
0.33
Total
N
222
270
Fish
gms/day
3.1
12.7
26.0
44.7
72.4
134.0
172.9
224.4
388.9
1.0
10.7
21.4
31.2
56.2
118.9
172.9
178.3
222.1
Methyl-
mercury
"g^Bbw
0.010
0.03
0.06
0.090
0.17
0.30
0.33
0.42
0.66
<0.01
0.02
0.05
0.07
0.13
0.29
0.30
0.36
1.06
Data are weighted to be representative of the U.S. population.
June 1996
4-34
SAB REVIEW DRAFF
-------
The second subpopulation identified in this risk characterization to be of concern consists of
children aged 14 years and younger. Although it would be of interest to have subdivided the groups
of children into at least two groups (e.g., ages 6 and younger; ages 7 through 14 years), data on
children ages 14 and younger are presented as a whole. No age-based subdivision was calculated
because of the limitations in the size of various categories. The basis for concern about children is
that intake of methylmercury from fish is greater than for adults when expressed on a per kilogram
body weight basis. When the methylmercury intake is expressed on a per kilogram body weight basis,
the exposure for children aged 14 years and younger is approximately two-to-three times that of the
adult. These data for children are presented in Tables 4-10 through 4-13. The higher estimated
exposure to methylmercury is the result of the higher intake of food on a per kilogram weight basis
among childrea A major uncertainty identified in this risk characterization is the absence of data to
assess health hazards of methylmercury for children who have low methyunercury exposures in utero.
One strength of the data from CSFn 89/91 is that individual body weight data were available
and were utilized in calculation of dietary intakes on a per kilogram body weight basis. Consequently
actual body weights rather than default values were utilized to estimate methylmercury exposure per
kilogram body weight. The methylmercury intakes of adult males and females are comparable on a
body weight basis. The maximum intakes on a per kilogram body weight basis are also provided for
each group considered. The intakes for the high-end fish consumer (the maximum reported in each
group of adults) is at least four times the intake for the individual at the 95th percentile.
Methyunercury intake from fish estimated in this way does not permit attribution to the
anthropogenic or "natural" sources of mercury. Because of the magnitude of anthropogenic, ambient
mercury contamination, the estimates of methyunercury from fish do not provide a "background"
value. "Background" values imply an exposure against which the increments of anthropogenic activity
could be added. This is not the situation due to release of substantial quantities of mercury into the
environment through human activities and to recycling of this previously released mercury.
43 Estimates of Sizes of At-Risk Populations
4.3.1 Human Populations
4.3.1.1 Number of Human Subjects in At-Risk Subpopulations in the United States
The number of human subjects potentially at risk of adverse effects from exposure to
methylmercury depends on the health-based endpoint(s) used in the risk assessment, and characteristics
of exposure to methylmercury (e.g., quantity of fish consumed, concentration of mercury in the fish
consumed). If paresthesias are the health-based endpoint of concern any adult male or female can be
considered potentially at-risk depending on the quantity and type of fish consumed. The total
population of the United States aged 15 years or greater is approximately 194,858,000 based on the
1990 United States Census data. The male population in these age groups numbers approximately
93,669,000. The female population in these ages numbers approximately 101,187,000. Approximately
30% of the adult population in the U.S. consumes fish on a regular basis.
The risk of paresthesia for children is difficult to estimate because of serious limitations of
data on effects of methylmercury exposure among children who were not exposed in utero. Initial
epidemiology investigations in Minamata and Niigata, Japan, where chronic exposure to
methylmercury contaminated fish was the source of mercury exposure, indicated that the highest
frequency of disease was observed among subjects ages 20-59 years. Fish consumption among
subjects in the age category birth to 10 years of age was lower than for older subjects (see page 64,
June 1996 4-35 SAB REVIEW DRAFT
-------
Tsubaki and Irukayama, 1977). Cases of fatal Minamata disease, however, included six children (ages
2.5,4.5, 5.0, 6.4, 7 and 8 years) among 38 cases (pages 132-133, Tsubaki and Irukayama, 1977).
Because the methylmercury contamination in Minamata area existed for a number of years clear
separation of prenatal from postnatal exposure is not possible. Harada (1977, page 220 in Tsubaki and
Irukayama, 1977) provided an analysis of the .frequency of occurrence of various symptoms and signs
hi Minamata disease. Adults had a 100 percent incidence of paresthesia. Occurrence of paresthesia
among congenital cases and children was considered not clear, but Harada noted that all patients had a
sensation of pain.
Children were also affected by methylmercury poisoning hi the Iraq epidemic. Rustam and
Hamdi (1974) included the age groups "birth through 10 years" and "11 through 20 years" hi the
patients they evaluated hi a neurological study of methylmercury poisoning hi Iraq. The pediatric
patients were not cases of in utero exposure because the youngest of this group was identified as 5
years of age. In their discussion of individual variation hi response to mercury, Rustam and Hamdi
observed that "in general, younger patients suffered heavier damage than the older ones" (page 509, *
Rustam and Hamdi, 1974).
Exposure patterns for children (see Volume HI and Section 4.7 of Volume VI) suggest that
their exposure to methylmercury on a "per kilogram body weight" basis is much higher compared with
adult exposures. Neuronal migration, a process specifically affected by methylmercury, begins at
about six weeks in utero, and that process continues until five months after birth (Chi et al., 1977).
Considering the broad-based impairment of nervous system metabolism that can be produced by
methylmercury (among others see Atchison and Hare, 1994); that nervous system development
continues post-natally through at least the third to fourth year of life [visual connections are complete
around 3 to 4 years of age (Hohman and Creutzfeld, 1975)]; and that the human brain is not fully
mature hi form until approximately age 20 (Rodier, 1994) children may be more sensitive to adverse
sensory-motor effects of methylmercury than are adults. If children are arbitrarily defined as persons
aged less than 15 years, the U.S. population is approximately 53,853,000 based on 1990 census data
(Table 4-14).
Table 4-14
Resident Population of the United States and Divisions, April 1, 1990 Census
by Gender and Age; in Thousands, including Armed Forces Residing in Region
Division/Gender
United States
Male
Female
Percent Female
Total
248,710
121,239
127,471
51.3
<15 Years of Age
53,853
27,570
26,284
48.8
15-44 Years of
Age
117,610
58,989
58,620
49.8
>45 Years of
Age
77,248
34,680
42,567
55.1
Developmental endpoints have also been used to establish the critical effects for
methylmercury. Estimates of the size of the population of women of reproductive age, number of live
June 1996
4-36
SAB REVIEW DRAFT
-------
births, number of fetal deaths, and number of legal abortions can be used to predict the percent of the
population and number of women of reproductive age who are pregnant in a given year. This
methodology has been previously used in the Agency for Toxic Substances and Disease Registry's
(ATSDR's) Report to Congress on The Nature'and Extent of Lead Poisoning in Children in the United
States (Mushak and Crocetti, 1990). To estimate the size of this population on a national basis Vital
and Health Statistics data for number of live births (National Center for Health Statistics of the United
States, 1990; Volume I, Natality, Table 1-60, pages 134-140), and fetal deaths (National Center for
Health Statistics of the United States, 1990; Volume II, Mortality; Table 3-10, pages 16, 18, and 20).
The incidence of fetal wastage, that is, spontaneous abortions prior to 20 weeks of gestation was not
considered since no systematically collected, nationally based data exist.
The estimate of number of women of child-bearing age includes some proportion of women
who will never experience pregnancy. However, substitution of the number of pregnancies in a given
year provides some measure of assessing the size of the surrogate population at risk. Estimates of the
size of the population were based on "Estimates of Resident Population of the United States Regions
and Divisions by Age and Sex" (Byerly, 1993). The Census data for 1990 were grouped by age and
gender. The sizes of these populations for the contiguous U.S. are shown in Table 4-15.
Women ages 15 through 44 are the age group of greatest interest in identifying a
subpopulation of concern for the effects of a developmental toxin such as methylmercury. This
population consisted of 58,222,000 women living within the contiguous United States (Table 4-15).
This population was chosen rather than for the total United States (population 58,620,000 women ages
15 through 44 years) because the dietary survey information from CSFII/89-91 did not include Hawaii
and Alaska. Based on estimates of fish consumption data for Alaska by Nobmann et al. (1992) the
quantities of fish eaten by Alaskans exceeds those of the contiguous U.S. population. It is also
estimated that residents of the Hawaiian Islands have fish consumption patterns that differ from those
of the contiguous United States.
The number of women of child-bearing age (15 through 44 years) in Alaska is estimated to be
approximately 138,000 and in Hawaii, the approximate number is 284,000 women. The percentage of
pregnant women for ages 15 through 44 eyars in Alaska was 9.1 percent, and in Hawaii this
percentage was also 9.7 percent.
Table 4-15
Resident Population of the Contiguous United States, April 1, 1990 Census
by Gender and Age; in Thousands, including Armed Forces Residing in Region
Division/Gender
Contiguous United
States
Male
Female
Percent Female
Total
247,052
120,385
126,667
51.3
<15 Years of Age
53,462
27,369
26,094
48.8
15-44 Years of
Age
116,772
58,548
58,222
49.9
>45 Years of
Age
76,817
34,467
42,348
55.1
June 1996
4-37
SAB REVIEW DRAFT
-------
The number of pregnancies per year was estimated by combining the number of live births,
number of fetal deaths (past 20 weeks of gestation) and the number of legal abortions. The legal
abortion data were based on information published by Koonin et al. (1993) in Morbidity and Mortality
Weekly Report. These totals are presented in Table 4-16. As noted in this table, the total of legal
abortions includes those with unknown age which were not included in the body of each table entry.
There were 2,929 such cases for the United States in 1990 or 0.2 percent of all legal abortions.
Another complication in the legal abortion data was for the age group 45 and older. The available
data provide abortion data for 40 years and older only. To estimate the size of the population older
than 45 years, the number of legal abortions for women age 40 years and older were allocated by
using the proportions of Live Births and Fetal Deaths for the two age groups 40-44 and 45 and older.
Table 4-16
Pregnancies by Outcome for Resident Females by Divisions and States,
U.S. 1990, by Age*
United States
Contiguous
United States
Females
Lave births
Fetal deaths
Legal abortions
Total pregnancies
Percent Pregnant
Females
Live births
Fetal deaths
Legal abortions
Total pregnancies
Percent Pregnant
Totalb
127,471,000
4,158,212
31,386
1,429,577
5,619,175
-
126,667,000
4,125,821
31,183
1,423,340
5,580,344
-
<15 Years
26,284,000
11,657
174
11,819
23,650
-
26,094,000
11,615
173
11,765
23,553
-
15-44 Years
58,620,000
4,144,917
31,176
1,413,992
5,590,085
9.5
58,222,000
4,112,579
30,974
1,407,830
5,551,383
9.5
>45 Years0
42,567,000
1,638
36
837
2,511
-
42,348,000
1,627
36
833
2,496
-
a Data sources: Byerly ER, State Population Estimates by Age and Sex: 1980-1992, U.S. Bureau of the Census. National Center for Health
Statistics of the U.S. Vol. I. Natality, Vol. H. Mortality, 1990. Koonin et al. Abortion Surveillance - US, 1990: MMWR 42:29-57, 1993.
Total of legal abortions includes those with unknown age which are not included in the body of each table entry. There were 2929 such
cases for the U.S. or 0.2 percent of all legal abortions.
c Cited sources provided abortion data for 40 years and older only. These were allocated by using the proportion of live births and fetal
deaths for the two age groups 40-44 and 45 and older.
It was estimated that within the contiguous United States 9.5 percent of women ages 15
through 44 years were pregnant hi a given year. The total number of live births reported in 1990 for
June 1996
4-38
SAB REVIEW DRAFT
-------
this age group was 4,112,579 with 30,974 reported fetal deaths and 1,407,830 reported legal abortions.
The estimated number of total pregnancies for women ages 15 through 44 years was 5,551,383 in a
population of 58,222,000 women (Table 4-16).
4.3.1.2 Populations Associated with High Consumption of Fish
Exposure to methylmercury depends on quantities of fish consumed, mercury concentrations in
the fish chosen, and duration of these consumption patterns. Three types of data were used to predict
the size of the adult female population potentially at-risk. In Section 5.3 of this Volume, comparison
are made between exposures and recommendations by the World Health Organization that women
consuming 100 grams or more of fish per day should be evaluated for the extent of methylmercury
exposures.
Estimated Population Based on Longitudinal Data from NPD. Inc. 1973/74
Using data from NPD, Inc. obtained in 1973 and 1974, it was estimated that 94% of women
consumed fish or shellfish at least once in a one-month period. If 9.5% of the female population is
pregnant every year, then from these data it is estimated that 8.9% of pregnant women consume fish at
least once per month. Among these consumers the 99th percentile consumption was 112 grams/day
for adult men and women between the ages 18 and 98 years (Rupp et al., 1980). For female subjects
estimated intake of fish at the upper ranges of intakes is shown in Table 4-17.
Table 4-17
Number of Pregnant Women Consuming Fish
at Various Intake Levels Based on Data from NPD,Inc. 1973/74
and 1990 U.S. Census Data
Estimated Quantity of Fish Consumed
> 60 grams/day
At the 99th Percentile or 112 grains of fish and shellfish per day -
> 120 grams/day
Estimated Number of Pregnant Women
55,785
51,867
1,484
Estimated Population Based on Cross-Sectional Data
The number of women of child-bearing age in the United States estimated to consume fish in
excess of 100 grams per day can be obtained by inference from the general U.S. population dietary
surveys (e.g., United States Department of agriculture's CSFII 89/91). Intake of fish and shellfish for
the general U.S. population (estimated by CSFH 89/91 data described in Appendix H, Volume ID)
based on "users" only was estimated to be 110 grams per day at the 95th percentile among women
ages 15 through 44 years. Female children (aged 14 years or younger) consumed 112 grams of fish
per day, at the 95th percentile while the overall intake at the 95th percentile among female subjects
regardless of age was 107 grams per day. Estimates of the number of women of child-bearing age
(ages 15 through 44 years) and the number of children (ages birth through 14 years) within the United
States population is shown in Table 4-18.
June 1996
4-39
SAB REVIEW DRAFT
-------
Table 4-18
Estimated United States Population Consuming Fish, Excluding Alaska and Hawaii
Estimates Based on the 1990 U.S. Census and the Continuing Surveys of Food Intake
by Individuals, 1989/1991
Population Group
Total U.S. Population
Total Female Population Aged 15 through 44 Years
Total Population of Children Aged <15 Years
Estimated Number of Persons
247,052,000
58,222,000
53,463,000
Percent of Respective Group
Reporting Fish Consumption during
the 3-Day Dietary Survey Period in CSFU 89/91*
Total Population
Females Aged 15 through 44 Years
Children Aged <15 Years
30.9 percent
30.5 percent
24.9 percent
Number of Persons Predicted to Consume Fish Based
on Percentage Consuming Fish in CSFII 89/91
Total Estimated Population
Total Estimated Number of Females Aged 15 through 44 Years
Total Estimated Number of Children Aged <15 Years
76,273,000
17,731,000
13,306,000
Number of Persons in
Highest 5 Percent of Estimated Population that Consumes Fish
Total Estimated Population
Total Estimated Female Population Aged 15 through 44 Years
Total Estimated Child Population
3,814,000
887,000
665,000
Estimated Number of Adult Pregnant Women in Highest 5 Percent Of Estimated Population that Consumes Fish
Number of Females Aged 15 through 44 Years x Percentage of Women Pregnant
in a Given Year
84,300
* Rounded to three significant figures.
The estimated number of women of child-bearing age (ages 15 through 44 years) in the
contiguous 48 states is approximately 17,731,000. It is estimated that in a given year 9.5% of
women in this age group are pregnant The estimated number of women ages 15 through 44 years in
the highest 5 percent of consumers identified hi a cross sectional survey with a 3-day sampling
window is 887,000. The estimated number of pregnant women in that highest 5 percent of fish
consumers is estimated to be approximately 84,300; this is the number of pregnant women expected to
consume an average of 100 g fish/day The type of fish consumed and the amount of mercury in those
fish will affect the over all mercury exposure.
June 1996
4-40
SAB REVIEW DRAFT
-------
Specific Subpopulations of Anglers and Subsistence Fishers
Specific subpopulations of anglers and subsistence fishers ingest fish substantially in excess of
the consumption by the general population. Fish consumption by these groups, by comparison to the
general U.S. population, is shown in Figure 4-5. For example, Puffer et al. (1981) in a study of
anglers in Los Angeles, California found a mean intake was 37 grams per day, but the 90th percentile
for this group was 225. grams per day. Orientals and Samoans had mean fish intakes of 70.6
grams/day (Puffer et al., 1981). Alaskan Natives from 11 communities averaged 109 grams of
fish/day (Nobbman et al., 1992). Wolfe and Walker identified a very high fish consumption rate
among persons living in remote Alaskan communities. The Columbia River Intertribal Fish
Commission (1994) reported that during the two months of highest average fish consumption average
intake was 108 grams per day. The Tribes of Puget Sound reported (Toy et al., 1995) an average
intake of 73 grams per day with a 90th percentile intake of 156 grams/day. West et al. (1989) found a
mean intake of approximately 22 grams/day, but a reported maximum value over 200 grams/day.
Peterson et al. 91994) in a study of Chippewa tribes found that 2 percent of 323 respondents ate at
° least one fish meal each day. In these individual tribal and angler studies, data were generally not
separately reported for women of child-bearing age.
Estimated Exposure to Methvlmercurv from Fish
The extent to which elevated fish consumption by adult women (e.g., highest 1 to 5% of
female consumers with sustained consumption of fish resulting hi intakes of 100 grams per day or
greater) is associated with increased exposure to methylmercury depends on the concentration of
mercury in the type of fish as well as the total quantity consumed. Detailed analysis of the NPD, Inc.
1973/74 data were not conducted because of changes in the quantity and types of fish consumed
between 1973/74 and the present; and because these data were not designed to be representative of the
United States population. Using the cross-sectional data from CSFII 89/91 the fish consumed were
analyzed to assess the top 25 species of fish consumed. These fish species and the average mercury
concentration used to calculate mercury exposures are shown in Table 4-18. These species represent
approximately 95% of the fish reported to be consumed by people surveyed in CSFII 89/91. When
weighted by the frequency of consumption, the mean mercury concentration was 0.134 micrograms per
gram wet weight of fish. Consumption of a 100 gram portion of fish with the mixture of fish species
shown hi Table 4-19 at mean mercury concentrations is estimated at 13.4 micrograms of mercury.
The CSFn 89/91 data were analyzed to determine broad categories of fish consumed for three
groups: all persons in the survey who reported consuming fish or shellfish; persons who reported
consuming fish or shellfish at 100 grams per day or greater; persons whose exposure to mercury was
estimated to be 1 microgram/kg body weight/day or greater. These data are shown hi Table 4-20 for
all consumers, Table 4-21 for women of child-bearing age, and 4-22 for children ages 14 and younger.
For adults, comparison of the types of fish consumed indicates that mercury exposures of 1 ug/kg
bw/day and higher are associated with much higher consumption of swordfish, shark and barracuda
than is seen for all adult consumers. For children, the basis for mercury exposures of 1 ug/kg bw/day
or higher is not intake of swordfish, shark, or barracuda. Based on the analysis of the CSFII 89/91 the
basis for mercury exposures of 1 ug/kg bw/day or higher is generally higher intake of fish relative to
body size.
Among anglers and members of some Native American Tribes intakes of fish higher than
those of the general population have been reported. Detailed descriptions of these studies have been
presented in Volume III of this Mercury Study Report to Congress. Figure 4-5 summaries the higher
end fish/shellfish consumers. The level of mercury exposure for these groups depends on the mercury
June 1996 4-41 SAB REVIEW DRAFT
-------
1
o
Q.
(0
O
O
CO
U.
2
452,
350-
250- -
Figure 4-5
Distribution of fish Consumption Rates
of Various Populations
Wolfe & Walker '87 Highest Response Group Mean in AK
CRITFC '94 99th %ile Adult
LEGEND - POPULATIONS
GENERAL U.S. POPULATION
NPD 73/74 •
CSFII •
RECREATIONAL ANGLERS
PUFFER
FIORE
CONNELY
SUBSISTENCE FISHERS
WOLFE & WALKER A
NATIVE AMERICANS
CRITFC A
TOY, TULALIP 9
NOBMAN
EPA '92 Wl TRIBES
| Toy '95 Tulalip Tribe 90th %ile
Fiore '89 95th %ile
Wl Anglers
NPD 73/74 Adult 99th %ile
| Nobmann '92 AK Tribes Mean
CRITFC '94 Adult Mean |
Toy '95 Tulalip Tribe Median
Puffer '81 Median
Fiore '89 75th %ile Wl Anglers
j NPD 73/74 Adult 90th %ile
Connoly '90 NY Anglers Mean|+
Fiore '89 Wl Anglers |
NPD 73/74 Adult 50th %ile
^ CSFII Age 15-44 Mean ? anddjl
o—
June 1996
4-42
SAB REVIEW DRAFT
-------
Table 4-19
Fish Species by Percent of Consumption in CSFII 89/91,
Mean Mercury Concentration and Mercury Levels in 100 gram Servings of Fish
Fish/Shellfish Species
Tuna
Shrimp
Flatfish
Cod
Catfish
Salmon
Pollock
Perch
Clams
Crab
Haddock
Trout
Ocean Perch
Scallops
Oysters
Sardines
Whiting
Mackerel
Pompano
Mullet
Herring
Swordfish
Squid
Croaker
Anchovies
Percent of Overall
Fish or Shellfish
Reported as Consumed
by Respondents in
CSFII 89/91
31.88
10.71
10.55
6.51
4.23
4.21
4.01
3.58
3.42
3.33'
2.56
2.04
1.58
1.30
1.06
0.90
0.780
0.5558
0.52
0.38
0.34
0.28
0.28
0.24
0.24
jig Hg/Gram of Fish or
Shellfish on a Wet
Weight Basis
0.206
0.047
0.144
0.121
0.035
0.035
0.150
0.110
0.121
0.117
0.089
0.149
0.116
0.042
0.023
0.050
0.050
0.081
0.104
0.009
0.013
0.95
0.026
0.125
0.047
Hg Hg/100 Gram
Serving of Fish or
Shellfish
6.567
0.503
1.519
0.788
0.148
0.147
0.602
0.394
0.414
0.390
0.228
0.304
0.183
0.055
0.024
0.045
0.039
0.047
0.054
0.004
0.004
0.27
0.01
0.03
0.01
June 1996
4-43
SAB REVIEW DRAFT
-------
Table 4-20
Percent of Dietary Intake by Species-Category of Fish/Shellfish
for Persons Consuming Fish or Shellfish (from CSFII 89/91)
Type of Fish
Finfish (Except as listed
below)
Tuna
Shellfish
Swordfish, Shark and
Barracuda
Freshwater Fish
Total Fish
All Persons Consuming
Fish or Shellfish
(%)
40.6
25.8
16.9
0.7
16.0
to 100.0
Persons Consuming
Fish or Shellfish at 100
grams/day or more
(%)
34.7
12.3
27.6
1.2
23.8
to 100.0
Persons Consuming
Fish or Shellfish at 1
fig Hg/kg body weight
or more
(%)
23.0
24.0
19.0
27.6
6.4
to 100.0
Table 4-21
Percent of Dietary Intake by Species-Category of Fish/Shellfish by Women of
Child-Bearing Age (from CSFII 89/91)
Type of Fish/Shellfish
Consumed
Finfish (Except as listed
below)
Tuna
Shellfish
Swordfish, Shark and
Barracuda
Freshwater Fish
Total Fish and Shellfish
All Women of Child-
Bearing Age Who
Reported Consuming
Fish/Shellfish
(*)
40.3
26.1
16.7
0.8
15.6
to 100.0
Women Consuming 100
grams or more
Fish/Shellfish per Day
(%)
37.3
10.5
27.9
1.9
22.5
to 100.0
Women with Mercury
Exposure of 1 ug/kg
body weight/day or
higher
(*)
19.7
9.1
14.6
45.7
10.8
to 100.0
June 1996
4-44
SAB REVIEW DRAFT
-------
Table 4-22
Percent of Dietary Intake by Species-Category of Fish/Shellfish by
Children Ages 14 Years and Younger (from CSFII 89/91)
Type of Fish/Shellfish
Consumed
•
Finfish *
Tuna
Shellfish
Swordfish, Shark or
Barracuda
Freshwater Fish
Total Fish and Shellfish
All Children Reporting
Fish Consumption
43.8
32.3
11.0
0
12.9
to 100.0
Children Consuming >
100 grams/day
35.0
23.2
37.2
0
4.7
to 100.0
Children with
Mercury Exposures >
1 ug/kgbody
weight/day or more
22.4
42.5
28.6
0
6.5
to 100.0
mercury/gram fresh weight identified from the mix of fish reported in CSFII 89/91 ) with individual
fish species provides an estimate for specific groups. If fish such as salmon with a mercury
concentration of 0.035 ug mercury/gram fresh weight or shellfish (scallops and oysters averaging
0.055 and 0.024 \ig mercury/gram fresh weight, respectively) are chosen, mercury exposure would be
lower than that predicted for the general diet. However, if fish species with substantially higher
mercury concentrations are chosen mercury intake will be increased above that predicted for the
general diet.
Recreational anglers and members of Native American Tribes consume fish from the general
food supply as well as self-caught fish. For example, Fiore et al. (1989) reported mean, 75th and 95th
percentile consumption of fish among licensed anglers in Wisconsin. A comparison of total fish intake
and daily intake of sports-fish is shown in Table 4-23. Sports-fish comprised less than 50% of the
intake of fish among this group of anglers. Toy et el. (1995) reported on fish and shellfish
consumption among the Tulalip and Squaxin Island Tribes living near Puget Sound. Both tribes rely
on commercial fishing as an important part of tribal income. Subsistence fishing and shell-fishing are
significant parts of tribal members economies and diets. Among consumers of anadromous fish, local
waters (i.e., Puget Sound) supplied a mean of 80% of the fish consumed. Survey respondents from
the Tulalip Tribes purchased approximately two-thirds of fish from grocery stores or restaurants, while
among the Squaxin Island Tribe, the source of fish was about 50% self-caught and 50% purchased
from grocery stores or restaurants.
June 1996
4-45
SAB REVIEW DRAFT
-------
Table 4-23
Total Fish Intake and Intake of Sports-fish by
Licensed Wisconsin Anglers as Reported by Fiore et al., 1989
Level of Fish Consumption
Mean
75th Percentile
95th Percentile
Intake of Sports-fish
grams/day
12.3
15.5
37.7
Total Fish Intake
gins/day
26.1
34.2
63.4
Data on fish purchases by members of Native American Tribes and by anglers indicate that
persons who "subsist" on self caught fish also purchase fish commercially. The National Academy of
Sciences-National Research Council report Seafood Safety indicated that approximately 20% of fish
and shellfish are obtained outside of commercial sources (NAS/NRC, 1990).
Using age-specific data on the number of women predicted to be pregnant in a given year, the
number of pregnant women consuming fish at various levels is shown in Table 4-24.
Table 4-24
EstimatedNumber of Pregnant Women Consuming Fish at Various Intake Levels
Based on Data from NPD,Inc. 1973/74 and 1990 U.S. Census Data
Estimated Quantity of Fish Consumed
> 60 grams/day
At
the 99th Percentile or 1 12 grams of fish and shellfish per day
> 120 grams/day
Estimated Number of Pregnant Women
55,785
51,867
1,484
concentration of the fish consumed. Comparison of the estimated average mercury concentration for a
broad mixture of fish species (e.g., 0.13 to 0.14 ug).
4.3.2 Estimates for the Size of the Piscivorous Wildlife Population
Five wildlife species were considered in the exposure and ecological risk Volumes of this
assessment. The five species were selected because they consumed fish. The selected species
consisted of three avian species (the bald eagle, osprey and belted kingfisher) and two mammalian
species (the river otter and mink). Estimates of the sizes of these populations in the U.S. are presented
as part of the risk characterization. These population size estimates are uncertain; generally a range or
an imprecise estimate is presented. For most of these population estimates, there is no good method
for corroboration. It should also be noted that these piscivorous wildlife populations are not the only
species potentially exposed through the fish consumption route.
June 1996
4-46
SAB REVIEW DRAFT
-------
4.3.2.1 Bald Eagle
An estimated 10,000 to 12,000 bald eagles inhabit the lower 48 United States. This total
represents combined estimates of the total number of breeding pairs and immature eagles. U.S. Fish
and Wildlife Service (1994) estimated that there are 4,016 breeding pairs in the lower 48 states. The
Peregrine Fund, Inc. estimates that there are several thousand sexually immature eagles dwelling in the
same geographic area (Petit, 1995).
4.3.2.2 Osprey
The size of the U.S. osprey population is estimated to be between 10,000 and 20,000
individuals. This estimate is based on a compilation of individual state population size estimates
reported in the literature (Petit, 1995).
4.3.2.3 Belted Kingfisher
Population estimates for small birds such as the belted kingfisher have ajarger degree of
uncertainty because they are based on species density estimates and it is not possible to assess the
accuracy of such predictions. Petit (1995) presents a rough estimate of approximately 170,000 belted
kingfishers in the lower 48 states. This estimate is the product of estimated kingfisher densities from
the breeding bird survey and total land area of the lower 48 United States.
4.3.2.4 Mink
The National Geographic Society (1960) estimated that approximately 1,000,000 mink are
trapped each year on the North American continent The source of this information is clearly dated.
If one assumes that 10 percent of the population is snared each year, then, roughly 10,000,000 mink
live on the North American Continent (Petit, 1995). There is a great deal of uncertainty in this
estimate.
4.3.2.5 River Otter
•>
Although the original otter range encompassed all the U.S. states on the North American
continent, the species range is presently more limited. Otter populations are considered stable across
the United States (Jenkins, 1983), although they are listed as endangered species in several states.
The book Wild Mammals of North America Biology, Management, and Economics edited by
Chapman and Feldhamer (1982) reports that otters are extremely difficult to count noting the
questionable accuracy of most index techniques. The book notes that most states base otter population
estimates on the reports of trapper and furbuyers. Jenkins (1983) estimated that, in a one-year period
over 1978 and 1979, 29,000 otters were harvested in the United States. Using the crude estimation
that 10 percent of the total population is eliminated by trapping in a given year, there are roughly
300,000 otters inhabiting the United States.
June 1996 4-47 SAB REVIEW DRAFT
-------
5. INTEGRATIVE ANALYSIS FOR METHYLMERCURY
5.1 Characterization of Risk: Quantitative Integration of Human and Wildlife Exposure and
Dose-Response
5.1.1 Introduction
In this chapter findings from the exposure analyses are integrated with those from the dose-
response assessments for both humans and wildlife. This integration is done only for methylmercury,
as the exposure assessment indicates this is the form to which the greatest exposure is likely. The
quantitative dose-response measures used for methylmercury are these: the human RfD of 1x10"4
mg/kg-day and the benchmark dose from which it was derived; the individual wildlife criteria and the
wildlife RfDs, LOAELs and NOAELs on which they were based.
%
a
The purpose of Section 5.2 was to determine which of the species considered to consume fish
from the hypothetical water body (developed in Volume III) is expected to be adversely affected by
the lowest methylmercury concentrations in fish (that is, individuals of which species are expected to
be the most at risk from methylmercury concentrations in fish). Comparisons of the fish consumption
rate assumptions for humans and the five wildlife species considered (presented in Volumes III and V)
and the health endpoint data (developed in Volumes IV and V) for the species considered are
presented. Assumptions employed to estimate the transport of mercury through the aquatic food chain
model (developed in Volumes HI and V) are described to illustrate the impact of selected uncertainties
underlying the assumptions. The fish consumption rate assumptions and the health endpoint data were
then integrated.to assess which species is most at risk from methylmercury in fish.
The aim of Section 5.3 was to compare quantitative dose-response estimates or
recommendations with measured mercury levels in fish and to determine the numbers of individuals
estimated to consume those mercury levels. This comparison gives an indication of the size of the
population that is not likely to be impacted by mercury. Comparisons with the total population
numbers gives an indication of the size of the "at risk" population.
5.1.2 Description of Subsistence Fishers
The term "subsistence fishers" has been used to describe various persons who rely on fish as a
major source of protein. "Subsistence fishers" are not defined by whether the fish/shellfish are self-
caught or obtained for money. Groups with high fish intake are typically determined by social,
economic, ethnic, and geographic characteristics. An additional group of people consume high levels
of fish in response to numerous health-based messages that have promoted the consumption of fish to
reduce the likelihood of disease, particularly of the cardio-vascular system. Further, there are large
numbers of people who simply prefer fish and shellfish as a source of protein. Consequently in the
following analyses, "high-end fish consumers" include these groups: anglers; members of some Native
American Tribes; members of ethnic groups who consume higher than typical intakes of fish; persons
who preferentially select fish for health-promotion purposes; individuals who relish the taste of fish;
and persons who rely on self-caught fish from local sources because of limited money to buy food.
Although humans have a degree of choice on their source of protein, the wildlife described
have much more restricted choices on protein sources because they are confined spatially or
territorially. Consequently all consumption by wildlife has been assumed to be locally caught,
although the highest predators in the aquatic food web cover wide territories.
June 1996 5-1 SAB REVIEW DRAFT
-------
5.2 Integration of Modeled Methylmercury Exposure Estimates for Humans and Wildlife
with the Dose-Response Assessments
5.2.1 Methylmercury Intake by Humans and Wildlife Based on Modeling of Fate and Transport of
Mercury and Patterns of Fish Intake
A comparison of pollutant exposure levels across the species in an ecosystem requires, among
other things, a knowledge of the environmental fate and transport of the pollutant (including chemical
transformation of the pollutant in the environment), the contact medium of interest, as well as the
contact rates and body weights of the wildlife species and the human subpopulations of interest in the
ecosystem.
»
Although methylmercury is found in other media and biota, it accumulates to the highest
concentrations in the muscle tissues of fish, particularly piscivorous fish. This conclusion is based
upon both the measurement data-and the results of the modeling presented in the Volume III of this
Report. Methylmercury remains essentially unchanged in fish tissue, when subjected to human
preparation methods (i.e., cooking). Although methylmercury exposure may occur through other
routes (for example, the ingestion of mernylmercury-contaminated drinking water and the consumption
of food sources other than fish such as amphibians or reptiles, the inhalation of atmospheric
methylmercury, and dermal uptake through contact with soil and water), the fish consumption pathway
dominates these other methylmercury exposure pathways in piscivores. This is clearly the result of the
bioaccumulation of methylmercury in their food source, fish, and because this compound is highly
bioavailable from fish. Other forms of mercury are also toxic, but since they are not known to
accumulate in commonly eaten foods, and since they are not as bioavailable in most media, they are
not of as great a concern. Consequently, the following comparison of methylmercury contact rates is
based solely on the daily ingestion rate of fish and assumptions pertaining to the transport of
methylmercury through the aquatic food chain.
The piscivores selected for analysis were these: human recreational angler, human high-end
local fish consumer (or subsistence fisher), child of a high-end local fish consumer, bald eagle, osprey,
kingfisher, mink and otter. All species were assumed to consume fish from the same lake and the
same concentrations of methylmercury were assumed to exist in the fish of the same trophic level.
The piscivore's estimated methylmercury contact rate from fish consumption was based on two
important factors: the methylmercury concentration in the contaminated fish and the daily amount of
fish eaten.
The methylmercury concentration in fish was estimated by multiplying the total dissolved
mercury concentration in water by a bioaccumulation factor (BAF). The predicted ranges and patterns
of mercury concentrations in surface waters at hypothetical locations near the hypothetical mercury
sources (the model plants) and under other deposition assumptions have been described in Volume III.
The type of model plant is predicted to be the most critical factor affecting mercury concentrations in
surface waters. Local water bodies in proximity to anthropogenic sources emitting substantial levels of
divalent mercury or close to sources with low stack heights or slow stack exit gas velocities were
predicted to be more highly impacted by stack mercury releases than those water bodies near sources
emitting lower levels of mercury or having higher effective stack heights. Based on the modeling
presented in Chapter 6 of Volume III, the closer a water body is to a mercury emissions source, the
higher the resulting mercury concentrations in the water body. (See the related discussion in Chapter
6 of Volume HI.) This also assumes that the size of the watershed and transport of mercury in the
June 1996 5-2 SAB REVIEW DRAFT
-------
watershed of each lake is the same. It is also important to remember that this modeling was
conducted assuming flat terrain surrounding the model plants.
For locations that are assumed to be influenced only by long-range atmospheric mercury
transport, the concentrations of mercury in surface waters were a function of overall proximity to
anthropogenic sources, increased soot and ozone levels in the atmosphere and elevated rainfall. (See
the related discussion in Chapter 5 of Volume HI.)
The concentrations of methylmercury in fish are also influenced by the fishes' diets. (The
bioaccumulation model is described in Volume V). Briefly, in the four-tier trophic food chain model
used in this effort, fish are assumed to feed at two levels: trophic level 3 fish were assumed to feed
on plankton which have lower levels of methylmercury, and trophic level 4 fish were assumed to feed
on trophic level 3 fish, which, due to bioaccumulation, had higher methylmercury concentrations than
the plankton upon which they feed. The bioaccumulation factor (BAF) (66,200) for trophic level 3
fish was estimated based on several sets of collected data. The BAF for trophic level 4 (335,000) was
estimated by applying a predator-prey factor (of approximately 5) to the bioaccumulation factor
estimated for trophic level 3 fish.
The BAF model assumed that the higher the dissolved total mercury concentrations in the local
waters, the proportionally greater the methylmercury concentration in the fish; and, as a consequence,
the higher the methylmercury exposure of the piscivores. The biomagnification of methylmercury as
modeled through the aquatic food web significantly impacts the exposure of piscivores. Those
piscivores consuming a diet primarily consisting of trophic level 3 fish would be predicted to receive
approximately 5 times less (20 percent) methylmercury per gram of fish eaten than those eating trophic
level 4 fish from the same site. Humans, which are assumed to eat only trophic level 4 fish, will have
a greater methylmercury exposure per gram of fish consumed than ospreys and kingfishers, which are
assumed to consume only trophic level 3 fish from the same water bodies. Similarly, otters, which are
assumed to consume an 80/20 mix of trophic level 3 and 4 fish will have a greater methylmercury
exposure per gram of fish consumed than minks, which are assumed to eat only trophic level 3 fish.
The ratio of grams fish consumed per day to piscivore body weight is also important in
estimating methylmercury exposure on a g/kg bw/day basis. The greater this ratio the higher the
resulting methylmercury exposure assuming methylmercury concentrations in consumed fish are
constant. For example, osprey and kingfishers each consume trophic level 3 fish only. Since
kingfishers daily consume 50 percent of their body weights hi fish and osprey roughly 20 percent of
their body weights hi fish of the same trophic level, the resulting average daily methylmercury intake
hi g/kg body weight will be higher among the kingfisher population.
Assuming that these piscivorous birds and mammals and the human fish-eating subpopulations
consume fish from the same lake, the estimates of daily consumption rates, the trophic level of the fish
consumed and the body weight of the animal all contribute significantly to methylmercury exposure
when expressed on a per kg of body weight basis. For example, the daily fish consumption of the
otter is approximately 16 percent of body weight and that of mink is 20 percent. Trophic level 4 fish
are assumed to make-up roughly 20 percent of the otter's total fish consumption with the other 80
percent consisting of trophic level 3 fish; on the other hand, minks are assumed to eat exclusively
trophic level 3 fish. As a result of percent of daily body weight consumed as fish and the trophic
level of fish consumed, otters will have a higher methylmercury contact rate than mink.
June 1996 5-3 SAB REVIEW DRAFT
-------
By using the relationship for methylmercury described by the four-tier trophic food chain
model (i.e., the different bioaccumulation factors for fish in trophic levels 3 and 4), the estimates of
the daily fish consumption rates from each trophic level and the body weight of the animal, the rates
of methylmercury exposure (m.mg/kg bw/day) for the animals in this hypothetical environment can be
ranked. To illustrate this, assume that for a lake at a given location all trophic level 3 fish are
contaminated with 0.1 ug methylmercury/g fish tissue; the trophic level 4 fish would be predicted to
have methylmercury concentrations of 0.5 ug/g. Eagles at this lake daily consume (370 g/day x 0.1 ug
methylmercury/g fish tissue) + (90 g/day x 0.5 ug methylmercury/g fish tissue)= 82 ug
methylmercury/day; given the body weight estimate 4.5 kg, the rate of exposure is estimated as 18
ug/kg bw/day.
Continuing the example exposure estimates for the other species:, ospreys at this lake: 0.1
ug/g x 300g/day/1.5 kg bw = 20 ug/kg bw/day; kingfishers at this lake: 0.1 ug/g x 75/0.15 = 50 ug/kg
bw/day; otters at this lake consume both trophic level 3 and 4 fish: (0.1 ug/g x 976 g/day + 0.5 ug/g
x 244 g/day)/7.4 =30 ug/kg bw/day; mink at this lake: 0.1 ug/g x 160.2 g/day/0.8 = 20 ug/kg bw/day;
and high-end fish-eating humans at this lake: 0.5 x 60 g/day/70 = 0.4 ug/kg bw/day.
From the modeling standpoint the methylmercury levels in the trophic level 3 fish or the total
dissolved mercury concentration in the water is irrelevant to the rank; only the relationship between
the aquatic trophic levels is critical. Using this model and the assumptions in Tables 5-1, 5-2 and 5-3,
the predicted piscivore exposure ranking from highest to lowest is: kingfisher > otter > osprey, mink
> bald eagle > human.
Table 5-1
Assumed Human Fish Consumption Rates and Body Weights Used in the Exposure Modeling
Subpopulation
Adult High-End Fish Consumer
Child High-End Fish Consumer
Adult Recreational Angler
Assumed
Body Weight
(kg)
70
17
70
Assumed Local Fish Consumption Rate
(Swet weight/^)
60
20
30
June 1996
5-4
SAB REVIEW DRAFT
-------
Table 5-2
Assumed Fish Consumption Rates and Body Weights of Piscivorous Birds and Mammals
Used in the Exposure Modeling
Animal
Bald Eagle
Osprey
Kingfisher
River
Otter
Mink
Body
Weight
4.6
1.5
0.15
7.4
0.8
Total Ingestion
Rate
(gwet weight^)
500
300
75
1220
178
Percent of Diet
Consisting of
Trophic Level 3
Fish
74
100
100
80
90
Percent of Diet
Consisting of
Trophic Level 4
Fish
18
0
vo
20
0
Percent of
Non-Aquatic
Foods in Diet
8
0
0
0
10
Table 5-3
Assumed Fish Consumption Rates of Piscivorous Birds and Mammals
Used in the Exposure Modeling
Annual
Bald Eagle
Osprey
Kingfisher
River Otter
Mink
Trophic Level 3 Fish Ingestion Rate
(g/day)
370
300
75
976
160.2
Trophic Level 4 Fish Ingestion Rate
(g/day)
90
0
0
244
0
The ranking demonstrates the importance of the trophic level of the fish which the piscivore
consumes, the daily consumption rate, and the ratio of daily fish consumption rate to body weight.
Despite consuming a comparatively small amount of trophic level 3 fish, the kingfisher ranked first in
this exposure ranking scheme; these birds consume large amounts of fish on a daily basis by
comparison to then- body weights. This use of this method also illustrates that within this hypothetical
ecosystem the human methylmercury contact rate based on fish consumption is much lower than that
of these piscivorous wildlife.
June 1996
5-5
SAB REVIEW DRAFT
-------
5.2.2 Comparison of Dose-Response Estimates Across Species
The second step for ranking species at risk from fish-related methylmercury exposure entails a
comparison of the health endpoints and the associated dose-response across species. The chemical
species of mercury (i.e., methylmercury) and the route of exposure (i.e., fish consumption) are the
same for all wildlife and human species. For the comparisons across health endpoints to be valid, the
health effects must be judged to be of similar concern for the species considered.
Methylmercury (as described in Volumes IV and V of this Report) has deleterious effects on
the chordate nervous system. Methylmercury also efficiently passes through the intestinal walls of
chordates and into the blood. Once in the blood, methylmercury may cross the blood brain and
placenta! barriers and impact the potentially affected neuronal tissues.
*The human health endpoint of concern is developmental neurotoxicity. The health endpoints
of concern for the avian wildlife species are reproductive and behavioral -deficits and for the
mammalian quadrupeds are neurological effects. For more details see Volumes IV and V.
Assuming that the effects are of similar concern for the well-being of individuals within a
species, the NOAELs, LOAELs and the human and wildlife WCs for these health endpoints can then
be compared across species.
U.S. EPA has on two occasions published RfDs for methylmercury which have represented the
Agency consensus for that time. These are described in the sections below. At the time of the
generation of the Mercury Study Report to Congress, it became apparent that considerable new data on
the health effects of methylmercury in humans were emerging. Among these are large studies of fish
or fish and marine mammal consuming populations in the Seychelles and Faroes Islands. Smaller
scale studies are in progress which describe effects in population s around the U.S. Great Lakes. In
addition, there are new evaluations of published work described in section 3.3.1.1 of Volume IV,
including novel statistical approaches and application of physiologically based pharmacokinetic
models.
As the majority of these new data are either not yet published or have not yet been subject to
rigorous review, it was decided that it was premature for U.S. EPA to make a change in the
methylmercury RfD at this time. An interagency process, with external involvement, will be
undertaken for the purpose of review of these new data, evaluations of these data and evaluations of
existing data. An outcome of this process will be assessment by U.S.EPA of its RfD for
methylmercury to determine if change is warranted.
The neurotoxicity of methylmercury hi children exposed in utero has been determined to be
the critical effect for the human RfD. The RfD was based on a statistical analysis of data from human
subjects exposed to methylmercury through the ingestion route in Iraq (Marsh et al., 1987). (See
Volume IV and Chapter 2 of this volume.) The RfD for humans was estimated to be IxlO"4 mg/kg-
day or 0.1 ug/kg bw/day. To compare methylmercury dose-response in the observed response range,
human NOAELs and LOAELs were estimated from the Marsh et al. (1987) data by using the hair-
mercury concentration groupings given hi the Seafood Safety report from NAS/NRC (NAS, 1991; see
Table 5-4). In this report each of the maternal-child pairs were assigned to one of five hair-mercury
concentration groups. The geometric means of each of the hair-mercury concentration groups were
1.4, 10.0, 52.5, 163.4 and 436.5 ppm. The incidence of combined developmental effects (late walking,
late talking, mental symptoms, seizures or neurological score greater than 3) in each of the groups was
June 1996 5-6 SAB REVIEW DRAFT
-------
18.5 percent, 21.4 percent, 46.2 percent, 66.7 percent and 93.3 percent for the 1.4, 10.0, 52.5, 163.4
and 436.5 ppm groups, respectively. The combined developmental effects incidence was determined
from Marsh et al. (1987) by scoring an individual as a responder if one or more of the developmental
effects was observed, summing the responders across each group and dividing by the number of
individuals in each group. These concentration groupings and incidence of combined developmental
effects were used in the calculation of the benchmark dose for the derivation of the methylmercury
RfD. The benchmark dose of 11 ppm mercury in hair was operationally equivalent to a NOAEL in
the derivation of the methylmercury RfD. A LOAEL of 52.5 ppm mercury in hair was estimated for
this risk characterization from inspection of data in Table 5-4. The NOAEL of 11 ppm mercury in
hair and the LOAEL of 52.5 ppm mercury in hair correspond to ingestion levels of 1 ^ig/kg-day and
5.3 ug/kg-day, respectively; these dose conversions were made by applying the methods for converting
hair mercury concentrations to ingestion levels used in the derivation of the RfD in Volume IV of this
Report.
Table 5-4
Incidence of Effects in Iraqi Children by Exposure Group3
Effect
Late walking
Late talking
Mental symptoms
Seizures
Neurological scores
>3
Neurological scores
>4
All endpoints
N (sample size)
Dose (ppm) Mercury in Hair
1.37
0
2
1
0
3
0
4
27
10
2
I
0
0
1
1
3
14
52.53
2
3
1
1
4
2
6
13
163.38'
3
4
3
2
3
2
8
12
436.60
12
11
4
4
9
•
6
14
15
* From Table 6-11 of Seafood Safety; dose is geometric mean
The RfDs for avian and mammalian wildlife are derived in Volume V of this Report. The
avian RfD was based on the data from a series of studies by Heinz and collaborators (Heinz, 1974,
1975, 1976a,b, 1979). Heinz and collaborators fed mercury contaminated grain to mallard ducks. A
NOAEL could not be identified. The estimated LOAEL, based on reproductive and behavioral effects,
was 64 ug/kg bw/day. The avian RfD was estimated by dividing the LOAEL by the uncertainty
factors.
The estimation of the RfD for the avian species utilized the following formula: RfD = TD x
[1/(UFA x UFS x UFL)], where
June 1996
5-7
SAB REVIEW DRAFT
-------
RfD = 64 |jg/kg bw/day x [l/(3 x 1 x 3)]
RfD = 7.1 ug/kg bw/day
where: TD - tested dose; here equal to the LOAEL of 64 ug/kg bw/day.
UFA - an uncertainty factor to indicate the uncertainty in applying a dose-response derived
for one species to another. A factor of 3 was applied.
UFS - an uncertainty factor which accounted for extrapolation from a subchronic dose-
response study to a chronic exposure. As the duration of the'Heinz studies was for the
animals' lifetime, a factor of 1 was applied.
UFL - an uncertainty factor employed to indicate uncertainty around the toxic threshold. A
factor of 3 was applied.
\
Note that in Volume V for the calculation of the Wildlife Criteria, the composite uncertainty factor
(UFL x UFA) was rounded to 10.
The mammalian RfD was based on the data from a series of studies by Wobeser and
collaborators (Wobeser, 1973; Wobesser et al., 1976a,b). Wobeser and collaborators fed
methylmercury to ranch mink. A NOAEL of 55 ug/kg bw/day was estimated from these studies. The
estimated LOAEL, based on damage to the nervous system and liver, was 180 ug/kg bw/day. The
mammalian RfD was estimated by dividing the NOAEL by the uncertainty factors. The uncertainty
factors utilized included: UFL-an uncertainty factor employed to indicate uncertainty around the toxic
threshold, UFA-an uncertainty factor to indicate the uncertainty in applying a dose-response derived for
one species to another, and UFS- an uncertainty factor which accounted for extrapolation from a
subchronic dose-response study to a chronic exposure.
The estimation of the RfD for the mammalian species utilized the following formula:
RfD = TD x [1/(UFA x UFS x UF^]
RfD = 55 ug/kg bw/day x [1/(1 x 10 x 1)]
RfD = 5.5 ug/kg bw/day
where: TD - tested dose; here equal to the LOAEL of 55 ug/kg bw/day.
UFA - an uncertainty factor to indicate the uncertainty in applying a dose-response derived
for one species to another. A factor of 1 was applied. Mink and otter are considered
to be similar.
UFS - an uncertainty factor which accounted for extrapolation from a subchronic dose-
response study to a chronic exposure. The Wobeser studies were judged to be
subchronic, and factor of 10 was applied.
UFL - an uncertainty factor employed to indicate uncertainty around the toxic threshold.
Since a NOAEL was estimated a factor of 1 was applied.
Based on the data developed for the health assessment, the human LOAEL and RfD are orders
of magnitude lower than the corresponding LOAELs and RfD of the other animals (Table 5-5). There
June 1996 5-8 SAB REVIEW DRAFT
-------
is a great deal of uncertainty in this comparison. It must be noted that the effects in humans are based
on the RfD definition of a critical effect; that is the most sensitive reported adverse effect or indicator
of adverse effect. The effects reported for mammals (i.e., neurologic damage in the mink) and birds
(i.e., reproductive effects in mallards) would be considered frank effects in the human RfD
methodology. The observations in laboratory animals indicate that it would be reasonable to expect
more subtle and less damaging effects of methylmercury to occur at lower doses than the wildlife
LOAEL AND NOAEL.
Table 5-5
Animal and Human Health Endpoints for Methylmercury in Mg/kg bw/day
Animal
Human
Mammalian
Quadrupeds
Avian
NOAEL
1.0
55
-
LOAEL
5.3
180
64
Health Effect
Related to LOAEL
Neuro-developmental
effects in children
Neurological damage
Behavioral and
reproductive effects
RfD or
Wildlife
we
0.1
5.5
7.1
Health Effect Related
to RfD oY Wildlife WC
Neuro-developmental effects
in children
Neurological damage
Behavioral effects
5.2.3 Integration of Modeled Methvlmercurv Intake Through Consumption of Fish for Hypothetical
Humans and Wildlife with Dose-Response Data
In this section the dose-response and exposure estimates are integrated to predict
concentrations of methylmercury in fish tissue which correspond to various health endpoints when
consumed by the piscivore. The methylmercury body burdens in fish which correspond to piscivore
health endpoints are estimated by dividing the product of the piscivore body weight (kg) and the
human or wildlife WC (jag/kg/day), or LOAEL (jig/kg/day) by the daily rate of fish consumption
(g/day). The units that result are expressed on the basis of fish muscle concentration (ng
methylmercury/g fish muscle tissue). The corresponding fish muscle concentrations could also account
for the differences in bioaccumulation between trophic level 3 and 4 fish. This was accomplished by
converting the concentrations calculated for consumers of trophic level 4 fish to the values expected in
trophic level 3 fish in the same lake and vice versa. The difference hi BAF 3 and BAF 4 is
approximately a factor of 5 and is based on the predator-prey factor. (See Volume V for more
details).
Table 5-6 shows that the RfDs for humans and wildlife are estimated to result from fish
concentration at least nine times lower than those for the estimated LOAELs. The RfD is, by
definition, a protective endpoint. There is predicted to be no risk at this level of exposure. The risks
posed by exceedance of this value cannot be estimated from the available data. The LOAEL, on the
other hand, is the lowest observed effect level. Exposure at the LOAEL is predicted to cause adverse
effects in some members of the populations. Since an avian NOAEL could not be estimated from the
data, an additional comparison of fish methylmercury concentrations not expected to adversely affect
June 1996
5-9
SAB REVIEW DRAFT
-------
the health of the piscivores was not done. The use of the human and wildlife WC was the only basis
for estimating the fish methylmercury concentrations not expected to have adverse effects on the health
of the piscivores.
Table 5-6
The Concentrations of Methylmercury in Trophic Level 3
and Trophic Level 4 Fish Which, If Consumed at the
Assumed Rates on a Daily Basis, Result in Exposure at the RfD or the LOAEL
Population
Eagle
Osprey
Kingfisher
Mink
Otter
Adult Human
Recreational Angler
Adult Human High-
End Fish Consumer
Child of High-End
Fish Consumer
Methylmercury
Concentration in
Trophic Level 3
Fish at RfD
(Mg/g).
0.04
0.036
0.014
0.027
0.019
0.05
0.02
0.02
Methylmercury
Concentration in
Trophic Level 4
Fish at RfD
(ng/g)
0.20
0.18
0.07
0.135
0.093
0.23
0.12
0.08
Methylmercury
Concentration in
Trophic Level 3
Fish at LOAEL
(ng/g)
0.36
0.32
0.13
0.90
0.61
2.47
1.24
0.90
Methylmercury
Concentration in
Trophic Level 4
Fish at LOAEL
(Mg/g)
1.8
1.6
0.64
4.49
3.04
12.4
6.18
4.51
Based on the results presented in Table 5-6, the most susceptible species can be determined.
The species with the lowest estimated fish methylmercury concentration which corresponds to a
LOAEL level is predicted to be the most susceptible to methylmercury concentrations in fish. The
kingfisher is the most susceptible species considered in this analysis based on either the LOAEL or the
Wildlife WC. This susceptibility is a function of the large amount of fish per body weight consumed
by this species.
Comparing both the human and wildlife WCs, the high-end fish-consuming humans and the
otter follow the kingfisher; mink, osprey, eagle, and human recreational angler follow on this basis. It
is interesting to note that fairly low concentrations of methylmercury in trophic level 3 fish are
predicted to result in wildlife and human exposure at the RfD or Wildlife WC. The predicted fish
methylmercury concentrations also exhibit a fairly narrow range; the range is from 0.01 to 0.05 ug
methylmercury/g fish tissue.
June 1996
5-10
SAB REVIEW DRAFT
-------
If the LOAEL is the basis for comparison, the kingfisher is followed by the osprey, eagle,
otter then mink and child and, finally, adult high-end human fish consumers. The piscivorous wildlife
are, on the basis of the LOAEL, generally more sensitive than the humans.
Note both the range between the RfD and the LOAEL for these organisms and the differences
in severity of the health end points. Changes in this range (e.g., a reduction of the mammalian
quadruped LOAEL) may change the predicted ranking. New avian and mammalian studies which
examine the effects of lower dietary methylmercury levels on neuronal tissues may provide a more
appropriate comparison for the human health data. *
Both modeled and measured methylmercury concentrations result in exceedence of the LOAEL
for all the wildlife species considered (See Volume HI Chapters 2 [measured concentrations in fish],
and 5 and 6 [predicted concentrations in fish]). Measured methylmercury concentrations in fish which
would result hi exceedence of the LOAEL for humans have been reported. These values are also
predicted to occur in a few combinations in the modeling exercise; for example, when the long range
transport and the predicted concentrations based on the local atmospheric modeling of the large
municipal waste combustor are combined for hypothetical locations in the Eastern United States.
Table 5-6 shows the fish methylmercury concentrations that are predicted to be protective of
selected wildlife species and the three hypothetical humans. Selection of the human RfD (based on an
estimate of 60 grams of fish consumption per day) as a protective basis for any risk management
action is expected to be protective of all wildlife considered except for the kingfisher.
There is a great deal of uncertainty in this comparisoa The uncertainty includes the
uncertainty and variability in the health endpoints and fish consumption rates. For example, there are
reports of fish consumption by persons hi the U.S. in exceedance of 60 grams/day; these are
documented hi Appendix H of Volume HI. There is also a great deal of variability pertaining to the
transport of mercury from the water body and sediments through the aquatic food chain. See related
discussions in Volume HI and Volume V.
9
For mercury the form or species of most concern for chordates in this assessment is
methylmercury (see Volumes IV and V of this Report). Although methylmercury is found in other
media and biota, it accumulates to the highest levels in the muscle tissues of fish, particularly
piscivorous fish. This is based upon both the measurement data and the modeling data also presented
in the Exposure Volume (Volume IE). Methylmercury remains essentially unchanged in fish tissue,
when subjected to human preparation methods (i.e., cooking). Methylmercury also efficiently passes
through intestinal walls into the blood. Other forms of mercury are also toxic, but since they are not
known to accumulate in commonly eaten foods, and since they are not as bioavailable hi most media,
they are not as great a concern.
S3 Potential Effects of Mercury Emission Sources on Local Fish Consumers
In the hypothetical sites constructed hi Volume III of this Report, some of the mercury emitted
from local sources is predicted to deposit on local watersheds and water bodies. The fate and transport
of the atmospheric mercury in the local area around the sources (within 50 Km) was modeled with a
modified version of the COMPDEP model. Since mercury emissions are also thought to be transported
across great distances, the RELMAP model was used to estimate the impacts of all mercury sources in
the continental U.S. The RELMAP model predicted a range of mercury air concentrations and mercury
deposition rates across the U.S. The U.S. was divided into Eastern and Western halves. The
June 1996 ' 5-11 SAB REVIEW DRAFT
-------
hypothetical local sites were placed at the 50th and 90th percentiles for the RELMAP-predicted
mercury air concentration and total deposition in both the Eastern and Western halves of the U-S.
Some of the mercury emitted from continental U.S. sources is also predicted to deposit on local
watersheds and water bodies.
The results of both the RELMAP and COMPDEP models are associated with some degree of
uncertainty. The uncertainties have been described in Volume III. It is important to note that the dry
deposition of divalent vapor-phase mercury, one component of the total mercury deposition, is highly
uncertain. The deposition velocity associated with this form was derived from that of a surrogate
compound with a similar Henry's Law Coefficient.
The movement of mercury through the environment is very complex. The deposited mercury is
predicted to be transported to the water bodyjand some of the total mercury that is dissolved in the
water column is predicted to be incorporated into the aquatic food chain. The amount of mercury that
is incorporated into the aquatic life at each trophic level is 'predicted through a bioaccumulation factor
(BAF). Uptake of dissolved mercury is highly variable, and there appears to be a great deal of
variability among water bodies. A number of factors, such as pH, appear to affect bioaccumulation.
The BAFs were derived from measured data; the uncertainty analysis presented in Volume V indicates
the range of values that can be associated with the BAF.
Piscivorous human or animal exposure to mercury results from the consumption of fish.
Essentially all the mercury in fish appears to be in the form of methylmercury. The uncertainties in the
modeling results were large and, as a result, only qualitative conclusions were derived. The
consequences of the predictions at these hypothetical sites can be examined.
5.3.1 Humans
Three types of human consumers were modeled: the adult subsistence fisher, the child of the
subsistence fisher and the adult recreational angler. The fish consumption rates associated with these
three hypothetical individuals was. 60, 20 and 30 grams/day, respectively.
The fish consumption rates for the subsistence anglers was that reported as the mean
consumption by Columbia River Intertribal Fish Commission (1995). Mean daily consumption for
females was 56 g and for males, 63 g. These values are corroborated by data from other studies. Fish
consumption studies such as those of Nobmann, 1992 (mean fish consumption of 3 Alaskan Tribes),
Fiore et al., 1989 (95th percentile of Wisconsin Anglers), Toy et al., 1996 (90th percentile fish
consumption rate for the Tulalip tribe) and Wolfe and Walker, 1987 (group mean of the highest
response group) report similar or higher fish consumption rates than those modeled. The CSFII 89/91
reports the 95th and 99th percentiles of fish consumption rates for estuarine and freshwater fish (the
types most likely consumed by subsistence anglers) as 47.3 and 113 g/day, respectively. Clearly, there
are individuals who consume high levels of fish, and they comprise a small percentage of the U.S.
population.
Given these fish consumption rates, the issue then becomes the concentrations of
methylmercury in the fish. The subsistence-level consumers modeled were assumed to derive all fish
consumed from a single or small number of geographically proximal water bodies. The water bodies
modeled were assumed to be a series of small lakes. Humans were assumed to consume trophic level 4
fish. Although the methylmercury concentrations of trophic level 4 fish are known to vary,
measurements near or exceding one ppm are not uncommon. As the predicted fish concentrations
June 1996 ' 5-12 SAB REVIEW DRAFT
-------
exceed 1 ppm in trophic level 4 fish and at daily fish consumption rates modeled, exposure to
methylmercury through fish consumption approaches or exceeds the product of 10 times RfD for the
high end consumer.
In the hypothetical Eastern site, when the 50th percentile RELMAP results and the results of
the local emission sources (i.e., model plants) are combined, two scenarios are predicted to result in
fish mercury concentrations greater than 1 ppm: the trophic level 4 fish in lakes at 2.5 Km from the
chlor-alkili plant (CAP) and the primary lead smelter (PLS) (See Table G-l from Volume III). At 2.3
and 1.9 ug mercury/g, fish a daily consumption rate of 60 g and a body weight of 70 Kg results hi
predictions of 2.0 and 1.6 ug methylmercury/Kg bw/day, respectively. In the hypothetical Eastern site,
when the 90th percentile RELMAP results and the results of the local emission sources (!.&, model
plants) are combined, five combinations are predicted to have fish mercury concentrations greater than
1 ppm: the trophic level 4 fish in lakes at 2.5 Km from the chlor-alkali plant (CAP), the primary lead
smelter (PLS), the large municipal waste combustor (large MWC) and the small municipal waste
combustor (Small MWC) and the trophic level 4 fish in lakes at 10 Km from the large MWC mercury
(See Table G-l from Volume III). See Table 5-7.
Table 5-7
Predicted Methylmercury Concentrations in Fish in the Eastern U.S.
combining 90th percentile RELMAP Estimates and Local Source Estimates
and the Resulting Human Exposure Estimates.
Combination
Large MWC 2.5 Km
Small MWC 2.5 Km
CAP 2.5 Km
PLS 2.5 Km
Large MWC 10 Km
Predicted Trophic Level 4 Fish
Concentration (ug mercury/g fish)
4.9
1.1
2.6
2.1
1.1
Estimated Human Intake
ug mercury/Kg bw/day
4.2
0.94
2.2
1.8
0.94
In the hypothetical Western site, when the 50th percentile RELMAP results and the results of
the local emission sources (i.e., model plants) are combined, two combinations are predicted to have
fish mercury concentrations greater than 1 ppm: the trophic level 4 fish in lakes at 2.5 Km from the
large MWC and the CAP (See Table G-2 from Volume III). At 1.9 and 2.1 ug mercury/g fish, a daily
consumption rate of 60 g and a body weight of 70 Kg results in predictions of 1.6 and 1.8 ug
methylmercury/Kg bw/day, respectively. In the hypothetical Western site, when the 90th percentile
RELMAP results and the results of the local emission sources (i.e., model plants) are combined, the
same two combinations are predicted to have fish concentrations greater than 1 ppm: the trophic level
4 fish in lakes at 2.5 Km from the large MWC and the CAP (See Table G-2 from Volume III). The
predicted exposure estimates 1.7 and 1.9 ug/ Kg bw/day, respectively.
The predicted methylmercury concentrations hi trophic level 4 fish occur generally hi the high
end combinations (e.g., 90th percentile RELMAP results combined with results predicted at 2.5 Km
from a local source). The local sources either have low effective stack height or emit vapor-phase
June 1996
5-13
SAB REVIEW DRAFT
-------
divalent mercury. Given the current understanding of the atmospheric fate and transport of emitted
mercury, qualitatively these results appear plausible.
The quantitative accuracy of the predicted fish methylmercury concentrations can not be
assessed with the available mercury monitoring data. The predicted value is on the high end of the
measured data as is expected given the modeling construct (i.e., 90th percentile RELMAP results in
the eastern half of the U.S. and 2.5 Km downwind from a large MWQ- The measured freshwater fish
mercury concentrations in the U.S. range from <0.01 ug mercury/gfto 5.94 ug mercury/g fish (NJDEP,
1994); typical values are between 0.11 and 0.26 ug mercury/g (Lowe et al., 1985; Bahnick et al.,
1994). If the predicted results are accurate (and again there are no data to conclusively demonstrate or
refuel this) then individuals who consume 60 grams of the fish at the predicted levels in Table 5-7 or
more per day may be adversely impacted from mercury emissions from the identified sources.
For the anglers who are assumed to consume a smaller quantity of local fish per day (i.e., 30
grams), the number of combinations from which they are potentially adversely impacted is fewer: 1)
in the Eastern U.S. the 50th percentile RELMAP and 2.5 Km from the CAP; 2) the 90th percentile
RELMAP and 2.5 Km from the large MWC and CAP; and 3) in the Western U.S. the 90th percentile
RELMAP results and 2.5 Km from the CAP.
Two issues must be considered for the freshwater anglers. The data of Fiore et al ., 1989
indicate that some members of the population consume both locally-caught freshwater fish and
commercial fish. This additional source of methylmercury exposure should be considered. If the
individuals are eating a variety of fish in commerce then the additional exposure will probably be
modest. If the individuals are consuming a more limited selection of commercial fish, (e.g., fish at the
apex of marine food web), which may have higher concentrations of mercury, these additional
exposures may result in adverse impacts. This again depends on the quantity and types of fish
consumed both from local waters and commercial.
The second issue for anglers is that they are assumed to fish from a variety of water bodies.
Several studies indicate that many anglers may travel extended distances to fish. These individuals
who consume fish from a variety of sources again probably decrease their chances of exposure to
methylmercury at lexicologically significant doses. There are a limited number of these mercury
emission sources and a limited number of locations predicted to be at br above the 90th percentile for
deposition of mercury.
5.4 Comparison with Other Recommendations
Recommended limits on methylmercury exposure have been expressed in these units: ug/kg
body weight/day; concentrations of mercury in tissues such as blood, hair, feathers, liver, kidney,
brain, etc.; grams of fish per day; number of fish meals per time interval (e.g., per week).
Reference values for mercury concentrations (expressed as total mercury) in biological
materials commonly used to indicate human exposures to mercury were published by the WHO/IPCS
(1990). The mean concentration of mercury in whole blood is approximately 8 ug/L, in hair about 2
ug/g, and in urine approximately 4 ug/L. Wide variation occurs about these values (WHO/IPCS,
1990).
Methylmercury exposures for general populations are reflected by hair mercury levels.
Because fish are the primary exposure pathway for methylmercury there is a broad-based scientific
June 1996 5-14 SAB REVIEW DRAFT
-------
literature associating increases in hair mercury concentrations with increases in fish consumption
(among other see: Abe et al., 1995; Akagi et al., 1995; Airey et al., 1983; Barbosa et al., 1995; Chai et
al., 1994; Girard and Dumont, 1995; Oskarsson et al., 1990; Wheatley and Paradis, 1995). The
WHO/TPCS (1990) evaluation and the U.S. EPA RfD of 0.1 ng/kg bw/day were based on
neurotoxicity to the fetus as hazard endpoint. Consequently the predictability of fetal mercury
exposure from maternal hair mercury concentrations is critical to the assessment of fetal risk from
maternal exposures. Maternal hair mercury concentrations predict mercury concentrations hi fetal
brain (Cernichiari et al., 1995); fetal blood (Cernichiari et al., 1995); umbilical cord blood (Wheatley
and Paradis, 1995; Girard and Dumont, 1995) and newborn hair (Chai et al., 1994).
There are no biological monitoring data to estimate mercury exposures (specifically hair
mercury concentrations) based on a survey that can be extrapolated to the general population of the
United States jdo not yot exist. To adequately predict methylmercury exposure for the general United
States population the data should be obtained from subjects who are chosen based on a sampling
strategy that can be extrapolated to the United States population, and must include appropriate quality
assurance/quality control procedures. There are some data available on hah- mercury concentrations
from persons living hi the United States including reports by the following: Creason et al., 1978a,
1978b, 1978c; Lasora and Citterman, 1991; and Crispin-Smith et al., 1985.
•Although data on hair mercury concentrations from a sample representative of the United
States population with adequate documentation of quality assurance/quality control do not exist, results
from individual studies conducted within the United States are shown in Table 5-8. These surveys
were conducted hi widely diverse geographic areas within the United States. Overall, the mean hair
mercury concentrations identified for subjects in these studies are typically under 1 ug/g or 1 ppm.
For a number of the surveys the detection limit was greater than 1 ppm indicating that a substantial
number of zero or trace values were included in the mean concentration. Calculating mean
concentrations when the analytical method has a detection limit near the overall mean presents
difficulties that have been treated with a variety of strategies. Although these approaches may present
a way to calculate a mean from existing data, the preferred approach is an analytical method
sufficiently sensitive and precise to provide data in the appropriate range.
The maximum values reported in these individual surveys range from 2.1 to 15.6 ppm. The
highest maximum value (15.6 ppm) was reported by Fleming et al. (1995) from a study that
specifically focussed on persons from the Florida Everglades who consumed wildlife from this area.
Until appropriate survey data for the general United States population exist, the overall pattern of hair
mercury concentrations for the United States remains unclear.
Although not from a United States population, hair mercury concentrations have been '
monitored among Canadian aboriginal peoples. Girard and Dumont (1995) summarized data on hair
mercury concentrations among the Cree Indians of Quebec sampled between 1983 and 1991. Of 626
hair samples, 112 had hair mercury concentrations greater than 2.5 ppm. In 1983 21% of mothers had
hair mercury concentrations > 6 ppm. Between 1983 and 1991 the prevalence of hair mercury
concentrations> 6 ppm decreased to 2%. Wheatley and Paradis (1995) reported on hair mercury
concentrations in Canadian Aboriginal Peoples providing cumulative results between 1970 and 1992.
During that period, 24.5% of people had hair mercury concentrations > 6 ppm, and 1.5% had hair
mercury concentrations > 10 ppm. Data were reported on 71,842 tests with regions including these:
Atlantic Provinces, Quebec, Ontario, Manitoba, Saskatchewan, Alberta, British Columbia, Northwest
Territories, and Yukon. The highest prevalence of hair mercury concentrations > 6 ppm was in
June 1996 5-15 SAB REVIEW DRAFT
-------
Table 5-8
Hair Mercury Concentrations (ug Hg/gram hair or ppm)
from Residents of Various Communities in the United States
Study
Creason et al.,
1978a
Creason et al..
1978b
Creason et al., 1978c
Airey, 1983
Airey, 1983
Community
New York
Metropolitan
Area'
•
Four communities in
New Jersey
(Ridgewood, Fairlawn,
Matawan and Elizabeth)
Birmingham, Alabama
and Charlotte, North
Carolina
USA Data cited by
Airey, 1983.
Community not
identified
USDA Data cited by
Airey, 1983.
Community identified:
LaJolla-San Diego
Mean
' Concentrationppm
Children (n=280),
0.67
Adults (n=203),
0.77
Children (n=204),
0.77
Adulfc (n-117)
0.78
Children (n=322),
0.46;
Adults (n=282),
0.47
1. Males (n=22), 2.7 ppm;
2. Females (n=16), 2.6
ppm;
3. Males and Females (24
subjects), 2.1 ppm;
4. Males and Females
(31 subjects), 2.2 ppm;
5. Males and Females
(24 subjects), Z9 ppm;
6. Males and Females
(79 subjects), 2.4 ppm;
1. 2.4 ppm (13 males)
2. 2.7 ppm (13 females);
3. 2.3 ppm (8 subjects
including males and
females);
4. 2.9 ppm (17 subjects
including males and
females);
5. 2.6 ppm ( 5 subjects
including males and
females);
6. 2.7 ppm (30 subjects
including males and
females)
Maximum
Concentration
ppm
Children, 11.3
Adults, 14.0
Children, 4.4
Adults, 5.6
^
Children, 5.4;
Adults, 7.5
1. 6.2 ppiu
2. 5.5 ppm
3. 5.6 ppm
4. 6.6 ppm
5. 7.9 ppm
6. 7.9 ppm
1. 6.2 ppm
2. 5.5 ppm
3. 4.5 ppm
4. 6.6 ppm
5. 6.2 ppm
6. 6.6 ppm
Additional
Information
on Study
Survey conducted
in 1971 and 1972
Survey conducted
in 1972 and 1973
Survey conducted
in 1972 and 1973
June 1996
5-16
SAB REVIEW DRAFT
-------
Table 5-8
Hair Mercury Concentrations (ng Hg/gram hair or ppm)
from Residents of Various Communities in the United States
Study
Airey, 1983
Airey, 1983
Crispin-Smith et aL,
1985
Lasora et aL, 1991
Lasora et al., 1991
Fleming et al., 1995
Community
USA Data cited by
Airey, 1983. Area
identified:
Maryland
USA data cited by
Airey, 1983.
Community identified:
Seattle.
USA, communities and
distribution not
identified.
Nome, Alaska
Sequim, Washington
Florida Everglades
Mean
Concentrationppm
1. 1.8 ppm (11 subjects,
males and females);
2. 1-5 ppm (11 subjects,
males and females);
3. 2.3 ppm (11 subjects,
males and females);
4. 1.9 ppm (33 subjects,
males and females)
1. 33 ppm (9 males);
2. 2.2 ppm (3 females)
3. 2.6 ppm (5 subjects,
males and females);
4. 1.5 ppm (3 subjects,
males and females);
5. 3.8 ppm (8 subjects,
males and females);
6. 3.0 ppm (16 subjects,
males and females).
0.48 ppm (1431
individuals);
0.52 (1009 individuals
reporting some seafood
consumption).
(80 females of child-
bearing age)
0.70 ppm (7 females of
child-bearing age)
1.3 ppm (330 male and
female subjects)
Maximum
Concentration
ppm
1. 3.8 ppm
2. 3.9 ppm
3. 4,5 ppm
4. 4.5 ppm
1. 5.6 ppm
2. 4.1 ppm
3. 5.6 ppm
4. 2.1 ppm
5. 7.9 ppm
6. 7.9 ppm
Maximum values in this
survey:
6.3 ppm
Maximum value:
15.2 ppm
Maximum value:
1.5 ppm
Maximum value:
15.6 ppm
Additional
Information
on Study
The 1009
individuals are a
subset of the 1431
individuals.
To be included in
the survey subjects
had to have
consumed wildlife
from the Florida
Everglades.
Northwest Territories where 42.6% of total population of 2273 had hair mercury concentrations > 6
ppm and, of those, 3.0% had hair mercury concentrations > 10 ppm. this study illustrates regional or
ethnic differences in hair mercury levels. """
5.4.1 Human Populations and Subpopulations
Cross-comparisons of limits on methylmercury exposure by various organizations are
facilitated by the work of Airey (1983) (Table 5-9) who analyzed mercury concentrations in 559
June 1996
5-17
SAB REVIEW DRAFT
-------
Table 5-9
Association Between Hair Mercury and Frequency of Fish Ingestion
Assocation of Hair Mercury Concentration (ng mercury/gram hair) with Frequency
of Fish Ingestion by Adult Male and Female Subjects Living in 32 Locations
within 13 Countries (Airey, 1983)
Frequency of Fish Meals
Once a month or less
Twice a month
Every week
Every day
Arithmetic Mean
1.4
1.9
23
11.6
Range
0.1 - 6.2
0.2 - 9.2
0.2 - 16.2
3,6 - 24.0
samples of human hair from 32 locations in 13 countries. The United States averaged 2.4 ug
mercury/gram hairjcompared with the lowest mean reported from Germany {b.5 ng mercury/gram J
and the highest from Japan(3.9 ng mercury/gram hair), Mean hair content of mercury varied with the
frequency of reported fish consumption.
5.4.1.1 Comparison with World Health Organization Recommendations
Grams of Fish per Day
The World Health Organization's International Programme for Chemical Safety (WHO/IPCS)
recommended as a preventive measure, that in populations that consume large amounts of fish (e.g.,
100 gram/day) hair levels of methylmercury in women of child-bearing age should be monitored. The
number of women of child-bearing age in the United States estimated to consume fish in excess of 100
grams per day can be obtained by inference from the general U.S. population dietary surveys (e.g.,
United States Department of Agriculture's CSFII 89/91). Intake of fish and shellfish for the general
US population (estimated by CSFII 89/91 data described in Appendix H, Volume III) based on "users"
only was estimated to be 110 grams per day at the 95th percentile among women aged 15 through 44
years. Female children (aged 14 years or younger) consumed 112 grams of fish per day, the 95th
percentile (Table 22, Appendix H, Volume in), while the overall intake at the 95th percentile among
female subjects regardless of age was 107 grams per day. Estimates of the number of women of
child-bearing age (ages 15 through 44 years) and the number of children (ages birth through 14 years)
within the United States population are shown in Table 5-10.
The number of women of child-bearing age consuming fish and/or shellfish in excess of 100
grams per day was also estimated from the NPD, Inc. 1973/74 data that recorded fish consumption for
a one-month period. Within this sample, 94% of people reported consuming fish or shellfish at least
once hi a one month period. Within this sample, the 99th percentile consumers reported an average
fish/shellfish Intake of 112 grams/day. Using the population size estimates for women of child-bearing
age from the 1990 United States Census described above estimates of the number of women of child-
bearing age (ages 15 through 44 years) and the number of children (ages birth through 14 years)
within the United States population are shown in Table 5-11.
June 1996
5-18
SAB REVIEW DRAFT
-------
Table 5-10
Estimated United States Population Consuming Fish, Excluding Alaska and Hawaii
Estimates Based on the 1990 U.S. Census and the Continuing Surveys of Food Intake
by Individuals, 1989/1991
Population Group
Total U.S. Population
Total Female Population Aged 15 through 44 Years
Total Population of Children Aged <15 Years
Estimated Number of Persons
247,052.000
58,222,000
53,463,000
Percent of Respective Group
» Reporting Fish Consumption during »
the 3-Day Dietary Survey Period in CSFU 89/91*
Total Population
Females Aged 15 through 44 Years
Children Aged <15 Years
30.9 percent
30.5 percent
24.9 percent
Number of Persons Predicted to Consume Fish Based
an Percentage Consuming Fish in CSFII 89/91
Total Estimated Population
Total Estimated Number of Females Aged 15 through 44 Years
Total Estimated Number of Children Aged <15 Years
76,273,000
17,731,000
13,306,000
Number of Persons in
Highest 5 Percent of Estimated Population that Consumes Fish*1
Total Estimated Population
Total Estimated Female Population Aged 15 through 44 Years
Total Estimated Child Population
3,814,000
887,000
665,000
Estimated Number of Adult Pregnant Women in Highest 5 Percent Of Estimated Population that Consumes Fish
Number of Females Aged 15 through 44 Years x Percentage of Women Pregnant
in a Given Year
84,300
" Rounded to three significant figures.
b Persons who consume an average 100 g or more of-fish/day.
June 1996
5-19
SAB REVIEW DRAFT
-------
Table 5-11
Estimated Fish-Consuming Population
in the United States, excluding Alaska and Hawaii
Estimates Based on the 1990 U.S. Census and the
National Purchase Diary Inc., 1973/74 Data on Fish/Shellfish Consumption
Population Group
Total U.S. Population
Total Female Population Aged 15
through 45 Years
Total Population of Children Aged < 15 Years
Estimated Number of Persons
247,052,000
58,222,000
53,462,000
Percent of Respective Group
Reporting Fish Consumption during
the One-Month Survey period in NPD, Inc. 1973/64 Survey
Total Population
Females Aged 15 through 45 Years
Children Aged < 15 Years
94%
94%
94%
Number of Persons Predicted to Consume Fish Based
on Percentage Consuming Fish or Shellfish in NPD, Inc. 1973/74
Total Estimated Population
Total Estimated Number of Females
Aged 15 through 45 Years
Total Estimated Number of Children
Aged < 15 Years
232,229,000
54,729,000
50,254,000
Number of Persons in
Highest One Percent of Estimated Fish-Consuming Population1
Total Estimated Population
Total Estimated Adult Female Population
Total Estimated Child" Population
2,322,000
547,000
503,000
Estimated Number of Adult Pregnant Women in
Highest One Percent of Estimated Fish-Consuming Population
Number of Adult Females x Percentage of Women Pregnant
in a Given Years
51,900
a Persons who consume an average 100 g or more or fish/day.
June 1996
5-20
SAB REVIEW DRAFT
-------
The estimated number of women of child-bearing age (ages 15 through 44 years) in the
contiguous 48 states is approximately 17,731,000. It is estimated that in a given year 9.5 percent of
women in this age group are pregnant. The groups of concern are women of child-bearing age who
are pregnant and children. Using consumption of 100 grams of fish/shellfish per day or more as a
screen for concern for mercury exposure estimates have been made of the number of women and
children whose fish/shellfish intake is at or above 100 grams/day. Within the CSFII 89/91 cross-
sectional data that identified 30.5% of females of child-bearing age consuming fish/shellfish at 100
grams or more in the 3-day sampling window, the number of persons with these consumption patterns
were identified. The estimated number of women ages 15 through 44 years in the highest 5 percent of
fish consumers (those consuming 100 g or more of fish per day) is approximately 887,000 based on
CSFH 89/91 data. The estimated number of pregnant women in that highest 5% of fish consumers is
estimated to be approximately 84,300.
Whether or not a 3-day sampling window adequately projects longer-term fish consumption
has been raised as a question. Data are available from the National Purchase Diary, Inc. 1973/94
sample that recorded fish/shellfish consumption for a month-long period. Within these data, 94% of
people reported ingesting fish at least once during the period. Within these 94%, the 99th percentile
intake was reported (Rupp et al., 1980) to be 112 grams/day. Using this top 1% of consumers the size
of the population of women of child-bearing age and children consuming 100 grams of fish/shellfish
per day or more was estimated. The estimated number of pregnant women consuming 100 grams of
fish/shellfish per day or more was estimated to be 51,900.
The USDA reports that overall fish/shellfish consumption in the United States has increased by
26% between 1970 and 1990. The estimated number of pregnant women who consume 100 grams/day
or more of fish is, thus, likely greater than 51,900. If the general increase of 26% applies at the
extremes of the distribution, the projected number of women would be approximately 65,000 women.
l
Local point sources for emissions of mercury can be most clearly linked to localized
deposition of mercury. An analysis of the CSFII 89/91 data by personnel from US EPA's Office of
Prevention, Pesticides, and Toxic Substances (personal communication, Helen Jacobs) determined that
at the mean 33% of total fish/shellfish intake identified in this survey came from freshwater and
estaurine fish and shellfish. Subpopulations of anglers and subsistence fishers have been assumed to
obtain most of their self-caught fish and shellfish from these local and estaurine sources.
Specific subpopulations of anglers and subsistence fishers and other high end fish consumers
ingest fish substantially in excess of the general population. Figure 4-5 (p. 4-41) summarizes grams of
fish consumed among specific subpopulations and highlights high end consumption. For example,
Puffer et al. (1981) in a study of anglers in Los Angeles, California found that mean intake was 37
grams per day, but the 90th percentile for this group was 225 grams per day. Orientals and Samoans
had mean fish intakes with a mean of 70.6 grams/day (Puffer et al. 1981). Alaskan Natives from 11
communities averaged 109 grams of fish/day (Nobbman et al., 1992). Wolfe and Walker identified a
very high fish consumption rate among persons living in remote Alaskan communities. The Columbia
River Intertribal Fish Commission (1994) reported that during the two months of highest average fish
consumption average intake was 108 grams/day. The Tribes of Puget Sound reported (Toy et al.,
1995) an average of 73 grams/day with a 90th percentile of 156 grams/day. West et al. (1989) found
a mean intake of approximately 22 grams/day, but a reported maximum value over 200 grams/day.
Peterson et al (1994) in a study of Chippewa tribes found that 2 percent of 323 respondents ate at least
one fish meal each day. In these individual tribal and angler studies, data were generally not
separately reported for women of child-bearing age.
June 1996 5-21 SAB REVIEW DRAFT
-------
Micrograms of Methylmercury per Day
The WHO/IPCS cited (WHO/IPCS, 1990) a reanalysis of the Iraqi data by Nordberg and
Strangert (1976, 1978, 1982 as cited in WHO/IPCS, 1990). This analysis included consideration of
inter-individual variation in mercury half-life and assumed a continuous distribution of individual
thresholds superimposed on a background frequency of symptoms such as paresthesia. These
calculations indicate that an intake of 50 ug methylmercury per day in an adult will not result in any
adverse effects to adults. A long term methylmercury intake of 3 to 7 ug/kg body weight/day would
adversely effect nervous system function as manifested by <5 percent increase in incidence of
paresthesia in adults. Hair mercury concentrations would be approximately 50 to 125 ug/gram at this
exposure. For the general population of the United States (Appendix H, Volume HI), among women
of child-bearing age, the 97th percentile ingestion level of methylmercury was estimated to be
approximately 0.48 ug/kg body weight/day. The intakes at the 95th percentile of methylmercury by
children ages 14 years or younger was 0.84 ug/kg/day for males and 0.81 ug/kg body weight/day for
females, based on the freshwater methytaercury concentration data of Bahnick et al. (1985). Using
the freshwater fish mercury concentration data of Lowe et al. (1994), the 95th percentile exposures
were 0.84 ug mercury/kg bw for males and 0.68 ug/kg bw for females ages 14 years and younger.
Micrograms of Mercury per Gram of Hair
Biological monitoring based on hair, blood and/or concentrations in other tissues has been used
as an index of body burden of mercury. These concentrations are used as a surrogate for mercury
concentrations in tissues proximate to the site of effect (e.g., mercury concentrations in nervous system
tissue). The WHO/IPCS has concluded (1990) that the general population does, not face a significant
health risk from methylmercury. When fish consumption is high enough for groups to attain a blood
methylmercury level of about 200 ug/L (corresponding to 50 ug/g hah-) a low (5 percent) risk of
neurological damage will occur. In 1995 Kinjo et al. reported threshold values hair mercury based on
logit and hockey stick analyses for calculated maximum hair mercury concentrations from human
subjects in the Niigata epidemic of Minamata disease in Japan. Male adults were calculated to have
threshold values (ug/g hair) (95 percent CI) of 46.5 (30,71) and 43.0 (27,67) depending on whether or
not patients with estimated maximum hair mercury concentrations of less than 20 ug mercury/gram
hair were included. Calculated threshold values for adult women were 24.7 (20,30) or 49.3 (30,64)
with and without inclusion of patients with estimated maximum values of less than 20 ug/g.
[Exclusion of hair mercury concentrations less than 20 fig mercury/gram hair was based on
unreliability of the analytical method (dithizone colorimetric techniques) at these concentrations]. Of
the 986 subjects reported by Kinjo et al. (1995) 26 had hair mercury concentrations less than 20 ug
mercury/gram hair.
Clinical observations in Iraq suggest that women during pemancy are more sensitive to the
effects of methylmercury with fetuses at particularly increased risk. The World Health
Organization/International Programme for Chemical Safety (1990) indicated, based on analysis of the
Iraqi data, a 30 percent or higher risk of abnormal neurological signs when maternal hair mercury
concentrations were above 70 ug/g. These abnormal neurological signs were the following: increased
muscle tone in the leg and exaggerated deep tendon reflexes, often accompanied by ataxia together
with a history of developmental delays. The WHO/IPCS (1990) evaluation indicated that data from
the Iraqi epidemic do not permit conclusions about risk of adverse effects below this level. However,
using statistical methods for biological modeling by Cox et al. (1989) and other data, WHO calculated
that a maternal hair concentration of 10 to 20 ug/g implies a 5 percent risk of neurological disorder.
June 1996 5-22 SAB REVIEW DRAFT
-------
Extrapolation of these data to lower mercury concentrations is uncertain, but psychological and
behavioral testing of subjects may identify subclinical effects.
5.4.1.2 Comparison with U.S. EPA's RfD and Benchmark Dose
The RfD and benchmark dose for methylmercury were based on the Iraqi data. Dose-
conversion calculations were used to convert data on hair mercury concentration to estimates of blood
mercury concentration and dietary intake (jag/day) of methylmercury. The RfD/RfC Work Group
chose a benchmark (lower bound for 10 percent risk) based on modeling of all nervous-system effects
in children. The 10 percent risk level was 11 ppm hair concentration for methylmercury. A dose-
conversion equation was used to estimate a daily intake of 1.1 ug methylmercury/kg body weight/day
that when ingested by a 60 kg individual is predicted to maintain a blood concentration of
approximately 44 \igfL or a hair concentration of 11 ug mercury/gram hair (11 ppm).
The benchmark dose can be compared with other recommended limits and with data on
methylmercury exposure via fish. The benchmark dose of 11 ug methylmercury/gram of hair
compares closely with the arithmetic mean of hair mercury concentration (Airey, 1983) for people
from many different communities who eat fish every day (11.6 ug mercury/gram hair). If a 100 gram
portion size for fish is assumed, this corresponds with quantities of fish eaten by approximately the
95th percentile fish consumer among persons who reported consuming fish in the CSFII 89/91.
Expressed another way the benchmark dose (see also Volume VI, Chapter 2, pg. 10) is 1.1 ug/kg body
weight/day assuming a 60 kg body weight individual. The benchmark dose was used as an estimate of
a NOAEL.
Comparison with the General Population
Among women of child-bearing age in the general U.S. population (as estimated from CSFII
89/91 food consumption data and fish mercury concentrations presented in Appendix H, Volume III),
only women reporting greater than 99th percentile consumption would exceed the benchmark dose for
methylmercury intake from fish, when methylmercury intake is expressed on a per kilogram body
weight basis.
Two issues need to be noted regarding these comparisons. Estimated dietary intakes at the
99th percentile and higher are at the extremes of the distribution. Short-term dietary intakes based on
three-days food consumption records are known to be quite subject to variability at the extremes of the
distribution. Consequently these data must be interpreted with caution until they can be confirmed or
repudiated with additional exposure assessments.
It must be noted that the benchmark was calculated to be a lower-bound on an effect level:
occurrence of neurological effects associated with prenatal exposure to methylmercury at a 10 percent
risk level. By contrast the RfD is an exposure level that is assumed to be protective against known
adverse effects for all members of the population including sensitive subpopulations. The RfD for
methylmercury is 0.1 fig/kg body weight/day. The RfD may be considered the midpoint in an
estimate range of about an order of magnitude. This range reflects variability and uncertainty in the
estimate.
June 1996 5-23 SAB REVIEW DRAFT
-------
Comparison with Populations Consuming Large Amounts of Fish
In the review of published data on fish-consumption among subpopulations who consume fish
more frequently than the general population, a number of reports were identified who consume
substantially higher quantities of fish than among the general population. These groups were identified
when the recommendation to monitor populations consuming one fish-meal a day (or 100 grams of
fish per day) was evaluated. Most of these reports do not provide a clear identification of the age and
gender of their subjects. However, to the extent that these subjects are women of reproductive age (15
through 44 years) the likelihood that they will exceed the benchmark dose for methylmercury depends
on the methylmercury concentration of the fish consumed.
Depending on whether'or not the fish obtained by a high-end fish consumer come from one
source (e.g., a small lake or local river) or from simply more of the general food supply, the mercury
concentration of the fish obtained may or may not be site-specific. Assuming a high-end fish
consumer obtains a broad mixture of fish sources, the mean mercury concentration of the fish
consumed is estimated to be about the mean or median value for the fish mercury concentrations used
hi the estimates for Appendix H of Volume IV. More precise estimates of mercury intake for these
subpopulations will require site-specific determinations of mercury in the fish consumed.
5.4.2 Wildlife Species
The Great Lakes Water Quality Initiative Criteria (GLWQI Criteria) were described in Volume
IV (Section 4.2) of this Mercury Study Report to Congress. The evaluation of data and calculation of
water concentrations (WC) in the Mercury Study Report to Congress was done in accordance with the
methods and assessments published in the draft GLWQI (U.S. EPA 1993a). Availability of additional
data and differences in interpretation of those data led to differences in the calculated values of the
WC in this Report and those published hi the final GLWQI (U.S. EPA, 1995b). Both evaluations used
the same methodology which was described in Section 4.2.1 of Volume IV. These two evaluations
relied on the same experimental studies as the basis for the WC calculation: for birds, the three
generation reproduction study in mallards (Heinz, 1974, 1975, 1976a,b, 1979); and for mammals the
subchronic dietary studies in mink (Wobeser et al., 1976a,b). In addition to these studies, the authors
of the Mercury Study Report to Congress were able to obtain Wobeser's dissertation (Wobeser, 1973);
this provided some additional information that was augmented by discussions with the author.
A comparison between the species-specific Wildlife Criteria Calculated in the Great Lakes
Water Quality Initiative and the Mercury Study Report to Congress was presented in Volume IV
(Table 4-3, pg. IV-15, repeated here as Table 5-12).
June 1996 5-24 SAB REVIEW DRAFT
-------
Table 5-12
Comparison of Wildlife Criteria Calculated by Great Lakes
Water Quality Initiative and by the Mercury Study
Species
Mink
Otter
Kingfisher
Osprey
Eagle
Wildlife Criterion
(pg/L) '
GLWQI
2880
1930
1040
Not done
1920
MSRC
415
278
193
483
538
All of the WC calculated in this Report are lower (more conservative) than those published iii
the GLWQI. All species-specific WC, however, differ less than an order of magnitude from one
another. Range in differences is from nearly 4-fold lower for the WC in this Report (eagle) to 7-fold
lower (mink and otter). Variation in the calculated WC are from two sources: evaluation of effects in
wildlife and evaluation of exposure to wildlife.
Details of differences between the GLWQI and this Report on evaluation of effects in birds
and piscivorous mammals have been presented in Volume V. For birds the GLWQI used a different
rate of food consumption 0.156 kg/kg-d compared with 0.128 kg/kg-d in this Report) and different
uncertainty factors than did the Mercury Study Report to Congress. In the effects assessment for
piscivorous mammals both the GLWQI and this Report used data on mink administered mercury in the
diet from the studies of Wobeser (1976a,b).
The Report also obtained the doctoral thesis of Wobeser (Wobeser, 1973). The GLWQI
identified a NOAEL of 1.1 ppm. At this dietary exposure there were changes in the liver, lesions in
the central nervous system and axonal degeneration; moreover, two of the animals in this treatment
group were observed at the end of the treatment of move slowly by comparison to other mink. The
study authors reported their opinion that mink treated at 1.1 ppm in the diet for longer than the study
(93 days) would be expected to show clinical signs of nervous system damage. Mink treated at the
next higher dose, 1.8 ppm, were observed with anorexia, ataxia and increased mortality. Based on
these considerations, this Report considered 1.1 ppm to be the LOAEL, and as described hi Section
4.2.2 of Volume IV, used data from the first part of the study to identify a NOAEL of 0.33 ppm. This
Report used data from Wobeser (1973) to establish the weights of female mink and kits used in-these
experiments; this results in slight differences in conversion of dose in ppm diet to ug/kg bw/day.
Another difference between the GLWQI and the Mercury Study Report to Congress was
through assessment of exposure to birds through consumption of prey. The GLWQI made assessments
specific to the Great Lakes region. Because the Mercury Study Report to Congress is a national
June 1996
5-25
SAB REVIEW DRAFT
-------
assessment use of region-specific assumption was not considered appropriate. Additional information
on these differences is found in Volume IV (Section 4.2).
5.5 Risk Characterization Issues
The results of this risk characterization method pose a series of questions applicable to
discussion of management of potential risk.
1. Should environmental mercury concentrations be managed based on the risks estimated
for humans that eat fish or for wildlife that each fish?
2. If potential piscivorous wildlife risks are the basis selected, then the risk manager must
determine which wildlife species are present and which wildlife species health endpoints
to protect
3. If potential risks to humans are the basis selected, then the risk manager must determine
the level of consumption of local fish by humans and which human health endpoints to
protect.
4. What level of uncertainty can be tolerated for a given situation?
Table 5-9 shows that the RfD or Wildlife Criteria Values are reached at fish concentration at
least nine times lower than for the estimated LOAELs for the wildlife species and humans. The RfD
is, by definition, a protective endpoint. The risks posed by exceedance of this value are uncertain.
The LOAEL, on the other hand, is the lowest observed effect level. Exposure at the rate of the
LOAEL is predicted to cause adverse effects in some members of the populations.
Table 5-9 shows that selection of the Wildlife Criteria Value for kingfisher protects for the
Wildlife Criterion Values of all other selected species and the human RfD, as well as the LOAEL at
the modeled rates for human local fish consumption. Similarly selection of the human RfD based on
an estimate of 60 grams of fish consumption per day is expected to be protective of all wildlife
considered at the Wildlife Criteria Values except for the kingfisher, as well as for the LOAELs.
Selection of any LOAEL is not protective for the human RfD or Wildlife Criteria Values. These are
decisions of risk management.
June 1996 5-26 SAB REVIEW DRAFT
-------
6. CONCLUSIONS
The following conclusions are presented in approximate order of degree of certainty in the
conclusion, based on the quality of the underlying database. The conclusions progress from
those with greater certainty to those with lesser certainty.
• There is a plausible link between methylmercury concentrations in freshwater fish and
anthropogenic mercury emissions. The degree to which this linkage occurs cannot be
estimated quantitatively at this time.
• Among humans and wildlife that consume fish, methylmercury is the predominant chemical
species contributing to mercury exposure.
• Methylmercury is known to cause neurotoxic effects in humans via the food chain.
• The human RfD for" methylmercury is calculated to be IxlO"4 mg/kg body weight/day. While
there is uncertainty in this value, there are data and quantitative analyses of health endpoints
that corroborate and support a reference dose within a range of an order of magnitude. A
quantitative uncertainty analysis indicates that the human RfD based on observation of
developmental neurotoxicity in children exposed to methylmercury in utero is likely to be
protective of human health.
• The RfD is a confident estimate (within a factor of 10) of a levels of exposure without adverse
effects on those human health endpoints measured in the Iraqi population exposed to
methylmercury from grain. These included a variety of developmental neurotoxic signs and
symptoms. The human RfD is for ingested methylmercury; no distinction was made regarding
the food hi or other media serving as the ingestion vehicle.
• U.S. EPA calculates that members of the U.S. population ingest methylmercury through the
consumption of fish at quantities of about 10 times the human reference dose. This amount of
methylmercury is equivalent to the benchmark dose used hi the calculation of the reference
dose; the benchmark dose was taken to be an amount equivalent to the NOAEL. The NOAEL
was an ingested amount of 1.1 ug per kg body weight per day. Consumption of mercury
equivalent to the NOAEL is predicted to be without harm for the majority of a population.
Individual risks cannot be determined from the available data.
• Prediction of risk cannot be made for ingestion of methylmercury above the benchmark dose,
given the currently available data in humans.
• Concentrations of mercury in the tissues of wildlife species have been reported at levels
associated with adverse health effects in laboratory studies hi the same species.
• Dietary survey data indicate that approximately 30 percent of the general U.S. population
consumes fish at least once during a three-day period. Among this group of fish consumers
roughly 50 percent are predicted to consume methylmercury at the RfD. Consuming
methylmercury at levels equal to the RfD is equated to be without harm.
• Dietary intake data from cross-sectional surveys indicate that approximately 30 percent of the
general U.S. population consumes fish at least once during a three-day period. Among this
June 1996 6-1 SAB REVIEW DRAFT
-------
group of fish consumers the majority are predicted to consume methylmercury at or below the
RfD. Consuming methylmercury at levels equal to the RfD is expected to be without harm.
• Based on longer-term data survey data that recorded fish consumption for a one-month period,
approximately 94% of the population consumes fish at least once during that period.
• Using both the longitudinal and cross-sectional survey data, it is estimated that 1 to 2 percent
of women of child-bearing age regularly consume fish and shellfish at average intakes of 100
grams per day or greater. Whether or not methylmercury intakes are elevated above the
estimated NOAEL depends on the concentration of methylmercury in the fish and shellfish
consumed.
• U.S. EPA estimates that approximately one-third of fish and shellfish consumed are from
freshwater/estaurine habitats that may be affected by local sources of mercury.
• Case reports in the literature document that sick and/or dying animals and birds with seriously
elevated tissue mercury concentrations have been found in the wild. These wildlife have
mercury concentrations elevated to a level documented in laboratory studies to produce adverse
effects in these species, for a specific case report concurrent exposure to other sources of ill
health cannot be excluded.
• Modeled estimates of mercury concentration in fish around hypothetical mercury emissions
sources predict exposures at the wildlife WC. The wildlife WC, like the human RfD, is
predicted to be a safe dose over a lifetime. It should be noted, however, that the wildlife
effects used as the basis for the WC are gross clinical manifestations or death. Expression of
subtle adverse effects at these doses cannot be excluded.
• Data are not sufficient for calculation of separate reference doses for children, in utero
exposure and the aged.
• Comparisons of dose-response and exposure estimates through the consumption of fish indicate
that certain species of piscivorous wildlife are more exposed.on a per kilogram body weight
basis than are humans. The implications for wildlife health are uncertain.
There are many uncertainties associated with this analysis. The sources of uncertainty include
the following:
• There is considerable uncertainty and apparent variability in the movement of mercury from
the abiotic elements of the aquatic system through the aquatic food chain.
• U.S. EPA has developed a BAF in an attempt to quantify the relationship between dissolved
mercury concentrations in the water column and methylmercury concentrations hi fish. This
BAF was developed using a four-tier food chain model and extant field data. A quantitative
uncertainty analysis of the BAF and the variability of the BAF was examined.
• There is considerable uncertainty in atmospheric processes that affect emitted mercury. U.S.
EPA has attempted to predict the fate and transport of mercury through the use of atmospheric
models. The results of these models are uncertain. For the regional (RELMAP) modeling,
June 1996 6-2 SAB REVIEW DRAFT
-------
predicted mercury concentrations are corroborated by measured data for certain areas of the
United States.
• A quantitative uncertainty analysis and qualitative considerations lead to the conclusion that
paresthesia in adults is not the most reliable endpoint on which to base a quantitative dose-
response assessment. A quantitative uncertainty analysis and qualitative considerations also
indicate that late walking in children is less reliable than combined developmental effects in
children exposed in utero.
• Total sources of exposure for selected populations may include occupational exposure
primarily to mercury vapor. Exposures from dental amalgam are expected to contribute to the
overall body burden of mercury. The association, however, between overall body burden of
mercury from these sources and methylmercury from the aquatic food chain is not established.
« Data estimating body burden of mercury based on biological monitoring of hair and blood
mercury levels among the general U.S. population have not been published. Such information
would permit firmer estimates of the risk of mercury toxicity in the general U.S. population.
• Data on body burden of mercury among populations that consume large quantities of fish are
also very limited. Such information would permit firmer estimates of risk of mercury toxicity
for these specific high-risk populations.
To improve the risk assessment for mercury and mercury compounds, U.S. EPA would need the
following:
• A monitoring program .to assess either blood mercury or feather/hair mercury of piscivorous
wildlife; particularly those in highly impacted areas. This program should include assessment
of health endpoints including neurotoxicity and reproductive effects.
• Collection of additional monitoring data on hair or blood mercury and assessment of health
endpoints among women of child-bearing age and children. This study should focus on high-
end fish consumers and on consumption of fish from contaminated water bodies.
• There is a need for improved data on effects that influence survival of the wildlife species as
well as on individual members of the species.
• There is a need for controlled studies on mercury effects in intact ecosystems.
• Monitoring data sufficient to validate or improve the local impact exposure models are needed.
June 1996 6-3 SAB REVIEW DRAFT
-------
7. RESEARCH NEEDS
The primary purpose of the Mercury Study Report to Congress was to assess the impact of
U.S. anthropogenic emissions on mercury exposure to humans and wildlife. The size of some
populations of concern have been estimated: namely women of child-bearing age and children who
eat fish. In the general population, people typically obtain their fish from many sources. The question
on whether or not the impact of mercury from anthropogenic ambient emissions can be proportioned
to the overall impact of methylmercury on wildlife is a much more difficult issue.
As with environmental monitoring data, information on body burden of mercury in populations
of concern (blood and/or hair mercury concentrations) are not available for the general U.S.
population. Data on higher-risk groups are currently too limited to discern a pattern more predictive
of methylmercury exposure than information on quantities of fish consumed. The selenium content of
certain foods has been suggestive as a basis for modifying estimates of the quantities of
methylmercury that produce adverse effects. Currently, data on this mercury/selenium association
form an inadequate basis to modify quantitative estimates of human response to a particular exposure
to mercury.
Available data for human health risk assessment have limitations as described in the Report
and in this summary. Studies of human fish-consuming populations in the Seychelles and Faroes
Islands address some of these limitations; they are expected to be published within a year of release of
this Report. Additional studies on U.S. populations who consume fish from the Great Lakes are in
progress. Public health agencies of the U.S. government as well as the U.S. EPA will evaluate these
new data when they are available. Risk management decisions beyond the ongoing activities specified
in the Clean Air Act Amendments of 1990 will be based on consideration of all human data including
results of these new studies.
The benchmark dose methodology used in estimating the RfD required that data be clustered
into dose groups. Most data on neurologically based development endpoints are continuous; that is,
not assigned to dose groups. For example, scoring on scales of IQ involves points rather than a
"yes/no" type of categorization. Measurements on the degree of constriction of the visual field involve
a scaling rather than a "constricted/unconstricted" type of variable. Although arbitrary scales can be
constructed, these groupings have generally not been done in current systems. Use of alternative dose
groupings (as described in Volume IV) had no significant effect on calculated benchmark doses. An
additional difficulty occurs in estimation of benchmark dose for multiple endpoints that have been
measured. Further research on appropriate methods for mathematical modeling is needed. For some
situations such information is known, but for methylmercury exposure and multiple endpoints
assessing the same system (i.e., developmentally sensitive neurological, neuromotor and
neuropsychological effects) the time-course/dose-response of such changes have not been clearly
established. Development of the mathematical models needs to be accompanied by understanding the
physiological/pathological processes of methylmercury intoxication.
Research to decrease the above uncertainties and to address characterization limitations include
the following:
• A monitoring program to assess either blood mercury or feather/hair mercury of
piscivorous wildlife; particularly those in highly impacted areas. This program should
include assessment of health endpoints including neurotoxicity and reproductive effects.
June 1996 7-1 GAB REVIEW DRAFT
-------
Collection of additional monitoring data on hair or blood mercury and assessment of
health endpoints among women of child-bearing age and children. This study should
focus on high-end fish consumers and on consumption of fish from contaminated water
bodies.
There is a need for improved data on effects that influence survival of the wildlife
species as well as on individual members of the species.
There is a need for controlled studies on mercury effects in intact ecosystems.
Monitoring data sufficient to validate or improve the local impact exposure models are
needed.
June 1996 7-2 SAB REVIEW DRAFT
-------
8. REFERENCES
Abe, T., R. Ohtsuka, T. Kongo, T. Suzuki, C. Tohyama, A. Nakano, H. Akagi and T. Akimichi. 1995.
High hair and urinary mercury levels of fish eaters in the nonpolluted environment of Papua New
Guinea. Archives of Environmental Health 50:367-373.
Airey, D. 1983. Total mercury concentrations in human hair from 13 countries in relation to fish
consumption and location. Sci. Total Enviroa 31:157-180.
Akagi, H., O. Malm, F.J.P. Branches, Y. Kinjo, Y. Kashima, J.R.D. Guimaraes, R.B. Oliverira, K.
Haraguchi, W.C. Pfeiffer, Y. Takizawa and Y. Kato. 1995. Human exposure to mercury due to gold
mining in the Tapajos River Basin, Amazon, Brazil: Speciation of mercury in human hair, blood and
urine. Water, Air, and Soil Pollution 80:85-94.
Albers, J.W., L.R. Kallenbach, LJ. Fine et al. 1988. Neurological abnormalities associated with
remote occupational elemental mercury exposure. Ann. Neurol. 24(5):651-659.
Allen, B.C., R.J. Kavlock, C.A. Kimmel and E.M. Faustman. 1994. Dose-response assessment for
developmental toxicity. II. Comparison of generic benchmark dose estimates with no observed effect
levels. Fund. Appl. Toxicol. 23:487-495.
Al-Shahristani, H. and K.M. Shihab. 1974. Variation of biological half-time of methylmercury in
man. Arch. Environ. Health. 28:342-344.
Amin-Zaki, L., S. Elhassani, M.A. Majeed, T.W. Clarkson, R.A. Doherty, M. Greenwood and T.
Giovanoli-Jakubczak. 1976. Perinatal methylmercury poisoning hi Iraq. Am. J. Dis. Child.
130:1070-1076.
Amin-Zaki, L., S. Elhassani, M.A. Majeed, T.W. Clarkson, M. Greenwood and R.A. Doherty. 1979.
Prenatal methylmercury poisoning. Am. J. Dis. Child. 133:172-177.
Amin-Zaki, L., M.A. Majeed, M.R. Greenwood, S.B. Effimsani, T.W. Clarkson and R.A. 1981.
Methylmercury poisoning in the Iraqi suckling infant: A longitudinal study over five years. J. Appl.
Toxicol. 1(4):210-214.
Anderson, O. 1983. Effects of coal combustion products and metal compounds on sister chromatid
exchange (SCE) in a macrophage-like cell line. Environ. Health Perspect. 47:239-253.
Anderson, Sue See page 8-19. Life Sciences Research Office, 1986.
Andres, P. 1984. IgA-lgG disease in the intestine of Brown-Norway rats ingesting mercuric chloride.
Clin. Immuno. Immunopath. 30:488-494.
Ashe, W.F., E.J. Largent, F.R. Dutra, D.M. Hubbard and M. Blackstone. 1953. Behavior of mercury
in the animal organism following inhalation. Ind.Hyg.Occup.Med. 17:19-43.
Aulerich, R.J., R.K. Ringer and S. Iwamoto. 1974. Effects of dietary mercury in mink. Arch.
Environ. Contain. Toxicol. 2(1):43-51.
June 1996 8-1 SAB REVIEW DRAFT
-------
Bahnick, D., C. Sauer, B. Butterworth and D. Kuehl. 1994. A national study of mercury
contamination of fish. Chemosphere. 29(3):537-546.
Bakir, F., S.F. Damluji, L. Amin-Zaki, M. Murtadha, A. Khalidi, N.Y. Al-Rawi, S. Tikriti, H.I. .
Dhahir, T.W. Clarkson, J.C. Smith and R.A. Doherty. 1973. Methylmercury poisoning in Iraq.
Science. 181:230-241.
Baranski, B. and I. Szymczyk. 1973. [Effects of mercury vapor upon reproductive functions of
female white rats]. Med. Pr. 24:248. (Czechoslovakian)
Barbosa, A.C., A.A. Boischio, G.A. East, I. Ferrrari, A. Goncalves, P.R.M. Silva and T.M.E. DaCruz.
1995. Mercury contamination in the Brazilian Amazon. Environmental and occupational aspects.
Water, Air and soil Pollution 80:109-121.
Barnes, D.G. and M.L. Dourson. 1988. Reference dose (RfD): Description and use in health risk
assessment. Reg. Toxicol. Pharmacol. 8:471-486.
Barr, J.F. 1986. Population dynamics of the common loon (Gavia immer) associated with
mercury-contaminated waters in northwestern Ontario. Occasional Paper No. 56, Canadian Wildlife
Service. Minister of Supply and Services, Canada, 1986. Catalogue No. CW69-1/56E.
Beliles, R.P., R.S. Clark, P.R. Belluscio, C.L. Yuile and L.J. Leach. 1967. Behavioral effects in
pigeons exposed to mercury vapor at a concentration of 0.1 mg/m3. Am. Ind. Hyg. J. 28(5):482-484.
Benoit, J.M. W.F. Fitzgerald and A.W.H. Damman. 1994. Historical atmospheric mercury deposition
in the Mid-Continental U.S. as recorded in an Ombrotrophic Peat Bog. In: Mercury Pollution
Integration and Synthesis, C.J. Watras and J.W. Huckabee, Ed. p. 187-202.
Berg, W., A. Johnels, B. Sjostrand and T. Westermark. 1966. Mercury content in feathers of Swedish
birds from the past 100 years. Oikos. 17:71. (Cited in Lindqvist, 1991)
Berlin, M., J. Fazackerley and G. Nordberg. 1969. The uptake of mercury in the brains of mammals
exposed to mercury vapor and to mercuric salts. Arch. Environ. Health. 18:719-729.
Bernard, A.M., H.R. Roels, J.M. Foldart and R.L. Lauwerys. 1987. Search for anti-laminin antibodies
in the serum of workers exposed to cadmium, mercury vapour or lead. Int. Arch. Occup. Environ.
Health. 59:303-309.
Bernaudin, J.F., E. Druet, P. Druet et al. 1981. Inhalation or ingestion of organic or inorganic
mercurials produces auto-immune disease in rats. Clin. Immunol. Immunopathol. 20:129-135.
Birke, G., A.G. Johnels, L-O. Plantin, B. Sjostrand, S. Skerrving and T. Westermark. 1972. Studies
on humans exposed to methylmercury through fish consumption. Arch. Environ. Health. 25:77-91.
Bleavins, M.R. and R.J. Aulerich. 1981. Feed consumption and food passage in mink (Mustela visori)
and European ferrets (Mustela putorius furo). Lab. Animal Sci. 31:268-269.
Borg, K., H. Wantorp, K. Erne and E. Nako. 1970. Alklylmercury poisoning in terrestrial Swedish
wildlife. Viltrevy. 6:301-379.
June 1996 8-2 SAB REVIEW DRAFT
-------
Bornhausen, M., M.R. Musch and H. Greim. 1980. Operant behavior performance changes in rats
after prenatal methyl mercury exposure. Toxicol. Appl. Pharmacol. 56:305-316.
Borst, H.A. and C.G. Lieshout. 1977. Phenylmercuric acetate intoxication in mink. Tijdschr.
Diergeneesk. 102:495-503.
Bowerman, W.W., E.D. Evans, J.P. Gisey and S. Postupalsky. 1994. Using feathers to assess risk of
mercury and selenium to bald eagle reproduction in the Great Lakes Region. Arch. Environ. Contain.
Toxicol. 27:294-298.
Buchet, J.P., H. Roels, A. Bernard and R. Lauwerys. 1980. Assessment of renal function of workers
exposed to inorganic lead, cadmium or mercury vapor. J. Occup. Med. 22(11):741-750.
Bunn, W.B., C.M. McGill, T.E. Barber, J.W. Cromer and L.J. Goldwater. 1986. Mercury exposure in
chloralkali plants. Am. Ind. Hyg. Assoc. J. 47(5):249-254. *
Burbacher, T.M., C. Monnett, K.S. Grant et al. 1984. Methyl mercury exposure and reproductive
dysfunction in the nonhuman primate. Toxicol. Appl. Pharmacol. 75:18-24.
Burbacher, T.M., M.K. Mohamed and N.K. Mottett. 1988. Methyl mercury effects on reproduction
and offspring size at birth. Reprod. Toxicol. l(4):267-278.
Burger, J., J.A. Rodgers and M. Goehfeld. 1993. Heavy metal and selenium levels in edangered
wood storks Mycteria americans from nesting colonies in Florida and Costa Rica. Arch. Environ.
Contain. Toxicol. 24:417-420.
Burger, J., C.T. Nisbet and M. Goehfeld. 1994. Heavy metal and selenium levels in feathers of
knownaged common terns (Sterna hirundd). Arch. Environ. Contain. Toxicol. 26:351-355.
Byerly, E.R. 1993. State Population Estimates by Age and Sex: 1980-1992, U.S. Bureau of the
Census, Current Population Reports P25-1106, U.S. Government Printing Office, Washington, DC.
Calder III, W.A. and E. J. Braua 1983. Scaling of osmotic regulation in mammals and birds. Am. J.
Physiol. 244:601-606.
Cappon, C.J. and J.C. Smith. 1981. Mercury and selenium content and chemical form in fish muscle.
Arch. Environ. Contain. Toxicol. 10:305-319.
Cappon, C.J. 1987. Uptake and speciation of mercury and selenium in vegetable crops grown on
compost-treated soil. Water Air Soil Pollut 34:353-361.
Cernichiari, E., R. Brewer, G.J. Myers, D.O. Marsh, L.W. Lapham, C. Cox, C.F. Shamlaye, M. Berlin,
P.W. Davidson and T.W. Clarkson. 1995. Monitoring methylmercury during pregnancy: Maternal
hair predicts fetal brain exposure. NeuroToxicology 16:705-710.
June 1996 8-3 SAB REVIEW DRAFT
-------
Chai, C, W. Feng, Q. Qian, M. Guan, X. Li, Y. Lu and X. Zhang. 1994. Total and methyl mercury
levels in human scalp hairs of typical .populations in China by NAA, GC(EC), and other techniques.
Biological Trace Element Research pgs. 423-433.
Chapman, J.A. and G.A. Feldhamer, Ed. 1982. Wild Mammals of North America Biology,
Management, and Economics. The Johns Hopkins University Press.
Charbonneau, S.M., I. Munro and E. Nera. 1976. Chronic toxicity of methyl mercury in the adult cat.
Toxicology. 5:337-340.
Chi, J.G., E.G. Dooling and F.H. Gilles. 1977. Gyral development of the human brain. Ann. Neurol.
1:86-93.
Chu, P., B. Nott and*W. Chow. 1993. Results and Issues from the PISC?ES Field Tests, Second
International Conference on Managing Hazardous Air Pollutants, Washington, DC.
Clark, W., R.G. Rizeq, D.W. Hansell and W.R. Seeker. 1993. Mechanisms and Control of Toxic
Metals Emissions, Second International Conference on Managing Hazardous Air Pollutants,
Washington, DC.
Commission on Life Sciences See page 8-19. National Research Council, 1986.
Cordier, S., F. Deplan, L. Mandereau et al. 1991. Paternal exposure to mercury and spontaneous
abortions. Br. J Ind. Med. 48(6):375-381.
Coordinating Committee on Evaluation of Food Consumption Surveys
Cox, C., T.W. Clarkson, D.E. Marsh, L. Amin-Zaki, S. Tikriti and G.G. Myers. 1989. Dose-response
analysis of infants prenatally exposed to methyl mercury: An application of a single compartment
model to single-strand hair analysis. Environ. Res. 49(2):318-332.
Cramer, G.M. 1994. Exposure of U.S. consumers to methylmercury from fish. p. 103-118. In:
DOE/FDA/EPA Wrkshop on Methylmercury and Huamn Health, P.O. Moskowitz, L. Saroff, M.
Bolger, J. Cicmanec and J. Durkee, Ed. Conference Number 9403156. Published through: Biomedical
and Environmental Assessment Group, Brookhaven National Laboratory, Upton, New York.
Creason, J.P., T.A. Hinners, J.E. Bumgarner and C. Pinkerton. 1978a. Human Scalp Hair: An
Environmental exposure Index for Trace Elements. I. Fifteen Trace Elements in New York, N.Y.
(1971-72). EPA-600/l-78-037a. U.S. Environmental Protection Agency, Office of Research and
Development, Health Effects Research Laboratory, Research Triangle Park, N.C. 27711.
Creason, J.P. et al. 1978b. Seventeen Trace Elements in Four New Jersey Communities (1972).
EPA-600/l-78-037b.
» Creason, J.P. et al. 1978c. Seventeen Trace Elements in Birmingham, AL and Charlotte, NC (1972).
EPA-600/1-78-037C.
June 1996 8-4 SAB REVIEW DRAFT
-------
Crispin-Smith, J., M.D. Turner and D.O. Marsh. Project III. Hair methylmercury levels in women of
childbearing age.
Custer, T.W. and W.L. Hohman. 1994. Trace elements in canvasbacks (Aythya valisineria) wintering
in Louisiana, USA. 1987-1988. Environ. Pollut. 84:253-259.
DeRosis, C.T., J. Stara and P.Durkin. 1985. Ranking chemicals based on toxicity data. Toxicol. Ind.
Health. 1:177-199.
Dourson, M.L. and J. Stara. 1983. Regulatory history and experimental support of uncertainty
(safety) factors. Reg. Toxicol. Pharmacol. 3:224-238.
9 *
Dourson, M.L., L. Knauf and J. Swartout. 1992. On reference dose (RfD) and its underlying toxicity
data base. Toxicol. Incfe Health. 8(3): 171-189.
Druet, P., E. Druet, F. Potdevin et al. 1978. Immune type glomerulonephritis induced by HgCl2 in
the Brown-Norway rat. Ann. Immunol. 129C:777-792.
Eisler, R. 1987. Mercury hazards to fish, wildlife and invertebrates: A synoptic review. U.S.
Department of the Interior. Division of Wildlife and Contaminant Research, Fish and Wildlife
Service, Washington, DC.
%
Engstrom, D.R., E.B.Swain, T.A. Henning, M.E. Brigham and P.L. Brezonick. 1994. Atmospheric
mercury deposition to lakes and watersheds: a quantitative reconstruction from multiple sediment
cores. In: Environmental Chemistry of Lakes and Reservoirs, L.A. Baker, Ed. American Chemical
Society, p. 33-66.
Evans, R.D. 1986. Sources of mercury contamination in the sediments of small headwater lakes in
south-central Ontario, Canada. Arch. Environ. Contain. Toxicol. 15:505-512.
Evans, H.L. and P.J. Kostyniak. 1972. Effects of chronic methylmercury on behavior and tissue
mercury levels in the pigeon. Fed. Proc. 3(12):A561. (Abstract)
Faustman, E.M., B.C. Allen, RJ. Kavlock and C.A. Kimmel. 1994. Dose-response assessment for
developmental toxicity. I. Characterization of database and determination of no observed effect levels.
Fund. Appl. Toxicol. 23:478-486.
Fawer, R.F., U. DeRibaupierre, M.P. Guillemin, M. Berode and M. Lobe. 1983. Measurement of
hand tremor induced by industrial exposure to metallic mercury. J. Ind. Med. 40:204-208.
Felsvang, K., R. Gleiser, G. Juip and K.K. Nielsen. 1993. Air Toxics Control by Spray Dryer
Absorption Systems, Second International Conference on Managing Hazardous Air Pollutants,
Washington, DC.
Fimreite, N. 1970. Effects of methylmercury treated feed on the mortality and growth of leghorn
cockerels. Can. J. Anim. Sci. 50:387-389.
Fimreite, N. 1971. Effects of methylmercury on ring-necked pheasants. Canadian Wildlife Service
Occasional Paper Number 9. Department of the Environment. 39 p.
June 1996 8-5 SAB REVIEW DRAFT
-------
Fimreite, N. 1979. Accumulation and effects of mercury on birds. In: The Biogeochemistry of
Mercury in the Environment, J.O. Nriagu, Ed. Elsevier, Amsterdam, The Netherlands, p. 601-628.
Fimreite, N. and L.M. Reynolds. 1973. Mercury contamination in fish in northwestern Ontario. J.
Wildl. Mgmt. 37(l):62-68.
Findley, M.T. and R.C. Stendell. 1978. Survival and reproductive success of black ducks fed methyl
mercury. Environ. Pollut. 16:51-64.
Findley, M.T., W.H. Stickel and RE. Christensen. 1979. Mercury residues in tissues of dead and
surviving birds fed methylmercury. Bull. Environ. Contain. Toxicol. 21:105-110.
*
Florida Panther Interagency Committee. 1989. Mercury Contamination in Florida Panthers. Status
Report oT the Technical Subcommittee.
Foa, V., L. Caimi, L. Amante et al. 1976. Patterns of some lysosomal enzymes in the plasma and of
proteins in urine of workers exposed to inorganic mercury. Int. Arch. Occup. Environ. Health.
37:115-124.
Food and Nutrition Board See page 8-19. National Research Council, 1986.
Foley, R.E., S.J. Jackling, R.J. Sloan and M.K. Brown. 1988. Organochlorine and mercury residues
in wild mink and otter: Comparison with fish. Environ. Toxicol. Chem. 7:363-374.
Forzi, M., M.G. Cassitto, C. Bulgheroni et al. 1976. Psychological measures in workers
occupationally exposed to mercury vapours. A validation study. In: Adverse Effects of
Environmental Chemicals and Psychotoxic Drugs, Amsterdam, Oxford, NY, Elsevier Science
Publishers. Vol. 2, p. 165-172.
Forzi, M., M.G. Cassitto, C. Bulgheroni and V. Foa. 1978. Psychological measures in workers
occupationally exposed to mercury vapors: A validation study. In: Adverse Effects of Environmental
Chemicals and Psychotropic Drugs: Neurophysiological and Behayioral Tests, Vol. 2, J.J.
Zimmerman, Ed. Appleton-Century-Crofts, New York, NY. p. 165-175.
Fossi, C., S. Focardi, C. Leonzio et al. 1984. Trace-metals and chlorinated hydrocarbons in birds'
eggs from the delta of the Danube. Environ. Conserv. 11:345-350. (Cited in Zilllioux, 1993)
Froslie, A., G. Hold and G. Norheim. 1986. Mercury and persistent chlorinated hydrocarbons in owls
(Strigiformes) and birds of prey (Falconiformes) collected in Norway during the period 1965-1983.
Environ. Pollut. 11:91-108.
Girard, M. and C. Dumont. 1995. Exposure of James Bay Cree to methylmercury during pregnancy
for the years 1983-91. Water, Air and Soil Pollution 80: 13-19.
Glass, G.E., J.A. Sorensen, K.W. Schmidt, J.K. Huber and G.R. Rapp, Jr. 1993. Mercury sources and
distribution in Minnesota's aquatic resources: Precipitation, surface water, sediments, plants, plankton,
and fish. Final report to Minnesota Pollution Control Agency and Legislative Commission on
Minnesota Resources, 1989-1991 (Contract Nos. 831479 and WQ/PDS020).
June 1996 8-6 SAB REVIEW DRAFT
-------
Gotelli. C.A.. E. Astolfi. C. Cox, E. Cernichiari and T. Clarkson. 1985. Early biochemical effects of
an organic mercury funcigicide on infants: "Dose makes the poison". Science. 277:638-640.
Greenwood, M.R., T.W. Clarkson, R.A. Doherty et al. 1978. Blood clearance half-times in lactating
and nonlactating members of a population exposed to methyl mercury. Environ. Res. 16:48-54.
Greenwood, M.R., T.W. Clarkson, R.A. Doherty, A.H. Gates, L. Amin-Zaki, S. Elhassani and M.A.
Majeed. 1978. Blood clearance half-times in lactating and nonlactating members of a population
exposed to methyl mercury. Environ. Res. 16:48-54.
Gunderson, V.M., K.S. Grant, T.M. Burbacher et al. 1986. The effect of low-level prenatal methyl
mercury exposure on visual recognition memory in infant crab-eating macaques. Child Devel.
57:1076-1083.
Gutenmann, W.H., J.G. Ebel Jr., H.T. Kuntz, K.S. Yourstone and D.J. Lisk. 1992. Residues of p,p'-
DDE and mercury in lake trout as a function of age. Arch. Environ. Contain. Toxicol. 22:452-455.
Hanson, P.J., S.E. Lindberg, K.H. Kim, J.G. Owens and T.A. Tabberer. 1994. Air/surface exchange
of mercury vapor in the forest canopy: I. Laboratory studies of foliar Hg vapor exchange.
International Conference on Mercury as a Global Pollutant, July 10-14, Whistler, British Columbia,
Canada.
Harada, H. 1978. Congenital Minamata disease: Intrauterine methyl mercury poisoning. Teratology.
18:285-288.
Harada, M. 1995. Minamata Disease: Methylmercury poisoning in Japan caused by environmental
pollution. Crit. Rev. Toxicol. 25(l):l-24.
Harada, Y. 1968. Congenital (or fetal) Minamata Bay disease. In: Minimata disease, Kumamoto,
Study Group of Minamata Disease, Kumamoto University.
Harada, Y. 1977. Congenital Minimata Disease. In: Minimata Disease: Methylmercury Poisoning
in Minamata and Niigata, Japan, R. Tsuback and K. Irukayama, Ed. Tokyo, Kodansha. p. 209-239.
Harada, Y. 1977. Congenital Minimata Disease. In: Minimata Disease, T. Tsubaki and K.
Irukayama, Ed. Published by Kodansha, Ltd. Tokyo and Elsevier Scientific Publishing Company,
Amsterdam/London/New York. Table 3.27 on page 220.
Harper, R.G., D.S. Hopkins, and T.C. Dustan (1988). Nonfish prey of wintering bald eagles in
Illinois. Wilson Bull. 100:688-690.
Harvey, T., K.R. Mahaffey, S. Velazquez and M. Dourson. 1995. Holistic risk assessment: An
emerging process for environmental decisions. Reg. Toxicol. Pharmacol. (In press)
Hattis, D. and K. Silver. 1994. Human inter-individual variability—A major source of uncertainty in
assessing risks for noncancer health effects. Risk Anal. 14:421-432.
Hays, H. and R.W. Risebrough. 1972. Pollutant concentrations in abnormal young terns from Long
Island Sound. Auk. 89(1): 19-35.
June 1996 8-7 SAB REVIEW DRAFT
-------
Heinz, G.H. 1974. Effects of low dietary levels of methylmercury on mallard reproduction. Bull.
Environ. Contain. Toxicol. 11:386-392.
Heinz, G.H. 1975. Effects of methylmercury on approach and avoidance behavior of mallard
ducklings. Bull. Environ. Contain. Toxicol. 13:554-564.
Heinz, G.H. 1976a. Methylmercury: Second-year feeding effects on mallard reproduction and
duckling behavior. J. Wildl. Manage. 40:82-90.
Heinz, G.H. 1976b. Methylmercury: Second-generation reproductive and behavioral effects on
mallard ducks. J. Wildl. Manage. 40(4):710-715.
Heinz, G.H. 1979. Methylmercury: Reproductive and behavioral effects on three generations of
mallard ducks. J. Wildl. Mgmt. 43:394-401.
Hirano, M., K. Mitsumori, K. Malta et al. 1986. Further carcinogenicity study on methyl mercury
chloride in ICR mice. 'Jap. J Vet Sci. 48(1):127-135.
Hohman, A. and O.D. Creutzfeld. 1975. Squint and the development of binocularity in humans.
Nature 254:613-614.
Hovart, M., L. Liang, N.S. Bloom. 1993. Comparison of distillation with other current isolation
methods for the determinations of methylmercury compounds in low level environmental samples.
Part II. Water Chimica Acta. 282:153-168.
Hultman, P. and S. Enestrom. 1992. Dose-response studies in murine mercury-induced autoimmunity
and immune-complex disease. Toxicol. Appl. Pharmacol. 113(2):199-208.
Interpoll Laboratories. 1990a. Results of the May 1, 1990 Trace Metal Characterization Study on
Units 1 and 2 at the Sherburne County Generating Station. Conducted for Northern States Power
Company, Report #0-3033E.
Interpoll Laboratories. 1990b. Results of the March 1990 Trace Metal Characterization Study on Unit
3 at the Sherburne County Generating Station. Conducted for Northern States Power Company,
Report #0-3005.
Interpoll Laboratories. 1991. Results of the September 10 and 11, 1991 Mercury Removal Tests on
Units 1 and 2 and Unit 3 Scrubber Systems at the NSP Sherco Plant in Becker, MN. Conducted for
Northern States Power Company, Report #1-3409.
Interpoll Laboratories. 1992a. Results of the November 5, 1991 Air Toxic Emission Study on the No.
1, 3, and 4 Boilers at the NSP Black Dog Plant Conducted for Northern States Power Company,
Report #1-3451.
Interpoll Laboratories. 1992b. Results of the January 1991 Air Toxic Emission Study on the No. 2
Boiler at the NSP Black Dog Plant. Conducted for Northern States Power Company, Report #2-3496.
Interpoll Laboratories. 1992c. Results of the July 1992 Air Toxic Emission Study on Unit 8 at the
NSP Riverside Plant. Conducted for Northern States Power Company, Report #2-3590.
June 1996 8-8 SAB REVIEW DRAFT
-------
Interpoll Laboratories. 1992d. Results of the December 1991 Air Toxic Emission Study on Units 6
and 7 at the NSP Riverside Plant. Conducted for Northern States Power Company, Report #1-3468.
Jackson, T.A. 1991. Biological and environmental control of mercury accumulation by fish in lakes
and reservoirs of northern Manitoba, Canada. Can. J. Fish. Aquat. Sci. 48:2449-2470.
Jenkins, J.H. 1983, The status and management of the river otter (Lutra canadensis) in North
America. Acta. Zool. Fennica. 174:233-235.
Karvetti, R. and L. Knuts. 19'85. Validity of the 24-hour recall. J. Am. Dietet. Assoc.
85:1437-1442.
Khera, S. 1973. Reproductive capability of male rats and mice treated with methyl mercury.
Toxicol. Appl. Pharmacol. 24:167-^77.
Kinjo, Y., Y. Takizawa, Y. Shibata, M. Watanabe and H. Kato. 1995. Threshold dose for adults
exposed to methylmercury in Niigata Minamata Disease outbreak. Environ. Sci. 3(2):91-101.
Kishi, R., K. Hashimoto, S. Shimizu and M. Kobayashi. 1978. Behavioral changes and mercury
concentrations in tissues of rats exposed to mercury vapor. Toxicol. Appl. Pharmacol. 46(3):555-566.
Kjellstrom, T., P. Kennedy, S. Walk's and C. Mantell. 1986a. Physical and mental development of
children with prenatal exposure to mercury from fish. Stage 1: Preliminary test at age 4. National
Swedish Environmental Protection Board, Report 3080 (Solna, Sweden).
Kjellstrom, T., P. Kennedy, S. Wallis et al. 1986b. Physical and mental development of children with
prenatal exposure to mercury from fish. Stage 2: Interviews and psychological tests at age 6.
National Swedish Environmental Protection Board, Report 3642 (Solna, Sweden).
Klaassen, C.D., M.O. Amdur and J. Doully. 1986. Casarett and Doull's Toxicology: The Basic
Science of Poisons. MacMillan Publishing Company, New York, NY.
Koonin, L.M., J.C. Smith and M. Ramick. 1993. Division of Reproductive Health, National Center
for Chronic Disease Prevention and Health Promotion: Abortion Surveillance - United States, 1990:
Morbidity Mortality Weekly Report, Vol 42/No. SS-6, p. 29-57, December 17.
Kozie, K.D. and R.K. Anderson. 1991. Productivity, diet, and environmental contaminants in bald
eagles nesting near the Wisconsin Shoreline of Lake Superior. Arch. Environ. Contam. Toxicol.
20:41-48.
Kucera, E. 1983. Mink and otter as indicators of mercury in Manitoba waters. Canad. J Zool.
61:2250-2256.
Lange, T.R., H.E. Royals and L.L. Connor. 1993. Influence of water chemistry on mercury
concentration in large-mouth bass from Florida lakes. Trans. Am. Fish. Soc. 122:74-84.
Langlois, C. and R. Langis. 1995. Presence of airborne contaminants in the wildlife of northern
Quebec. Sci. Total Environ. 160/161:391-402.
June 1996 . 8-9 SAB REVIEW DRAFT
-------
Langolf, G.D., D.B. Chaffin, R. Henderson and H.P. Whittle. 1978. Evaluation of workers exposed
to elemental mercury using quantitative tests of tremor and neuromuscular functions. Am. Ind. Hyg.
Assoc. J. 39:976-984.
Langworth, S. 1987. Renal function in workers exposed to inorganic mercury. In: Occupational
Health in the Chemical Industry: Papers presented at the XXIIICOH Congress, Sydney, Australia,
September 27-October 2, 1987. Copenhagen, World Health Organization, p. 237.
Lasora, B.K. and R.J. Citterman. 1991. "Segmental analysis of mercury in hair in 80 women of
Nome, Alaska"'. OCS Study MMS 91-0065. PNL-7880, DE 92-003656. National Technical
Information Service. From U.S. Department of the Interior. Minerals Management Service, Alaska,
OCS Region.
Lauwerys, R., A. Bernard, H. Roels, J.P. Buchet, J.P. Gennart, P. Mahieu and J.M. Foidard.
1983. Anti-laminin antibodies in workers exposed to mercury vapour. Toxicol. Lett. 17:113-116.
Lauwerys, R. H. Roels, P. Genet, G. Toussaint, A. Bouckaert and S. De Cooman. 1985. Fertility of
male workers exposed to mercury vapor or to manganese dust: A questionnaire study. Am. J. Ind.
Med. 7(2): 171-176.
Lee, I.P. and R.L. Dixon. 1975. Effects of mercury on spermatogenesis studies by velocity
sedimentation cell separation and serial mating. J Pharmacol. Exp. Ther. 194:171-181.
Levin, M., J. Jacobs and P.G. Polos. 1988. Acute mercury poisoning and mercurial pneumonitis from
gold ore purification. Chest. 94(3):554-558.
Levine, S.P., G.D. Cavender, G.D. Langolf and J.W. Albers. 1982. Elemental mercury exposure:
Peripheral neurotoxicity. Br. J. Ind. Med. 39:136-139.
Life Sciences Research Office See page 8-19. Life Sciences Research Office, 1986.
•
Lilis, R., A. Miller and Y. Lerman. 1985. Acute mercury poisoning with severe chronic pulmonary
manifestations. Chest. 88(2):306-309.
Lindberg, P., T. Odsjo and L. Reuterardh. 1985. Residue levels of polychlorobiphenyls, DDT, and
mercury in bird species commonly preyed upon by the peregrine falcon (Falco peregrinus Tumi) in
Sweden. Arch. Environ. Contain. Toxicol. 14:203-212.
Lindquist, O., K. Johansson, M. Aastrup et al. 1991. Mercury in the Swedish environment. Recent
research on causes, consequences and corrective methods. Water Air Soil Pollut. 55:1-261.
Linscombe, G.N., N. Kinler and R.J. Aulerich. 1982. Mink. In: Wild Mammals of North America,
J.A. Chapman and G.E. Feldhamer, Ed. Johns Hopkins University Press, Baltimore, MD. p. 629-643.
Lonzarich, D.G., T.E. Harvey and J.E. Takekawa. 1992. Trace element and organochlorine
concentrations in California Clapper Rail (Rallus longirostric obsoletus) eggs. Arch. Environ. Contam.
Toxicol. 23:147-153.
June 1996 8-10 SAB REVIEW DRAFT
-------
Lowe, T.P., T.W. May, W.G. Brumbaught and D.A. Kane. 1985. National Contaminant
Biomonitoring Program: Concentrations of seven elements in fresh-water fish, 1978-1981. Arch,
Environ. Contam. Toxicol. 14:363-388.
MacCrimmon, H.R., C.D. Wren and B.L. Gots. 1983. Mercury uptake by lake trout, Salveiinus
namaycush, relative to age, growth and diet in Tadenac Lake with comparative data from other
Precambrian Sheild lakes. Can. J. Fisher. Aq. Sci. .40:114-120.
Marsh, D.O., G.J. Myers, T.W. Clarkson et al. 1981. Dose-response relationship for human fetal
exposure to methyl mercury. Clin. Toxicol. 10:1311-1318.
Marsh, D.O., T.M. Clarkson, C. Cox, G.J. Myers, L. AminZaki and S. Al-Tikriti. 1987. Fetal
methylmercury poisoning: Relationship between concentration in single strands of maternal hair and
child effects. Arch. Neurol. 44:1017-1022.
Mathers, R.A. and P.M. Johansen. 1985. The effects of feeding ecology on mercury accumulation in
walleye (Stizostedion vitreum) and pike (Esox lucius) in Lake Simcoe. Can. J. Zool. 63:2006-2012.
McFarland, R.B. and H. Reigel. 1978. Chronic mercury poisoning from a single brief exposure. J.
Occup. Med. 20(8):532-534.
»
McKeown-Eyssen, G.E., J. Ruedy and A, Neims. 1983. Methyl mercury exposure in northern
Quebec: II. Neurologic findings in children. Am. J Epidemiol. 118:470-479.
*
McKim, J.M., G.F. Olson, C.W. Holecombe and E.O: Hunt. 1976. Long term effects of
methylmercuric chloride on three generations of brook trout (Salveiinus fontinalis): Toxicity,
accumulation, distribution and elimination. J. Fish. Res. Bd. Can. 33:2726-2739.
Miller, J.M., D.B. Chaffin and R.G. Smith. 1975. Subclinical psychomotor and neuromuscular
changes in workers exposed to inorganic mercury. Am. Ind. Hyg. Assoc. J. 36:725-733.
Mishonova, V.N., P.A. Stepanova and V.V. Zarudin. 1980. Characteristics of the course of pregnancy
and births in women with occupational contact with small concentrations of metallic mercury vapors in
industrial facilities. Gig truda i prof zabolel. 24:21-23.
Mitsumori, K., K. Maita, T. Saito et al. 1981. Carcinogenicity of methyl mercury chloride in ICR
mice: Preliminary note on renal carcinogenesis. Cancer Lett. 12:305-310.
Mitsumori, K., M. Hirano, H. Ueda et al. 1990. Chronic toxicity and carcinogenicity of methyl
mercury chloride in B6C3F1 mice. Fund. Appl. Toxicol. 14:179-190.
Mohamed, M., T. Burbacher and N. Mottet. 1987. Effects of methyl mercury on testicular functions
in Macaca fascicularis monkeys. Pharmacol. Toxicol. 60(l):29-36.
June 1996 8-11 SAB REVIEW DRAFT
-------
Mosbaek, H., J.C. Tjell and T. Sevel. 1988. Plant uptake of airborne mercury in background areas.
Chemosphere. 17:1227-1236.
Munro, I.C., E.A. Nera, S.M. Charbonneau et al. 1980. Chronic toxicity of methyl mercury in the rat.
J. Environ. Pathol. Toxicol. 3:437-447.
Mushak, P. and A.M. Crocetti. 1990. The Nature and Extent of Lead Poisoning in Children in the
United States: A Report to Congress. Agency for Toxic Substances and Disease Registry, United
States Public Health Service, United States Department of Health and Human Services.
Mutti, A., S. Lucertini, M. Fornari et al. 1985. Urinary excretion of-a brush-border antigen revealed
by monoclonal antibodies in subjects occupationally exposed to heavy metals. Heavy Met. Environ.
International Conference 5th. Vol. 1, p. 565-567.
Nagy, K.A. 1987. Field metabolic rate and food requirements scaling in mammals and birds. Ecol.
Monogr. 57(2): 111-128.
NAS (National Academy of Sciences/National Research Council). 1983. Risk Assessment in the
Federal Government: Managing the Process. National Academy Press, Washington, DC.
NAS (National Academy of Sciences/National Research Council). 1994a. Risk Assessment in the
Federal Government: Managing the Process. National Academy Press, Washington, DC.
NAS (National Academy of Sciences/National Research Council). 1994b. Science and Judgment in
Risk Assessment. National Academy Press, Washington, DC.
National Academy Press, Washington DC 1986.
National Center for Health Statistics of the United States. 1990a. Volume I. Natality: Table 1-60;
p. 134-140.
National Center for Health Statistics of the United States. 1990b. Volume II. Mortality; Table-10,
p. 16, 18 and 20.
National Research Council See page 8-19. National Research Council, 1986.
NIEHS (National Institute of Environmental Health Sciences). 1993. Report to Congress on
Methylmercury. NIEHS. Research Triangle Park, NC, USA.
Noblett, Jr., J.G., F.B. Meserole, D.M. Seeger and D.R. Owens. 1993. Control of Air Toxics from
Coal-fired Power Plants Using FGD Technology, Second International Conference on Managing
Hazardous Air Pollutants, Washington, DC.
Nobmann, E.D., T. Byers, A.P. Lanier, J.H. Hankin and M.Y. Jackson. 1992. The diet of Alaska
Native adults: 1987-1988. Am. J. Clin. Nutr. 55:1024-1032.
(i
NRC/NAS (National Research Council/National Academy of Sciences). 1991. (Committee on
Evaluation of the Safety of Fishery Products). Seafood Safety, F.E. Ahmed, Ed. National Academy
Press, Washington, DC.
June 1996 8-12 SAB REVIEW DRAFT
-------
NRC/NAS. 1994. Science and Judgment in Risk Assessment. National Academy Press, Washington.
DC.
NTP National Toxicology Program). 1993. Toxicology and carcinogenesis studies of mercuric
chloride in F344 rats and B6C3F1 mice. U.S. Department ot Health and Human Services, Research
Triangle Park, NC.
O'Connor, D.J. and S.W. Nielsen. 1980. Environmental survey of methylmercury levels in wild mink
and otter from the northwestern United States, and experimental pathology of methylmercurialism in
the otter. In: Worldwide Furbearer Conference Proceedings, LA. Chapman and D. Pursley, Ed.,
Frostburg, MD, 2-11 August 1980. Worldwide Furbearer Conference, Frombert MD, p. 1726-1745.
O'Connor, D.J. and S.W. Nielsen. 1981. Environmental survey of methylmercury levels in wild mink
(Mustela visori) and otter (Lutra canadensis) from the northeastern United States and experimental
pathology of methylmercurialism in the otter. Worldwide Furbearer Conference Proceedings,
p. 1728-1745.
Odsjo, T. 1982. Eggshell thinning and levels of DDT, PCB and mercury in the eggs of osprey
(Pardion haliaetus L.) and marsh harrier (Circus aeruginosus L.) in relation to their breeding success
and population status in Sweden. Ph.D. Dissertation University of Stockhol'm, Sweden.
Oskarrson, A., B. Ohlin, E.M. Ohlander and L. Albanus. 1990. Mercury levels in hair from people
eating large quantities of Swedish freshwater fish. Food Additives and Contaminants 7:555-562.
Peterson, D.E., M.S. Kanarek, M.A. Keykendall, M. Diedrich, H.A. Anderson, P.L. Remington and
T.B. Sheffy. 1994. Fish consumption patterns and blood mercury levels in Wisconsin Chippewa
Indians. Arch. Environ. Health. 49(ll):53-58.
Petit, D. (Office of Migratory Bird Management, U.S. Fish and Wildlife Service, U.S. Department of
the Interior). 1995. Personal Communication to Glenn Rice, U.S. Environmental Protection Agency.
Request for Population Estimates for Selected Avian and Mammalian Species. June 7. Memorandum.
Piikivi, L. 1989. Cardiovascular reflexes and low long-term exposure to mercury vapor. Int. Arch.
Occup. Environ. Health. 61:391-395.
Piikivi, L. and U. Tolonen. 1989. EEG findings in chlor-alkali workers subjected to low long term
exposure to mercury vapor. Br. J. Ind. Med. 46:370-375.
Piikivi, L. and H. Hanninen. 1989. Subjective symptoms and psychological performance of chlorine-
alkali workers. Scand. J. Work Environ. Health. 15:69-74.
Piikivi, L. and A. Ruokonen. 1989. Renal function and long-term low mercury vapor exposure.
Arch. Environ. Health. 44(3): 146-149.
Prichard, A.L. P.A. McAnulty, M.J. Collier and J.M. Tesh. 1982b. The effects of inorganic mercury
on fertility and survival in utero in the rat. Teratology. 26(3):20A.
Putnam, J.J. (1991). Food Consumption, 1970-1990. Food Review 14(3):2-12. July-September.
June 1996 8-13 SAB REVIEW DRAFT
-------
Radian Corporation. 1993a. Preliminary Draft Emissions Report for EPRI Site 102, Field Chemical
Emissions Monitoring Project. Prepared for Electric Power Research Institute, February 1993.
Radian Corporation. 1993b. Preliminary Draft Emissions Report for EPRI Site 21, Field Chemical
Emissions Monitoring Project. Prepared for Electric Power Research Institute, May 1993.
Rentes, P. and E. Seligman. 1968. Relationship between environmental exposure to mercury and
clinical observation. Arch. Environ. Health. 16:794-800.
RfD Work Group Notes of 13 October 1994.
Rice, D.C. 1989a. Delayed neurotoxicity in monkeys exposed developmentally to methyl mercury.
Neurotoxicology. 10(4): 645-650.
Rice, D.C. 1989b. Brain and tissue levels of mercury after chronic methyl mercury exposure in the
monkey. J. Toxicol. Enviroa Health. 27(2): 189-198.
Rodier, P.M. 1994. Vulnerable periods and processes during central nervous system development.
Environ. Health Perspect 102(Suppl 2): 121-124.
Roelke, M.H., D.P. Schultz, C.F. Facemire, S.F. Sundlof and H.E. Royals. 1991. Mercury
contamination in Florida panthers. A report of the Florida Panther Technical Subcommittee to the
Florida Panther Interagency Committee.
Roels, H., R. Lauwerys, J.P. Buchet et al. 1982. Comparison of renal function and psychomotor
performance in workers exposed to elemental mercury. Int Arch. Occup. Environ. Health. 50:77-93.
Roels, H., R-P. Gennart, R.L. Lauwreys et al. 1985. Surveillance of workers exposed to mercury
vapor: Validation of a previously proposed biological threshold limit value for mercury concentration
in urine. Am. J. Ind. Med. 7:45-71.
Roels, H., S. Abdeladim, E. Ceulemans and R. Lauwreys. 1987. Relationships between the
concentrations of mercury in air and in blood or urine in workers exposed to mercury vapour. Ann.
Occup. Hyg. 31:135-145.
Roels, H., S. Abdeladim, M. Braun, J. Malchaire and R. Lauwerys. 1989. Detection of hand tremor
in workers exposed to mercury vapor: A comparative study of three methods. Environ. Res.
49:152-165.
Rosenman, K.D., J.A. Valciukas, L. Glickman, B.R. Meyers and A. Cinotti. 1986. Sensitive
indicators of inorganic mercury toxicity. Arch. Environ. Health. 41:208-215.
Rustam, H. and T. Hamdi. Methyl mercury poisoning in Iraq. A neurological study. Brain.
97:499-510.
Scheuhammer, A.N. 1991. Effects of acidification on the availability of toxic heavy metals and
calcium to wild birds and mammals. Environ. Pollut. 71:329-376.
June 1996 8-14 SAB REVIEW DRAFT
-------
Scott, D.P. and F.A.J. Armstrong. 1972. Mercury concentration in relation to size in several species
of freshwater fishes from Manitoba and northwestern Ontario. J. Fish. Res. Bd. Can. 29:1685-1690.
Scott, M.L. 1977. Effects of PCBs, DDT, and mercury compounds in chickens and Japanese quail.
Fed. PrOc. 36:1888-1893.
Sheffy, T.B. and J.R. St. Amant. 1982. Mercury burdens in furbearers in Wisconsin. J. Wildl.
Manage. 46:1117-1120.
Sikorski, R., T. Juszkiewicz, T. Paszkowski, et al. 1987. Women in dental surgeries: Reproductive
hazards in occupational exposure to metallic mercury. Int Arch Occup Environ Health 59:551-557.
Singer, R., J.A. Valciukas and K.D. Rosenman. 1987. Peripheral neurotoxicity in workers exposed to
inorganic mercury compounds. Arch. Environ. Health. 42(4):181-184.
Skurdal, J., T. Quenild, and O.K. Skogheim. 1985. Aquatic bacterial populations and heavy metals.
I. Composition of aquatic bacteria in the presence of copper and mercury sits. Water Res. 11:639-
642.
Sloss, L.L., 1993. Emissions and Effects of Air Toxic from Coal Combustion: An Overview, Second
International Conference on Managing Hazardous Air Pollutants, Washington, DC.
Smith, R.G., A.J. Vorwald, L.S. Patil and T.F. Mooney, Jr. 1970. Effects of exposure to mercury in
the manufacture of chlorine. Am. Ind. Hyg. Assoc. J. 31:687-700.
Sorenson, J.A., G.E. Glass, KW. Schmidt et al. 1990. Airborne mercury deposition and watershed
characteristics in relation to mercury concentrations in water, sediments, plankton, and fish of eighty
northern Minnesota lakes. Environ. Sci. Technol. 24:1716-1727.
Spalding,.M.G., R.D. Bjork, G.V.N. Powell and S.F. Sundlof. 1994. Mercury and cause of death in
great white herons. J. Wildl. Mgmt. 58(4):735-739.
Steffek, A.J., R. Clayton, C. Siew and A.C. Verrusio. 1987. Effects of elemental mercury vapor
exposure on pregnant Sprague-Dawley rats (abstract only). Teratology. 35:59A.
Stem, A.H. 1993. Reevaluation of the reference dose for methylmercury and assessment of current
exposure levels. Risk Anal. 13(3):355-364.
Stewart, W.K., H.A. Guirgis, J. Sanderson and W. Taylor. 1977. Urinary mercury excretion and
proteinuria in pathology laboratory staff. Br. J. Ind. Med. 34:26-31.
Stonard, M.D., B.V. Chater, D.P. Duffield, A.L. Nevitt, J.J. O'Sullivan and G.T. Steel. 1983. An
evaluation of renal function in workers occupationally exposed to mercury vapor. Arch. Occup.
Environ. Health. 52:177-189.
Subcommittee on Criteria for Dietary Evaluation See page 8-19. National Research Council, 1986.
Suchanek, T.H., P.J. Richerson, L.A. Woodward, D.G. Slotten, J.L. Holts and C.E.E. Woodmansee.
1993. Preliminary lake study report. A survey and evaluation of mercury in sediment water,
June 1996 8-15 SAB REVIEW DRAFT
-------
plankton, periphyton, benthic invertebrates and fishes within the aquatic ecosystem of Clear Lake
California. Institute of Ecology, U.C. Davis.
Sundlof, S.F., M.G. Spalding, J.D. Wentworth and C.K. Steible. 1994. Mercury in livers of wading
birds (Ciconiformes) in Southern Florida. Arch. Environ. Contamin. Toxicol. 27:299-305.
Suter, K.E. 1975. Studies on the dominant lethal and fertility effects of the heavy metal compounds
methyl mercuric hydroxide, mercuric chloride, and cadmium chloride in male and female mice.
Mutat. Res. 30:365-374.
Swain, E.B., D.A. Engstrom, M.E. Brigham, T.A. Henning and P.L. Brezonik. 1992. Increasing rates
of atmospheric mercury deposition in midcontinental North America. Science. 257:784-787.
Swedish EPA. 1991. Mercury in the Environment: Problems and Remedial Measures in Sweden.
ISBN 91-620-1105-7.
Tamashiro, H., M. Arakaki, H. AkagL M. Futastsuka and L.H. RohL 1985. Mortality and survival for
Minamata Disease. Int. J Epidemiol. 14(4):582-588.
Thompson, D.R., R.W. Furness and R.T. Barrett. 1992. Mercury concentrations in seabirds from
colonies in the Northeast Atlantic. Arch. Environ. Contain. Toxicol. 23:383-389.
Toweill, D.E. and J.E. Tabor. 1982. River otter. In: Wild Mammals of North America, J.A.
Chapman and G.A. Feldhamer, Ed. Johns Hopkins University Press, Baltimore, MD. p. 688-703.
Tsubaki, T. and K. Irukayama. 1977. Minamata Disease. Methylmercury poisoning in Minamata and
Niigata, Japan. Kodansha, Ltd and Elsevier Scientifi Publishing Company, Amsterdam.
U.S. EPA. 1984. Risk Assessment and Management: Framework for Decision Making. Office of
Policy, Planning, and Evaluation, Washington, D.C. EPA/600/9-85/002.
U.S. EPA. 1986a. Guidelines for Carcinogen Risk Assessment. Federal Register. 51:33992-34005.
(September 24)
U.S. EPA. 1986b. Guidelines for Mutagenicity Risk Assessment. Federal Register. 51:34006-34012.
(September 24)
U.S. EPA. 1987a. The Risk Assesment Guidelines of 1986. Office of Health and Environmental
Assessment, Washington DC 20460. EPA/600/8-87/045.
U.S. EPA. 1987b. Peer Review Workshop on Mercury Issues. October 26-27, 1987, Summary
Report. Environmental Criteria and Assessment Office, Cincinnati, OH.
U.S. EPA. 1991. Guidelines for Developmental Toxicity Risk Assessment. Federal Register.
56:63798-63826. (December 5)
U.S. EPA. 1992. Framework for Ecological Risk Assessment Risk Assessment Forum, Washington,
DC. EPA/630/R-92-001.
June 1996 8-16 SAB REVIEW DRAFT
-------
U.S. EPA. 1993a. Ecological Impacts of Some Heavy Metals Related to Long-Range Atmospheric
Transport. April 1993 report by the Secretariat prepared with the assistance of H. Andreae (consultant,
Germany).
U.S. EPA. 1993b. Great Lakes Water Quality Initiative Criteria Documents for the Protection of
Wildlife (PROPOSED). DDT; Mercury; 2,3,7,8-TCDD; PCBS. Office of Water, Office of Science
and Technology, Washington DC. EPA-822-R-93-007. April.
U.S. EPA. 1994. Integrated Risk Information System (IRIS). Online. Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati, OH.
U.S. EPA. 1995. Policy for Risk Characterization at the U.S. Environmental Protection Agency.
Memorandum from Carol M. Browner, March 21.
&
U.S. EPA. 1995a. - Trophic level and exposure analysis for selected piscivorous birds and mammals.
Volume I. Analysis for species of the Great Lakes Basin (Draft). U.S. EPA Office of Science and
Technology, Washington, DC.
U.S. EPA. 1995b. Water quality guidance for the Great Lakes system and correction. Proposed
rules. Federal Register.
U.S. Fish and Wildlife Service. 1994. Biologue Series. Odsjo. (Cited in Lindqvist, 1991)
Verbeck, M.M., H.J.A. Salle and C. Kemper. 1986. Tremor in workers with low exposure to metallic
mercury. Hyg. Assoc. J. 47(8):559-562.
Watanabe, T., T. Shimada and A. Endo. 1982. Effects of mercury compounds on ovulation and
meiotic and mitotic chromosomes in female golden hamsters. Teratology. 25:381-384.
Weil, C.S. and McCollister. 1963. Relationship between the short- and long-term feeding studies in
designing an effective toxicity test. Agric. Food Chem. 11:486-491.
Wheatley, B. and S. Paradis. 1995. Exposure of Canadian aboriginal peoples to methylmercury.
Water, Air, and Soil Pollution 80: 3-11.
WHO (World Health Organization). 1990. Methyl mercury. Vol. 101. Geneva, Switzerland: World
Health Organization, International Programme on Chemical Safety.
Willett, W. 1990. Nature of variation in diet In: Nutrition Epidemiology, W. Willett, Ed.
Monographs in Epidemiology and Biostatistics, Vol. 15. Oxford University Press, New York/Oxford.
p. 34-51.
Wobeser, G.A. 1973. Ph.D. Dissertation. Aquatic Mercury Pollution: Studies of its occurrence and
pathologic effect on fish and mink. University of Saskatchewan (Canada). Dissertation Number 73-
24, 819. University Microfilms, Ann Arbor, MI.
Wobeser, G., N.D. Nielsen and B. Schiefer. 1976a. Mercury and mink I: The use of mercury
contaminated fish as a food for ranch mink. Can. J. Comp. Med. 40:30-33.
June 1996 8-17 SAB REVIEW DRAFT
-------
Wobeser, G., N.D. Nielsen and B. Schiefer. 1976a. Great Lakes Water Quality Initiative Criteria
Documents for the Protection of Wildlife (PROPOSED). DDT; Mercury; 2,3,7,8-TCDD; RGBs.
Office of Water, Office of Science and Technology, Washington DC. EPA-822-R-93-007. April.
Wobeser, G., N.D. Nielsen and B. Schiefer. 1976b. Mercury and mink II: Experimental
methylmercury intoxication. Can. J. Comp. Med. 40:34-45.
Wolfe, R.J. and R.J. Walker. 1987. Subsistence economies in Alaska: Productivity, geography and
development impacts. Arctic Anthropol. 24:56-81.
Wren, C.D., H.R. MacCrimmon and B.R. Loescher. 1983. Examination of bioaccumulation and
biomagnification of metals in a Precambrian shield lake. Water Air Soil Pollut. 19:277-291.
Wren, C.D. 1985. Probable case of mercury poisoning in a wild otter in northwestern Ontario. Can.
Field-Nat. 99:112-114.
Wren, C.D. and H.R. MacCrimmon. 1986. Comparative bioaccumulation of mercury in two adjacent
freshwater ecosystems. Water Res. 20:763-769.
Wren, C.D., P.M. Stokes and K.L. Fischer. 1986. Mercury levels in Ontario mink and otter relative
to food levels and environmental acidification. Canad. J Zool. 64:2854-2859.
Zillioux, E.J., D.B. Porcella and J.M. Benoit. 1993. Mercury cycling and effects in freshwater
wetland ecosystems. Environ. Toxicol. Chem. 12:2245-2264.
June 1996 8-18 SAB REVIEW DRAFT
-------
ADDITIONAL REFERENCES
Cox, C., D. Marsh, G. Meyers, and T. Clarkson. 1995. Analysis of data on delayed development from
the 1971-72 outbreak of methylmercury poisoning in Iraq: assessment of influential points.
NeuroToxicology 16:727-730.
Crump, K., J. Viren, A. Silvers, H. Clewell, J. Gearhart, and A. Shipp. 1995. Reanalysis of dose-response
data from the Iraqi methylmercury poisoning episode. Risk Analysis 15:523-532.
Gearhart, J., H. Clewell, K. Crump, A. Shipp, and A. Silvers. 1995. Pharmacokinetic dose estimates of
mercury in children and dose response curves of performance tests in a large epidemiological study.
Water,Air, and Soil Pollution 80:49-58.
\
Life Sciences Research Office. 1986. Guidelines for Use of Dietary Intake Data. Editor: Anderson, S.
Contract Number FDA 223-84-2059. Federations for American Societies of Experimental Biology,
Publishers. Bethesda, Maryland.
Marsh D., M. Turner, J. Crispin Smith, P. Allen, and N. Richdale. 1995. Fetal methylmercury study in
a Peruvian fish-eating population. NeuroToxicology 16:717-726.
Myers G., P. Davidson, C. Cox, C. Shamlaye, M. Tanner, D. Marsh, ET. Cernichiari, L. Lapham, M.
Berlin, and T. Clarkson. 1995. Summary of the Seychelles child development study on the relationship
of fetal methylmercury exposure to neurodevelopment. NeuroToxicology 16:711-716.
National Research Council. Subcommittee on Criteria for Dietary Evaluation. 1986. Nutrient adequacy:
Assessment using food consumption surveys. National Academy Press, Washington, DC.
U.S. Environmental Protection Agency
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12th Floor
Chicago, IL 60604-3590
June 1996 8-19 SAB REVIEW DRAFT
------- |