ornl
OAK RIDGE
NATIONAL
LABORATORY
HTI*I MM fri
ORNL-6051
Report on the Workshop on Food-Chain
Modeling for Risk Analysis
J. E. Breck
C. F. Baes I
ENVIRONMENTAL SCIENCES DIVISION
Publication No. 2340
OPERATED BY
MARTIN MARIETTA ENERGY SYSTEMS, INC.
FOR THE UNITED STATES
DEPARTMENT OF ENERGY
-------
Printed in the United States of America. Available from
National Technical Information Service
U.S. Department of Commerce
5285 Port Royal Road, Springfield, Virginia 22161
NTIS price codes—Printed Copy: A05; Microfiche A01
This report was prepared as an account of work sponsored by an agency of the
United States Government. Neither theUnitedStatesGovernment nor any agency
thereof, nor any of their employees, makes any warranty, express or implied, or
assumes any legal liability or responsibility for the accuracy, completeness, or
usefulness of any information, apparatus, product, or process disclosed, or
represents that its use would not infringe privately owned rights. Reference herein
to any specific commercial product, process, or service by trade name, trademark,
manufacturer, or otherwise, does not necessarily constitute or imply its
endorsement, recommendation, or favoring by the United States Government or
any agency thereof. The views and opinions of authors expressed herein do not
necessarily state or reflect those of the United States Government or any agency
thereof.
-------
ORNL-6051
Report on the
Workshop on Food-Chain Modeling
for Risk Analysis
J. E. Breck
C. F. Baes III
Environmental Sciences Division
Publication No. 2340
EPA Project Officer: A. A. Moghissi
Date of Issue - February 1985
Prepared for
Office of Research and Development
U.S. Environmental Protection Agency
Washington, D.C. 20460
EPA Interagency Agreement No. DW 89930292-01-0
(DOE 40-740-78)
Prepared by the
OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee 37831
operated by
MARTIN MARIETTA ENERGY SYSTEMS, INC.
for the
U.S. DEPARTMENT OF ENERGY
under Contract No. DE-AC05-840R21400
-------
DISCLAIMER
Although the research described in this report has been funded by
the United States Environmental Protection Agency through Interagency
Agreement Number DW 89930292-01-0 with the Oak Ridge National
Laboratory, it has not been subjected to the EPA's required peer and
policy review and, therefore, does not necessarily reflect the views of
the agency, and no official endorsement should be inferred.
n
-------
CONTENTS
Page
DISCLAIMER ii
LIST OF FIGURES v
LIST OF TABLES vii
ABSTRACT ix
ACKNOWLEDGEMENTS xi
1. INTRODUCTION 1
1.1 OBJECTIVES 1
1.2 ACTIVITIES 3
1.3 PRESENTATIONS 8
2. AQUATIC FOOD CHAIN MODELS 13
2.1 MODELS BEST SUITED TO SYNFUELS RISK ANALYSIS 14
2.1.1 Concentration Factor Model 14
2.1.2 Dynamic, Bioenergetics-based Models 16
2.1.3 Comments For All Models 21
2.2 DATA SOURCES AND PARAMETER ESTIMATION METHODS 26
2.3 MAJOR LIMITATIONS ON EXISTING DATA AND METHODS 30
2.4 CONCLUSIONS 34
3. TERRESTRIAL FOOD CHAIN MODELS 36
3.1 MODELS BEST SUITED TO SYNFUELS RISK ANALYSIS 36
3.2 DATA SOURCES AND PARAMETER ESTIMATION METHODS 42
3.3 MAJOR LIMITATIONS ON EXISTING DATA AND METHODS 47
4. REFERENCES 55
APPENDIX A. AGENDA 61
APPENDIX B. LIST OF PARTICIPANTS 64
ill
-------
LIST OF FIGURES
Fig. 1. Components of the overall human health risk assessment
methodology for synfuels technologies 2
Fig. 2. Bioaccumulation factor vs octanol-water partition
coefficient for different trophic levels in a linear
food chain 20
Fig. 3. Pathways addressed in the terrestrial food chain
computer code TERREX 37
-------
LIST OF TABLES
Page
Table 1. Risk analysis units (RACs) 4
Table 2. Variable names and descriptions of parameters in the
SITt* data base 38
Table 3. Transport parameters for synfuels food-chain exposure
assessment 43
Vll
-------
ABSTRACT
BRECK, J. E., and C. F. BAES III. 1985. Report on the
workshop on food-chain modeling for risk analysis.
ORNL-6051. Oak Ridge National Laboratory, Oak Ridge,
Tennessee. 82 pp.
The Workshop on Food-Chain Modeling for Risk Analysis was held in
Washington, U.C., on March 22-24, 1983. The workshop was sponsored by
the U.S. Environmental Protection Agency, Office of Research and
Development (EPA/ORD), and supported under the Integrated Health and
Environmental Risk Analysis Program (IHERAP) for Synfuels. Atmospheric
and aquatic dispersion models, aquatic and terrestrial food-chain
transport models, and models that estimate risks from calculated
environmental exposures to synfuels effluents (dose-response models)
are used to assess health risk from sunfuels effluents. The workshop
focused on the aquatic and terrestrial food-chain models currently
being used in the risk assessment process.
The workshop was attended by both modelers and experimentalists -
all with some experience in risk assessment. During the first day,
each participant presented a summary of recent work or a review of work
of interest. After the presentations, topics of discussion included
(1) models best suited for synfuels risk assessment, (2) best data
sources and methods of estimating model input parameters, and
(3) limitations inherent in the models used for risk analysis.
However, discussions touched on all areas of food-chain modeling for
risk assessment. Discussions on the complementary roles of the modeler
and the experimentalist, estimation of model uncertainties, and
IX
-------
integration of food-chain models with other aspects of the assessment
approach were particularly relevant. On the second day, discussions
continued and participants logged the discussions and conclusions in
which they took part. On the last day, a collective exposition was
presented, discussed, and edited. This report details the
presentations made by the workshop participants, the collective
exposition on topics discussed, and the conclusions reached.
The workshop concluded that in aquatic food-chain modeling of
chronic low-level releases of synfuels effluents, a simple
concentration factor approach is appropriate. For terrestrial
fooa-chain models the need for greater model complexity was recognized
to account for location-specific variations in agricultural practice,
although the estimation of terrestrial transport is also achieved
through the use of concentration factors. For both aquatic and
terrestrial models, using field data is the best method for estimating
concentration factors, but where such data do not exist, laboratory
data can be used. If no data exist for a particular compound or class
of compounds, estimates can be made using partition coefficients based
on structure-activity relationships. Finally, the workshop recognized
the need to estimate the uncertainty associated with model output.
-------
ACKNOWLEDGMENTS
Dr. Lawrence W. Barnthouse, Charles F. Baes III, and
Dr. James E. Breck, all of the Environmental Sciences Division of
Oak Ridge National Laboratory (ORNL), planned and organized the
technical program. Barnthouse was to be the Workshop Moderator, but
because he was unable to attend, this duty was assumed by Mr. Baes and
Dr. Breck.
Special thanks go to Dr. Melvin W. Carter of the Georgia Institute
of Technology for making the meeting arrangements and coordinating
travel plans and to Barbara Burns for typing assistance during the
workshop. The workshop organizers also wish to thank the participants
(listed in Appendix B) for their hard work at the conference.
This workshop was sponsored and supported by the Integrated Health
and Environmental Risk Analysis Program for Synfuels of EPA's Office of
Research and Development under Interagency Agreement No. DOE 40-740-78
(EPA No. DW 89930292-01-1) with the U.S. Department of Energy under
Contract No. DE-AC05-840R21400 with Martin Marietta Energy Systems,
Inc. The Project Officer is Dr. A. Alan Moghissi.
XI
-------
1. INTRODUCTION
1.1 OBJECTIVES
The Workshop on Food-Chain Modeling for Risk Analysis was held in
Washington, D.C., March 22-24, 1983, and was sponsored and supported by
the Integrated Health and Environmental Risk Analysis Program for
Synfuels, U.S. Environmental Protection Agency Office of Research and
Development (EPA/ORD). This particular workshop focused on
applications in the area of synfuels and considered both terrestrial
arid aquatic food chains leading to man. The purpose of the workshop
was to obtain the recommendations of experts on (1) terrestrial and
aquatic food-chain models best suited to synfuels risk analysis,
(2) data sources and parameter estimation methods best suited to
synfuels risk analysis, and (3) major limitations on existing data and
methods. Appendixes A and B, respectively, contain the agenda and list
of participants.
The aquatic and terrestrial food-chain exposure models are parts
of an overall assessment process that estimates the human health risk
attributable to alternative synfuels technologies (Fig. 1.). These
particular models estimate the concentrations of chemicals (or chemical
groups) released from synfuels facilities in aquatic and terrestrial
foods. From the concentrations in foods, exposures to individuals and
populations are calculated; exposures (or intakes) are related to risks
by applying models that incorporate dose/response relationships. The
risk analyses in Fig. 1. are based on a characterization of synfuels
-------
SOURCE TERMS
EMISSION RATE
(Mg/year)
AQUATIC
TRANSPORT
I I
CONCENTRATION
IN WATER (pg/L)
ATMOSPHERIC
TRANSPORT
H
DEPOSITION RATE
((jg/m3/s)
1
AQUATIC
FOOD-CHAIN
TRANSPORT
CONCENTRATION
IN AIR (pg/m3>
TERRESTRIAL
FOOD-CHAIN
TRANSPORT
CONCENTRATION
IN AQUATIC
FOODS (pg/kg)
DRINKING WATER
AND CONTACT
CONCENTRATION
IN TERRESTRIAL
FOODS (pg/kg)
HUMAN
EXPOSURE
ASSESSMENT
I
INHALATION
HUMAN HEALTH
RISK
ASSESSMENT
Fig. 1. Components of the overall human health risk assessment
methodology for synfuels technologies.
-------
plant waste streams by their risk analysis categories (RACs) content
formerly Rish Analysis Units (RAUs). RACs are chemicals grouped for
the purpose of risk analysis (Moghissi and Foley 1982). For this
project, 39 RAUs have been defined (Table 1.).
The overall purposes of this risk assessment methodology for
synfuels are to identify chemical groups (RACs) that could pose a
problem for human health and to compare the health risks of alternative
synfuels technologies. Several approaches could be taken in assessing
the potential health effects of each RAC.
First, one could use the conservative approach used in certain
screening assessments to identify potential problems. Parameter values
used in the risk calculations could be chosen conservatively so that
the human health risk is not likely to be underestimated. Chemical
groups that do not produce significant risk, even with conservative
calculations, are unlikely to cause problems. In this application,
some RACs identified as being potential risks will not be shown to be
of great risk upon more detailed investigation. However, for this type
of screening, false positives (safe RACs identified as health risks)
are more acceptable or tolerable than missed faults (problem RACs
passed as safe).
A second approach is to use a single number for each parameter in
the calculations that represents a typical or average value — a best
guess. Deterministic calculations would then produce a typical,
average, or "best-guess" value for the risk associated with each RAC.
-------
Table 1. Risk Analysis Categories (RACs)
RAC No.
Category
Description/Chemical Makeup
1
2
3
4
5
6
Carbon monoxide
Sulfur oxides
Nitrogen oxides
Acid gases
Alkaline gases
Hydrocarbon gases
CO
SOX
NOX
H2S, HCN
NH3
Methane through butanes,
acetylene, ethene
7 Formaldehyde
8 Volatile organochlorines
9 Volatile carboxylic acids
10 Volatile O&S heterocyclics
11 Volatile N-heterocyclics
12 Benzene
13 Aliphatic/alicyclic
14 Morio/diaromatic hydro-
carbons (excluding
benzene)
15 Polycyclic aromatic
hydrocarbons
Ib Aliphatic amines(excluding
N-heterocyclics)
17 Aromatic amines (exluding
N-heterocyclics)
18 Alkaline nitrogen hetero-
cyclics [azaarenes]
(excluding volatiles)
ly Neutral N, 0, & S hetero-
cyclics (excluding
volatiles)
20 Carboxylic acids
(excluding volatiles)
21 Phenols
22 Aldehydes and ketones
(carbonyls) (excluding
formaldehyde)
23 Nonheterocyclic organo-
sulfur
24 Alcohols
25 Nitroaromatics
through butenes; Ci-C4 alkanes, alkynes
and cyclo compounds; bp < ~20°C
CHO
To bp~120°C; C^CLz, CHCL3, CC14
To bp ~120°C; Formic and acetic acids only
To bp~120°C; Furan, THF, thiophene
To bp~120°C; pyridine, piperidine,
pyrrolidine, alKyl pyridines
Benzene
Cs (bp ~40°C) and greater; paraffins,
olefins, cyclopcompounds, terpenoids, waxes,
hydroaromatics
Toluene, xylenes, naphthalenes, biphenyls,
alkyl derivatives
Three rings and greater; anthracene, BaA, BaP,
alkyl derivatives
Primary, secondary and tertiary nonhetero-
cyclic nitrogen, MeNH2, DiMeNH, TriMeN
Anilines, napthylamines, amino pyrenes;
nonheterocyclic nitrogen
Quinolines, acridines, benzacridine; excluding
pyridines
Indoles, carbazoles, benzofurans, dibenzo-
thiophenes
Butyric, benzoic, phthalic, stearic
Phenol, cresols, catechol, resorcinol
Acetaldehyde, acrolein, acetone
Mercaptans, sulfides, disulfides, thiophenols,
CS2
Methanol, ethanol
Nitrobenzenes, nitropyrenes
-------
Table 1. (continued)
RAC No.
26
27
28
29
30
31
32
33
34
3b
36
37
38
39
Category
Esters
Amides
Nitriles
Tars
Respirable particles
Arsenic
Mercury
Nickel
Cadmium
Lead
Other trace elements
Radioactive materials
Photochemical oxidants
Other materials
Description/Chemical Makeup
Acetates, phthalates,
Acetamide, formamide,
formates
benzamides
Acrylonitrile, acetonitrile
As, all forms
Hg, all forms
Ni, all forms
Cd, all forms
Pb, all forms
Ra-226
Ozone
-------
A third approach, similar to the second, attempts to propagate the
variance associated with each parameter. In the best case, the
uncertainty or variability associated with each parameter would be
characterized by a distribution of values, and the resulting assessment
calculations would produce a probability distribution of the risk
associated with each RAC. This approach is more complicated and
requires acquisition of more data and, consequently, requires more
effort. However, in the best case, it could produce results that
include and surpass the results of both methods mentioned previously:
the mean or median of the distribution of risk is similar to the result
from the second approach mentioned, and the 95th or 99th (or 99.99th or
other selected) percentile of the risk distribution may approximate the
result obtained by the first screening method.
The purpose of these synfuels risk assessments is not regulatory
screening in which regulatory action would be initiated or recommended
if conservative screening levels are exceeded. Rather, the purposes of
these synfuels risk assessments are to identify RACs likely to cause
significant health risks and to compare alternative synfuels
technologies using the best available estimates of their risks to human
health. At this stage of the evaluation of the alternative
technologies, both missed faults and false positives are to be
avoided. (Reducing the probability of one type of error tends to
increase the probability of the other; a single method cannot
simultaneously minimize each type of error.) Users of the food-chain
risk assessment methodology described in this report will need to
decide the approach that best fits their assessment goals.
-------
This workshop is one in a series covering different aspects of the
Integrated Health and Environmental Risk Analysis Program (IHERAP) for
synfuels. The workshop considered contaminant accumulation and
transfer in food chains, not physical transport in the ambient
environment. A previous workshop (Georgia Institute of Technology,
Atlanta, Georgia, January 18-20, 1983) covered aquatic transport
modeling for risk analysis (Donigian and Brown 1983).
1.2 ACTIVITIES
The agenda for the workshop is presented in Appendix A. After a
welcome by Dr. Mel Carter and some introductory remarks by
C. Fred Baes III, Or. Alan Moghissi gave a charge to the workshop.
Several of the attendees presented short talks to the entire group
about recent research they have conducted in workshop-related areas.
These presentations are summarized in Sect. 1.3 of this report.
The workshop was then divided into aquatic and terrestrial
discussion groups. Each group reviewed and discussed the issues and
wrote a several-paragraph summary of their discussion. These summaries
were combined, typed, and reviewed by the group on the final day of the
workshop. This report is an edited and expanded version of this
combined report.
1.3 PRESENTATIONS
After a welcome from Dr. Melvin W. Carter (Georgia Institute of
Technology), Dr. A. Alan Moghissi, EPA Project Officer for the
Integrated Health and Environmental Risk Analysis Program (IHERAP) for
-------
8
Synfuels of EPA's Office of Research and Development, gave a charge to
the workshop participants. He briefly described the synfuels
activities currently supported by IHERAP. Dr. Moghissi explained the
use of RACs in relation to the risk analysis of synfuels chemicals. He
defined the objective of the workshop as using the expertise of the
participants to assist risk analysts in deciding which food-chain model
to use.
In his presentation titled "Conceptual Framework for Foodchain
Exposure Assessment," Charles F. Baes III (ORNL) described the
terrestrial food-chain model he has been using (see Sect. 3.1) as it
has been applied in assessments of exposures to synfuels chemicals.
The four key issues Baes suggested for consideration in examining
food-chain models for risk analysis of synfuels were:
1. Is the model structure appropriate?
2. How should parameters be quantified?
3. What are the major sources of uncertainty, and how can
they be quantified?
4. Where should future efforts be directed?
James E. Breck (ORNL) discussed estimation of food-chain uptake of
chemicals by fish by expansion of a structure-activity relationship
that considers only chemical uptake from water. The structure-activity
relationships used are regressions between some chemical parameter,
such as the octanol-water partition coefficient or solubility in water,
and some biological activity of the compound, such as bioaccumulation.
By applying some simplifying assumptions, the octanol-water partition
-------
coefficient K can be used to compute the steady-state contaminant
concentration in fish resulting from both direct uptake from water and
uptake from food (see Sect. 2.1). The results of such computations
suggest that for chemicals that bioaccumulate because they are
lipophilic, a K exists below which food-chain uptake is negligible
ow
and may be ignored. For such chemicals, a bioconcentration factor
(BCF) model should adequately predict the contaminant concentration in
fish.
James Falco (EPA/ORD) described recent work in estimating human
exposures from hazardous waste sites and from contaminated foods.
Recent assessments have moved away from "worst-case" analyses toward
methods that estimate upper and lower bounds of exposure or,
preferably, the entire distribution of exposure. Human exposures from
contaminated foods are estimated by (1) monitoring food contaminated by
chemicals such as pesticides and (2) modeling chemical transport and
transformation up the human food chain. Bioconcentration factors have
been used to predict the concentration of chemicals in fish.
Concentration ratios of chemicals in fish to chemicals in sediment are
useful when the contaminant is undetectable in the water; however, such
ratios are not accurate for running waters, where concentrations
observed in catfish and other bottom feeders are only 5% of the
calculated values.
Jerry Eisele (Oak Ridge Associated Universities, Comparative
Animal Research Laboratory) presented information on his experiments
with poultry, swine, and dairy cattle. He described movement of several
-------
10
RAUs [naphthalene (RAC 14), o-naphthol (RAC 21), and 7-methyl
benz(c)acridine (RAC 18)] through animals to human food products (e.g.,
meat, eggs, and milk). Chemical distribution and concentration
differed among the animals tested. For example, naphthalene fed to
dairy cows reached an approximate equilibrium concentration in milk
within 30 d, but naphthalene fed to chickens remained far from
equilibrium concentration in eggs after 30 d.
Craig McFarlane (EPA Corvallis) described his planned experiments
to quantify the rates of transfer of chemicals to plants. After some
introductory remarks on plant physiology, he described the laboratory
apparatus constructed for plant uptake experiments. He plans to study
chemicals that span a range of values for K and Henry's Law
constants, beginning with the compound bromacil. The results are
intended for use in terrestrial food-chain models.
Chuck Garten (ORNL) discussed his study of empirical and
statistical terrestrial food-chain models. Working with John Trabalka
(ORNL), he developed regression models that predict the chemical
concentration factor (CF) for ruminant fat, nonruminant fat, and avian
fat based on a chemical's water solubility or its K . They found
that the coefficient of determination for the regressions decreased as
additional chemicals were included in the analysis and that error
bounds on the predicted CF were large -- plus or minus two orders of
magnitude. The CFs for mammalian fat, nonrodent fat, and avian fat
2
could be estimated (r = 0.91 and 0.72, respectively) from a
chemical's CF in rodent fat. A chemical's persistence in the soil
-------
11
half-life, however, could not be predicted adequately from K or
ow
water solubility.
John Connolly (Manhattan College) presented the results of
research on modeling the bioaccumulation of contaminants by several
trophic levels in aquatic systems. Using a bioenergetics approach, a
detailed model simulates the assimilation of polychlorinated biphenyls
(PCBs) by lake trout in Lake Michigan. The model accounts for direct
uptake from the water as well as food-chain transfer from phytoplankton
to the opossum shrimp (Mysis) to the alewife to the lake trout. The
model has also been applied to the accumulation of PCBs by yellow perch
in Saginaw Bay of Lake Huron and kepone by striped bass and croakers in
the James River. Connolly compared field data and simulated kepone
concentrations in croakers; good agreement existed between the two.
John Nagy, Brookhaven National Laboratory (BNL), presented
concerns of human health risk analysts about the uncertainty of the
data used in food-chain models. As data are passed from scientists
involved in basic research to environmental modelers to risk analysts,
information about the uncertainty associated with the data and the
applicability of the data for particular purposes should be passed on.
Nagy critically examined the reported BFs for vinyl chloride and
polyaromatic hydrocarbons (PAHs) in fish. He suggested that risk
analysts should be aware of the variation among reported values and the
importance of considering the influence of the fish lipid level in
selecting the appropriate value to use in a particular application.
-------
12
Paul Moskowitz (BNL) described the use of food-chain model output
in assessing the risk of synfuels technologies to human health. The
health risk is initially being assessed by estimating, for each route
of exposure (aquatic foods, terrestrial foods, inhalation, drinking
water, etc.). the annual cancer incidence rate induced by emissions
from synfuels plants. Dose-response models are used to predict the
annual cancer incidence rate from the total mass of each RAC in aquatic
or terrestrial foods consumed by people. Comparisons can then be made
of the differences in cancer incidence rate among the pathways of
exposure, among RACs, and among alternative synfuels technologies.
F. Owen Hoffman (ORNL) discussed uncertainties associated with the
output of models. He distinguished between research models, that are
used to enhance understanding of mechanisms and processes and
assessment models that are used to make predictions upon which
decisions are made. Several sources of model uncertainty are (1) a
particular model's abstraction of reality, (2) variability in the
model's parameters, (3) correlations among parameters,
(4) site-specific capabilities, and (5) deterministic estimates that
ignore system variability.
2. AQUATIC FOOD-CHAIN MODELS
Aquatic food-chain models are used to estimate the concentration
and total amount of each RAC in the edible fraction of aquatic foods
consumed by man. Estimates of RAC concentrations in water and sediment
and other relevant physiochemical variables are calculated using an
-------
13
aquatic transport model (e.g., Travis et al. 1983). The eventual
impact of ingesting a particular RAC, taking into account the effects
of any food processing on toxicity, is addressed in the human exposure
and human health risk components of the overall assessment methodology
(Fig. 1.).
Some very similar and frequently used terms are defined here to
facilitate understanding of the subsequent discussion. Concentration
factor (CF) refers to the ratio between the concentration of a chemical
in an organism and in water. There is no implication that the
concentration in the organism is at steady state, though it may be.
All uptake pathways are considered, as they are in CFs derived from
field data. Bioconcentration factor (BCF) also refers to the ratio
between the concentration of a chemical in an organism and in water.
The concentration in the organism is assumed to be at steady state, and
only uptake from water is considered. (This same term is .used in some
terrestrial papers to mean the ratio between a concentration in an
organism's fat and the concentration in its diet.) Biomagnification
factor (BMF) is the ratio of the steady-state concentration in an
organism to the concentration in its food when food is the only route
of exposure (i.e., when the concentration in water is negligible).
Bioaccumulation factor (BAF) is the ratio of the steady-state
concentration in an organism to the concentration in water, if both
food and water contribute to the organism's exposure.
-------
14
2.1 MODELS BEST SUITED TO SYNFUELS RISK ANALYSIS
Estimation of the risk posed to humans by the components of
synfuels released to natural water systems requires estimation of the
concentration of these components in aquatic species consumed by
humans. It was the opinion of the human health risk analysts at the
workshop that the uncertainty associated with the human health
dose-response estimates was approximately plus or minus two orders of
magnitude (i.e., given an accurate estimate of dose, that the response
could be predicted within a factor of 100). It should be noted that
the level of knowledge about the dose-response relationship varies
among RACs; thus, the level of uncertainty associated with the response
estimates varies accordingly. Knowing this level of uncertainty may be
helpful in selecting an appropriate aquatic food-chain model.
2.1.1 Concentration Factor Model
The analysis of chemical accumulation in aquatic species has been
approached from several levels of detail. The simplest method computes
the concentration of a chemical in a species as the product of a CF and
the concentration of dissolved chemical in water. This method was used
in recent assessments of synfuels wastes (Moskowitz et al. 1983) and
has been used by the U.S. Environmental Protection Agency (EPA) in
developing water quality criteria (EPA 1980) and by the U.S. Nuclear
Regulatory Commission (NRC) for radionuclide assessments (NRC 1977).
The CF is obtained from field data, from laboratory testing, or from an
empirical relationship between CF and the octanol-water partition
coefficient (K ). For these assessments of synfuels technologies, a
OW
-------
15
CF is needed that makes the best prediction of the contaminant
concentration in fish chronically exposed under field conditions. The
CF should account for all important routes of uptake. A CF estimated
from field data is preferred only if the exposure is chronic. If such
a field CF is not available, an appropriate CF must be estimated. If
the elimination rate of the chemical from fish is rapid, (1) the food
chain is unlikely to be an important route of uptake and (2) the BCF is
likely to be a good approximation of the CF (Macek et al. 1979; Thomann
1981; Bruggeman et al. 1981; Oliver and Niimi 1983). Chemicals having
rapid elimination rates will include those that are not highly
lipophilic. [Chemicals, such as methyl mercury, which have a low K
ow
but are accumulated by mechanisms unrelated to lipophilic
characteristics must be given separate and special consideration.]
The experiments and field data of Oliver and Niimi (1983) suggest
that the BCF will approximate the field CF for chemicals having a log
Knul less than about 4.3; Bruggeman et al. (1981) suggest that the
OW
food chain will make an important contribution only for chemicals
having log BCF greater than about 5. Since the workshop, approximately
60 possible synfuels chemicals have been assessed at ORNL, and the
concentrations in fish appear to be adequately estimated by the CF
method; for these chemicals, either a field CF is available, the
chemical is not highly lipophilic, or the chemical is rapidly
metabolized by fish (e.g., PAHs, RAC 15). Section 2.2 discusses the
estimation of BCFs and field CFs, and Sect. 2.1.2 discusses models that
can be used when the food-chain pathway to fish is likely to make an
important contribution to the total contaminant concentration in fish.
-------
16
2.1.2 Dynamic, Bioenergetlcs-based Models
Dynamic, bioenergetics-based models consider the uptake and
depuration dynamics of one or more trophic levels in the food chain.
These models can account for the influence of factors such as body size
and water temperature on contaminant dynamics through effects on growth
rate, respiration rate, and consumption rate. Detailed models based on
a bioenergetics approach can give good estimates of contaminant
concentration in fish (estimated by the aquatics group to be within a
factor of three), given a good estimate of depuration rates and the
contaminant concentrations in water and food (e.g., Weininger 1978;
Thomann and St. John 1979; Thomann 1981; Thomann and Connolly 1984).
These models may be required for the best estimates of concentrations
in fish of compounds having very low depuration rates (e.g., PCBs)
because these compounds take a long time to reach equilibrium. Because
such models can account for effects of fish age, size, and fish lipid
content, they can provide the high-level resolution required for risk
analysis of critical human populations.
Bioenergetics-based models have already been developed for several
fish species and have been applied to several specific situations:
PCBs in lake trout, coho salmon, and alewives in Lake Michigan; kepone
in striped bass and croakers in the James River; methyl mercury in pike
and roach in Swedish lakes; and PCBs and methyl mercury in Ottawa River
yellow perch [Connolly and Tonelli (in press); Fagerstrom and Asell
1973; Fagerstrom et al. 1974; Norstrom et al. 1976; Weininger 1978;
Thomann and St. John 1979; Thomann and Connolly 1984]. Because of the
-------
17
level of detail used in these models, the resulting equations are
complex and application to a specific food chain requires knowledge of
many parameters, some of which are hard to obtain and/or are site
specific. However, once the biological data are accumulated for a
particular species and location, it is relatively easy to apply the
model to additional chemical contaminants having similar properties.
These models can give detailed, time-varying estimates of contaminant
concentration in fish of several ages. The risk analyst should
consider the following criteria in deciding whether or not to use these
detailed models:
1. chemicals of interest should have low depuration and
biotransformation rates [i.e., long times to reach
steady state (Thomann 1981)] such that the steady-state
assumption of certain simpler models is violated;
2. a model and paramter, or the resources to develop them,
should be available for the chemical, species, and
location of interest; and
3. the level of accuracy or detail that these models can
provide should be worth the increased effort required.
Considering these criteria, it appears that the detailed models are
currently inappropriate for synfuels risk analysis.
One of the possible shortcomings of BCF models is that the
potential contribution of the food chain to bioaccumulation may be
ignored. A relatively simple model that includes the food chain can be
constructed based on the following assumptions:
-------
18
1. the food chain is linear,
2. each trophic level is represented by fixed bioenergetic
parameters,
3. the concentration of chemical in the food chain is in
dynamic equilibrium with the concentration of chemical
in water, and
4. the rate of direct uptake from water is independent of
trophic level (it may be easy to relax this assumption
by accounting for body size).
The resulting model may be termed a bioenergetic-based equilibrium
food-chain model or an "equilibrium food-chain model." The model
presented by Thomann (1981) did not assume a constant BCF. Bruggeman
et al. (1981) independently derived a similar model. The Thomann model
calculates the steady-state BAF for the n trophic level (BAF ) in
a linear food chain. The "food-chain transfer number" f (Thomann
1981) or "biomagnification factor" (Bruggeman et al. 1981) depends on
the assimilation efficiency of ingested chemical, the consumption rate,
the contaminant elimination rate, and the growth rate of organisms in
trophic level n (Thomann 1981). The model is highly sensitive to
elimination rate. Errors in its estimation will produce rather large
errors in BAF .
To simplify further it may be inferred from Bruggeman et al.
(1981) that the food-chain transfer number f is (approximately)
directly proportional to the bioconcentration factor (i.e., f will
be on the order of 10 to 10 times BCF ). As Bruggeman et al.
-------
19
(1981) point out, this suggests that chemicals in the food will make an
important contribution to the concentration in an organism only for
chemicals with a BCF of 10 or greater. This result is shown in
Fig. 2, where the additional assumption has been made that the
food-chain transfer numbers (f ) and the BCF from water (BCF ) are
the same for all trophic levels. In this figure, the log BCFi vs log
K relation is taken from Oliver and Niimi (1983), who corroborated
uw
their laboratory measurements of BCF with field data on fish/water
contaminant ratios in Lake Ontario rainbow trout. This figure shows
the importance of considering the food-chain pathway to fish for
chemicals with very low elimination rates (i.e., chemicals having very
high BCFs).
The accuracy of the equilibrium food-chain model is dependent on
the validity of the four assumptions indicated earlier and the
uncertainty associated with the input parameters. The amount of
reduction in uncertainty resulting from the increase in process
specification in this model, relative to the simple water BCF approach,
is unknown and remains to be quantified.
A major assumption of this model is that equilibrium or steady
state is closely approached. Work in progress by Breck (ORNL) suggests
that chemicals for which the food-chain pathway is a dominant uptake
route are likely to have very low depuration rates (and high K
UW
values), so that a fish may not reach a steady state within its
lifetime. The simple equilibrium models of Thomann (1981) and
Bruggeman et al. (1981) are easily modified to account for this
deviation from steady state without greatly increasing model complexity.
-------
20
UJ
UJ
_l
o
I
Q.
O
o:
o:
2
DC
o
10
9
8
7
6
5
ORNL- DWG 83-12539
8 °
8 -i
-2
1
345
LOG KOW
8
Fig. 2. bioaccumulation factor (BAF) vs octanol-water
coefficient (Kow) for different trophic levels
food chain. The line for n = 1 reflects
bioconcentration from water only, as measured a
experiment (Oliver and Niimi 1983). The line
includes contaminant uptake from water plus
transfer from trophic level 1. The line for n :
direct uptake from water plus food-chain transfer
levels 1 and 2. Lines for trophic levels 4 and 5
similarly.
partitioning
in a linear
contaminant
low-exposure
for n = 2
food-chain
= 3 includes
from trophic
are computed
-------
21
For the 60 or more synfuels chemicals considered thus far in
synfuels assessments at ORNL, concentrations in fish appear to be
adequately estimated by a simple CF model (see Sect. 2.1.1). However,
if chemicals are encountered that do not fit the requirements of the CF
model (e.g., chemicals for which there are no field estimates of CFs
and for which depuration and biotransformation rates are very low and
values for K very high, suggesting that the food chain is the
uw
dominant uptake pathway), then a dynamic bioenergetic model would be
appropriate. The scientists at the workshop believed that if use of
this type of model were necessary, then data currently available in the
literature could be used.
2.1.3 Comments For All Models
Concentration factors for aquatic organisms can vary among taxa,
trophic levels, and sizes of individuals. If aquatic foods represent a
major source of exposure or risk for a particular chemical (or group),
as determined by preliminary analyses, such differences should be
accounted for in synfuels risk analyses. For example, some crustaceans
and molluscs lack the ability of fish to rapidly metabolize PAHs (RAC
15), so that BCFs are higher for certain of these organisms than for
fish (Maiins et al. 1979; Southworth et al. 1980; Farrington et al.
1983). Differences among taxa in growth rate and trophic level also
contribute to variations in CFs (Thomann 1981). Larger body size is
correlated with a lower elimination rate; this has been observed for
methyl mercury in fish (Norstrom et al. 1976), and for PCBs in
macrozooplankton and phytoplankton (Brown et al. 1982). Because
-------
22
elimination rate influences the CF, and most laboratory measurements of
BCF have been made on much smaller sizes of fish than those typically
consumed by humans, it would be useful to know the contribution of body
size to increased uncertainty in estimated CF. A literature survey
should be done to quantify the uncertainty in CF (or BCF) resulting
from variations among taxa, trophic levels, and sizes of individuals.
If justified by the results of such a literature survey, the
aquatic food of humans could be categorized as follows:
Level 1, a phylogenetic classification. Data on the harvest and
consumption of aquatic foods from commercial and sport fisheries and
aquaculture would be directly or indirectly available on a taxonomic
basis. Species specification would allow determination of the feeding
habit class (Level 2) and the parameter values (e.g., lipid content)
required for estimating the CFs. The important taxa in the aquatic
food chain are (1) Pisces, (2) Mollusca, (3) Crustacea, and
(4) Aves/Anseriformes.
Level 2, feeding habit (or some trophic-level scheme).
Recognizing that combinations are the rule rather than the exception,
these might include (1) detritivore, (2) phytoplanktivore,
(3) zooplanktivore, (4) macroherbivore, (5) benthic invertebrate
feeder, and (6) piscivore. Other distinctions may be necessary (e.g.,
feeding on aquatic benthic invertebrates vs feeding on invertebrates of
terrestrial origin). Not all feeding habit classes will be represented
in a given phylogenetic group.
-------
23
Other possible, important classifications would distinguish
between freshwater, saltwater, and estuarine species and between
flowing- and standing-water species or populations.
The method of classification described here is general, possibly
exceeding the needs of an assessment program. IF a preliminary
analysis using the simplest approach showed that greater resolution is
required for particular chemicals in aquatic foods, then, depending on
the results of a literature survey, the method could be expanded to a
more complex form.
A workshop participant suggested that a data book documenting the
derivation of CFs should be kept. This would include sections on each
substance/RAC, including pertinent math, biology, and chemistry,
computations performed; a qualitative review of uncertainty, both at
individual compound and RAC levels; and, if possible, quantification of
uncertainty via confidence intervals and/or probability distribution
functions (PDFs) for possible use in stochastic assessment models. It
is recognized that these intervals or PDFs will vary according to the
question addressed by the model; larger uncertainties will be
associated with initial group assessments than with assessments of risk
to large populations.
As discussed in the Introduction, the overall purposes of this
risk assessment methodology are to identify chemical groups (RACs) that
could pose human health problems and compare the health risks of
alternative synfuels technologies. Therefore, it is suggested that
calculations of food-chain exposure to each RAC use values for a
-------
24
typical chemical or a value obtained by averaging across several
chemicals in the RAC, noting and carrying through the variance or
variability in output; an alternative method would be to do the
calculations for several representative chemicals within each RAC to
account for the variability among chemicals within a RAC. The purpose
here is not regulatory screening (i.e., regulatory action would not be
initiated or recommended if conservative screening levels were
exceeded). For the regulatory screening case, upper-bound estimates
would be more appropriate than average or typical values for CFs.
Depending upon the goals of a particular risk assessment, different
methods might be useu to select parameter values and to account for the
variability among chemicals.
For the purposes of the EPA synfuel assessment, then, at least
best guess estimates of uncertainty are needed although ideally PDFs
would be available for all substances and aquatic food types. The
workshop aquatic food-chain group did not, however, reach a consensus
on this point. The following two paragraphs represent a strongly felt
minority opinion.
Currently, uncertainties in 95% upper-bound risk estimates warrant
only an order-of-magnitude accuracy for estimates of BAFs. However,
even when ranges are specified for BAF values, their derivation has
been usually associated with conservative bias. Therefore, because of
limitations in available data, current models are not capable of
producing a best-estimate value or of values. Thus, the most
appropriate use of BAF values to date is in screening calculations.
-------
25
In all cases, however, it is desirable to preserve the range or
distribution of uncertainty of BAFs with stochastic modeling. It is
not appropriate to use single values in deterministic models unless the
approach is known to be sufficiently conservative for screening
calculations. Deterministic calculations based on indiscriminate use
of geometric means may be particularly misleading.
In view of the location of possible synfuel plants, specification
of the following variables/parameters for freshwater fishes have the
highest priority: CF (else BCF) by RAC, a factor to convert whole-body
concentration to concentration in the edible portion of the aquatic
food, and a factor to account for changes in concentrations resulting
from food processing and cooking. A survey of the literature should be
done to quantify the uncertainty, in CF (else BCF) resulting from
variations among taxa, trophic levels, and sizes of individuals. To
estimate the mass of each RAC in aquatic foods ingested by humans, the
following information is needed as well to complete the exposure
assessment: the mass of each aquatic food type taken from each stream
reach or water body, the fraction that goes to human consumption, the
edible fraction, and the number of people eating food of each subtype
from each stream reach/water body.
2.2 DATA SOURCES AND PARAMETER ESTIMATION METHODS
When a CF method is to be used in risk analysis as suggested here,
the best way to estimate the factor is to use field data on the ratio
of the concentration of contaminant in fish to the concentration of
-------
26
contaminant in water. This approach has been found to be useful for
estimating BAFs for radionuclides in freshwater biota (Vanderploeg et
al. 1975). Data on the concentration in fish categorized by species,
lipid level, and weight and age will be needed for later, more detailed
risk analyses.
Field data should be included in risk assessment only if:
1. the concentration of a chemical in aqueous solution is
measured,
2. the concentration in fish is measured (may be whole
organism or a portion), and
3. the exposure concentration is reasonably constant for an
adequate period of time (i.e., the rate of approach to
steady-state CF is chemical-dependent) — not the result
of a local spill or other temporary or localized
condition.
Tnis approach to the estimation of concentration ratios includes all
pathways of exposure for aquatic organisms under the most realistic
conditions. Unfortunately, such measurements are not available for
many of the synfuels organics. Data from acute field exposures are
also valuable but need to be analyzed carefully to yield CFs
appropriate for the risk analyses of chronic exposures discussed here.
As previously mentioned (Sect. 2.1.1), if an appropriate field CF
is not available, then an appropriate CF must be estimated. If either
the elimination rate or the rate of biotransformation of the chemical
is rapid, the BCF is likely to be a good approximation of the CF
-------
27
(Thomann 1981; Bruggeman et al. 1981), and estimates are needed of the
BCF of the chemical. This is likely to be the case for most synfuels
chemicals, according to recent assessments of synfuels chemicals at
ORNL (Breck, in preparation). If both the elimination rate and the
rate of biotransformation are very low, the food-chain uptake pathway
may contribute significantly to the chemical's concentration in fish,
and additional information will be needed to estimate an appropriate
CF, as discussed in the following.
The laboratory and field data of Oliver and Niimi (1983) suggest
that the steady-state BCF measured in the laboratory will be a good
approximation of the field CF for chemicals having a log K of about
ow
4.3 or less. (It should be noted that the Oliver and Niimi values for
BCF were higher than thse measured by some other workers, perhaps
because Oliver and Niimi used the large-size fish or very low exposure
concentrations.) Bruggeman et al. (1981) suggest that the food-chain
uptake pathway is probably not significant for chemicals having a log
BCF of less than about 5.
A measured BCF is generally preferred over an estimated BCF. The
fish/water concentration ratio should be measured very close to steady
state or the steady-state fish/water ratio (i.e., BCF) should be
estimated from the ratio of uptake rate to elimination rate (including
the biotransformation rate). The results of Kosian et al. (1981)
suggest that the BCF is appreciably influenced by the method of
calculation of BCF from laboratory measurements. The ratio of uptake
rate to elimination rate seems to be more appropriate for estimating
-------
28
the BCF of very lipophilic chemicals than are the 28-d values of the
fish/water concentration ratio.
If field or laboratory CFs are not available for the compound of
interest, BCFs should be estimated from regression models developed for
homologous chemicals (these will often be members of the same RAC).
For example, separate regression models for PAHs, chlorinated benzenes
(Oliver and Niimi 1983), and phenols (Saarikoski and Viluksela 1982),
could be used. Estimates from other homologs should be especially
useful for cases like the PAHs, in which BCFs are relatively low,
despite high log K s, as a result of metabolism of the compound by
OW
fish. If data on homologs are not sufficient to estimate a BCF, BCF
should be estimated from the octanol-water partition coefficient by
using regression models based on a wide variety of chemicals (e.g.,
Trabalka and Garten 1982; Veith et al. 1980).
Chemicals that ionize to a significant extent at field pHs have a
reduced potential for accumulation, as Saarikoski and Viluksela (1982)
found for certain substituted phenols. For such chemicals, the
reduction in BCF resulting from ionization at a given pH can be
estimated if the dissociation constant is known (Saarikoski and
Viluksela 1981; Goldstein et al. 1974).
BCF data for fat-soluble RACs should be accompanied by lipid
content information so that the BCF values can be expressed on either a
whole-body or lipid-weight basis, as necessary. BCFs represented on
the basis of lipid content can be easily converted to either
edible-portion or whole-weight basis using reasonable estimates of
-------
29
lipid content in tissues. For example, if the BCF on a lipid basis is
10,000, whole fish are 3% lipid, and the edible portion is 5% lipid,
then the BCF in whole fish would be 300 and the BCF in edible portion
would be 500. General estimates of the percentage of lipid content are
available for many commercial species (Sidwell et al. 1974; Kinsella et
al. 1977; Rottiers and Tucker 1982). The workshop group felt that such
estimates are generally useful for a risk assessment.
In the event that a linear food-chain model is required (Sect.
2.1.2), values for the BCF, elimination rate, consumption rate,
assimilation efficiency of ingested chemical, and growth rate would be
required for each trophic level. The BCF and .the excretion rate can be
determined from laboratory experiments or estimated from the K
ow
(e.g., Oliver and Niimi 1983; Spacie and Hamelink 1982; Trabalka and
Garten 1982; Mackay 1982; Veith et al. 1980; Kenaga and Goring 1980).
Growth rate can be obtained from the literature or may be estimated
from general functions relating these parameters to the 'size of the
organism. Assimilation efficiency of the chemical can be obtained from
laboratory experiments or estimated from the K (see Schanker
OW
1960). Consumption rate, which varies with body size, can be obtained
from the literature or estimated from the growth rate, respiration
rate, and food assimilation efficiency (Norstrom et al. 1976; Thomann
1981).
Models having two body compartments for the organism are not
necessary if the bioaccumulation ratio is derived from a log K
(J W
correlation, chronic laboratory exposure, or field data. However, they
-------
30
may be considered if the food-chain model is developed from uptake- and
elimination-rate constants measured in the laboratory (Ellgehausen
et al. 1980). In some cases, the elimination-rate constant can be
overestimated by making observations only during the beginning of
depuration. Elimination of organics is often made up of "fast" and
"slow" phases. If the initial slope of a biphasic depuration curve is
used as an estimate of the elimination-rate constant, then the
half-life of the chemical will be underestimated and the calculation:
uptake rate
= elimination rate (2.1)
will produce an underestimation of BCF.
2.3 MAJOR LIMITATIONS ON EXISTING DATA AND METHODS
Concentration factors measured from chronic field exposures would
provide the most appropriate values for use in synfuels risk analyses,
given the conditions listed in Sect. 2.2. Such field exposures include
several uptake pathways and more realistic conditions than laboratory
exposures. Measurements of CFs observed in the field are valuable
sources of information about the behavior of organic chemicals, and
such measurements provide a way to calibrate and verify the models.
However, it should be understood that the results will probably vary
from one ecosystem to another because of factors such as food-chain
length, bioavailability of the chemical, and species characteristics.
Compilation of results from different systems will allow the magnitude
of this variation to be estimated.
-------
31
If Thomann's (1981) equilibrium food-chain model (see Sect. 2.1.3)
is used to calculate the CF, estimates will be needed of the BCF from
water only. Several limitations exist on the use of structure-activity
relationships (SARs) to estimate BCF. First, an appropriate equation
must be used (Trabalka and Garten 1982; Oliver and Niimi 1983; Veith et
a.l. 1980), as discussed in Sect. 2.2. Second, for benthic species, a
concentration ratio relative to the concentration of chemical in
sediment may be more appropriate than a ratio relative to the water
concentration. Third, it must be recognized that there is generally
uncertainty associated with the value of the K (the independent
ow
variable in the regression) as well as error associated with the
dependent variable (here, BCF) in the regression (Trabalka and Garten
1982; Veith et al. 1980).
The regressions relating BCF and K that are currently used
OW
were developed mainly for organo-chlorines and other organics that are
poorly metabolized by fish. To a close approximation, these chemicals
remain unchanged throughout the partitioning process and throughout the
food chain. These regressions work rather well for such materials.
However, many of the organics associated with synfuels (e.g., PAHs) are
known to undergo biotransformation in some aquatic organisms
(Southworth et al. 1980; Spacie et al. 1983). Usually, but not always,
the metabolites formed are more polar and are, therefore, eliminated
more quickly from the body. In effect, the organism has the ability to
lower the partition coefficient of the parent material. Thus,
estimates based on the log K of the original chemical may produce
OW
-------
32
overestimates (conservative estimates) of BCF and food-chain
transport. Enzyme induction may also lower BCF.
It would be extremely useful to collect information to group
aquatic organisms on the basis of their biotransformation abilities.
For example, fish and chironomid midges readily metabolize many PAH
compounds, whereas bivalves and Daphnia are poor hydrocarbon
converters. Categorization of this type would dramatically facilitate
our estimates of BCF and food-chain transfer for synfuel chemicals.
Currently, no good structure-activity models exist for predicting in
advance which taxa will or will not metabolize a particular organic
compound.
A fourth limitation on the use of SARs to estimate BCFs concerns
the variaole(s) used to quantify a chemical's structure. The most
commonly used regressions predict BCF from K . As others have
ow
pointed out (e.g., Trabalka and Garten 1982), such relations will
underestimate the bioconcentration of chemicals such as methyl
mercury. This is because methyl mercury bioaccumulation is not related
to lipid partitioning, but to bonding with the sulfhydryl groups of
certain proteins (Reichert and Mai ins 1974). Therefore, its
bioaccumulation would not be expected to correlate with a variable
(K ) related to lipid partitioning. Fortunately, however, there
OW
appears to be only a very small portion of chemicals that behave like
methyl mercury.
A final limitation on the use of regressions to estimate BCFs is
the relatively large degree of uncertainty about the predicted values.
-------
33
Regression equations developed for a homologous group of chemicals will
have a smaller range of uncertainty about the predicted values than
will regressions developed from a wide variety of chemicals. Trabalka
and Garten (1982) compiled bioconcentration information on more than
100 chemicals. In their regression of log BCF vs log K "the 95%
ow
confidence limits about the predicted value at the mean log [K ] are
OW
greater than +2 orders of magnitude." They note, however, that "the
bulk of the extreme data scatter occurs below the regression line...and
overpredictions by the model are more acceptable than underpredictions."
A limitation of the structure of the models (presented in Sect.
2.1) is that they are based on the concentration of the chemical in
water. Situations in which the concentration of sediment-associated
chemical is important (e.g., through a benthic component of the food
chain) may require that this benthic component be separated and
referenced to the sediment rather than to the water column. Factors
such as adsorption to dissolved organic matter (e.g., humic acid) may
also reduce the concentration available in water and, therefore, reduce
the amount of chemical absorbed by the organism. Such factors should
be accounted for by the aquatic transport model.
2.4 CONCLUSIONS
The aquatic food-chain group at the workshop agreed that for
chronic low-level releases of synfuels effluents, the concentration
factor approach is relevant and appropriate for assessing risks.
Application of field data based on the ratio of a chemical's
-------
34
concentration in fish to the concentration in water is the best way to
estimate the CF. If field data are not available, laboratory
measurements of BCFs should be used. If no measured BCF is available,
then the factor can be estimated using regressions of log BCF vs log
Kow'
The data of Oliver and Niimi (1983) suggest that simple BCF models
adequately predict the concentration in fish of chemicals with a log
K of about 4.3 or less. The great majority of synfuels chemicals
ow
fall in this category. Those synfuels chemicals that have a log K
OW
greater than about 4.3 are typically metabolized rapidly (e.g., PAHs,
RAC 15), consequently, they are not accumulated significantly through
food chains and are adequately described by a BCF model. Chemicals
having low depuration and biotransformation rates are candidates for
significant uptake via the food chain, making models that explicitly
account for food-chain uptake applicable. No such synfuels chemicals
have yet been identified, except for those for which field estimates of
CFs are already available (i.e., mercury, RAC 32).
Although information is currently available on the uncertainty
associated with regression models (i.e., log BCF vs log K ), further
OW
research is needed to quantify (1) the level of uncertainty associated
with CFs estimated from field data and (2) the differences among taxa
and among trophic levels.
-------
35
3. TERRESTRIAL FOOD-CHAIN MODELS
3.1 MODELS BEST SUITED TO SYNFUELS RISK ANALYSIS
The appropriateness of a terrestrial food-chain model for synfuels
risk assessment may be judged by its usefulness to the risk analyst,
who must not only perform the initial risk estimates but also identify
critical pathways, uncertainties, and research needs. The model
currently used at ORNL for terrestrial food-chain exposure calculations
is based on a modification of the traditional multiplicative chain
approach used by the NRC in making radionuclide dose assessments (NRC
1977; Travis et al. 1983). The computer code incorporating this model,
TERREX (Baes et al. in preparation), is a modified version of the TERRA
computer code for radionuclide transport (Baes et al. 1983). The
various pathways adoressed in the TERREX code are shown in Fig. 3. The
breakdown and interrelationships among these pathways are based on a
review and analysis of agricultural practice in the United States, by
Shor et al. (1982).
The terrestrial food-chain transport model includes produce, milk,
and beef pathways to humans. Input into the terrestrial model is via
atmospheric dispersion and deposition. Plant compartments are
contaminated via root uptake of material that has deposited and
accumulated in soil and via direct contamination of foliar surfaces.
The root uptake component of plant contamination is calculated through
the use of soil/plant CFs that are specific for either leafy (B. ) or
reproductive/storage (Bir) organs of the plant. These CFs relate the
dry weight concentration of compound i in the edible plant part to its
-------
ORNL-DWG 83-12989
CONCEPTUAL GENERIC TERRESTRIAL FOODCHAIN
Air
—'/»—
I
r
LEAFY »
VEGETABLES '^ "^
I 1
I* 1
t
Ir
NON-IRRIGATED SOIL
f
PROTECTED
PRODUCE
,
•
i •
L_
-'«-•
F,
MILK.
-*"1
-
.
>\
GRAIN
..
A r
MAN
"EXPOSED-
PRODUCE
IRRIGATED SOIL
-A.,-
KEY
INTERCEPTION FRACTION
ATMOSPHERIC LOSS
SOIL LOSS
BIOACCUMULATION 'VEGETATIVE-
BIOACCUMULATION -REPRODUCTIVE'
INGESnON-TO-MILK
INGESTION-TO-BEEF
Co
O!
Figure 3. Pathways addressed in the terrestrial food-chain computer code TERREX.
-------
37
dry weight concentration in root zone soil (assumed to be to a 15-cm
depth). Meat- and milk-producing cattle are assumed to feed on grain
and locally-contaminated forage. The grain may be grown locally or
imported from offsite, depending on sizes of local livestock herds and
grain production within the assessment area. Beef and milk
concentrations are calculated via beef and milk transfer coefficients
(Ff and F , respectively), which relate daily intake of the
contaminant to equilibrium concentrations in the food products. The
derivation of the soil/plant CFs and beef and milk transfer
coefficients relies heavily on structure/activity relationships (based
on log K ) reported in the literature (Briggs 1981; Kenaga 1980;
ow
Baes 1982).
The terrestrial transport model is coupled with a data base (SITE)
of default location-specific parameters describing agricultural,
meteorological, and land-use characteristics for each cell of a
rectangular grid superimposed on the conterminous United States. The
dimension of each grid cell is 0.5°x 0.5° latitude-longitude; and for
each cell, 36 parameters are defined (Table 2). In performing an
assessment of a synfuels plant, the assessment grid defined by the
atmospheric dispersion model (for distances less than 50 km, this is
usually a polar grid) is superimposed on the data base grid and
parameters are recalculated to the assessment grid. This process
allows the assessment model to reflect, in a general way, the actual
agricultural practices and characteristics of the area being modeled.
The details of the data base and the sources of the parameter values
within it are described by Baes et al. (1984b).
-------
38
Table 2. Variable names and descriptions of parameters in the SITE* data base.
Variable name
CELLON
CELLAT
ET
IRRI
PRECIP
YEV
AYBF
YLV
YSF
AREAP
NUrtCC
SALFC
NUMMC
NUMSHP
PGF
PHF
P3F
HUM ID
ARE AT
POP
FRUIMF
FRUFM
FURbN
PLV
PEV
PPV
PGH
MWIXHT
AMIXHT
YPV
YGF
YGH
FFDAYS
NUMBC
Units
°W
°N
mrn/y
mm/y
mm/y
kg(fresh)/m2
kgtdryj/y/m2
kg(fresh)/m2
kg(dry)/m2
rnZ
head
heaa/y
head
head
kg
kg
kg 3
m2
number
unitless
unitless
unitless
kg
Kg
kg
kg
m
m
kg(fresh)/m2
kg(fresh)/m2
kg(dry)/m2
number
head
Description
longitude of the southeast corner of the SITE
data cell
latitude of the southeast corner of the SITE
data cell
evapotranspiration
irrigation
precipitation
yield of exposed produce
areal yield of hay feed
yield of leafy vegetables
yield of (corn and sorghum) silage feed
total area of pasture
cattle and calf inventory
number of cattle on feed sold
number of milk cows
number of sheep
production of grain for feed
production of hay feed
production of (corn and sorghum) silage feed
average annual absolute humidity
total area of cell
population of cell based on 1980 census
fraction of 1980 population classed as rural
non-farm
fraction of 1980 population classed as rural
farm
fraction of 1980 population classed as urban
production of leafy vegetables
production of exposed produce
production of protected produce
production of grain for human consumption
morning mixing height
after noon mixing height
yield of protected produce
yield of grain food
yield of grain feed
number of frost-free days in a year
number of beef cattle
-------
Table 2. Continued.
39
Variable name Units
Description
CFLAli
number
UOMLF
number
the caution flag: 1 means cell on atlantic
coast. 2 means cell on Mexican border. 3
means 1 and 2. 4 means cell has interior
body of water. 5 means 1 and 4. 6 means 2
and 4. 8 means cell on pacific coast. 12
means 4 and 8. 16 means cell on Canadian
border. 20 means 4 and 16. 21 means 1 and
4 and 16. 24 means 8 and 16. 32 means
cell has desert or barren land. Finally,
36 means 4 and 32.
A five digit number of the form FLPPP is
printed out. F = 1 if the cell is more
than 50% federal land, i.e., land type is
undefined. F = 0 if less than 50% of the
cell is federal land. L = 1 means tall row
crops. L = 2 means short row crops. L = 3
means has or tall grass. L = 4 means urban
area. L = 5 means small lakes. L = 6
means short grass. L = 7 means forest.
Finally, PPP = the percentage of the
dominant land type (may be 0 if the SITE
cell is 100% federal land).
*baes et a I., 1984.
-------
40
Although the food-chain model represented by Fig. 3 appears to be
more complex than the aquatic food-chain model, both are based on the
use of BCFs to calculate accumulation in various food-chain trophic
levels. The apparent complexity of the terrestrial model arises from
the necessity of considering the many varied components of agricultural
food chains (e.g., produce exposed to and produce protected from direct
contamination of edible parts and management practices for milk,
feedlot, and other cattle). From the standpoint of the uncertainty in
model predictions, the participants at the workshop felt that increased
complexity did not improve the model's accuracy as compared to a
simpler model. However, those workshop participants who have performed
risk calculations felt that a simpler approach would not provide
sufficient detail in addressing the complex differences, identifying
important exposure pathways, and highlighting research needs. Workshop
attendees involved in work in which model input parameter values are
experimentally determined argued that the model is too simple to
address real-world transport processes and provides little guidance in
their work. However, they agreed that, within the context of the
overall assessment procedure, the model is a reasonable approach,
provided there is an ongoing dialogue between the modeler and the
experimentalist. This latter point will be addressed subsequently in
Sect. 3.2.
Because of the limited data base on terrestrial transport and
behavior of synfuels effluents, many potentially important pathways are
not included in the terrestrial food-chain model. Recognizing the
-------
41
difficulty in quantifying many of these transport process, the workshop
attendees, nevertheless, identified several potentially important
issues that should be considered at some point in the modeling
process. These issues include foliar adsorption and translocation to
edible produce parts; the effects of food processing (especially
cooking) on human exposures; the prediction of soil degradation
kinetics based on structure/activity relationships; the contribution of
animal products other than beef and milk (e.g., eggs, chicken, and
pork); differences in transfer coefficients as a result of livestock
management practice; ingestion of water and soil by livestock; and
irrigation water as a source term to the terrestrial food chain. The
attendees ranked the first three issues as being more important than
the latter four. Each of these issues is discussed in Sect. 3.3.
3.2 DATA SOURCES AND PARAMETER ESTIMATION METHODS
Early in the workshop, the methods for estimating model parameter
values and the available data sources were discussed. The agricultural
production and practice parameters in the SITE data base are derived
from a Census of Agriculture conducted by the U.S. Department of
Commerce (DOC 1977). These data are summarized by county and were
converted to the uniform 0.5°x 0.5° cell grid by methods described in
Baes et al. (1984b). A more recent Census of Agriculture is available,
but differences in agricultural practice and production represented by
the more current census were thought to be relatively small (less than
10%). The meteorological data are 30-year averages collected from over
-------
42
200 U.S. weather stations maintained by the National Oceanic and
Atmospheric Administration (Ruffner 1978; DOC 1968; DOC 1979). Other
information, including land-use data, are taken from ORNL's Geoecology
Data Base (01 sen, Emerson, and Nungesser 1980).
These data sources were generally viewed as being appropriate to
the risk assessment methodology, which is concerned with annual-average
or equilibrium conditions and population exposures. Of greater concern
are the methods used to estimate the terrestrial transfer coefficients
in the model. Because information in the literature on food-chain
transport characteristics for synfuels compounds is scarce, most
transport parameters must be estimated from structure/activity
relationships. Table 3 lists the soil/plant, milk, and beef transport
parameters currently used at ORNL and indicates which are derived from
literature sources and which have been estimated from
structure/activity relationships. The parameters derived from
literature sources represent an "average" value for an "average"
compound within the kAC. That is, if references were found for more
than one compound within an RAC, a geometric mean (exponential of the
mean of the log-transformed data) of all values found was used.
Typically, this procedure required determination of a geometric mean
value for each literature reference and determination of a geometric
mean of all of these mean values. However, in most cases, multiple
references or information on more than one compound within an RAC was
not available; thus, most parameters in Table 3 are based on only one
reference for each compound.
-------
43
Table 3. Transport parameters for synfuels food-chain exposure
assessment
, *d2 Kd3
RAC # Log Kow ] (s-1) (ml/g)
Fm 6 Ff 7
(d/kg) (d/kg)
9
10
11
12
13
14
15
16
17
18
19
20
2}
22
23
24
25
27
28
31
32
33
34
35
[-0.36]8
[2.00]
[-1.00]
[2.09]
[4.00]
L3.18J
[5.28J
[0.57]
0.19]
L2.65]
[2.96]
[1.97]
[1.55]
[0.90]
[2.52]
[-0.50]
[2.31]
[0.64]
[-0.92]
2
1
7
1
8
1
L5
3
7
4
1
2
L&
2
4
3
1
4
3
E-69
E-5
E-6
E-6
E-6
E-5
£-8]
E-6
E-6
E-6
E-5
E-6
E-8]
E-5
E-6
E-6
E-6
E-6
E-6
7
[1
[3
[1
0
[5
[7
2
4
[3
[4
1
7
3
2
6
2
2
3
[2
1
2
[7
[9
E-2
E+0]
E-2]
E+OJ
E+l]
E+0]
E+l]
E-l
E-l
E+0]
E+0]
E+0
E-l
E-l
E+0
E-2
E+0
E-l
t-2
E+2]
E+l
E+2
E+0]
E+2]
8 E+2
3 E+l
2 E+3
3 E+l
2 E+0
5 E+0
[2 E-l]
2 E+2
9 E+l
1 E+l
7 E+0
3 E+l
6 E+l
1 E+2
1 E+l
1 E+3
2 E+l
2 E+2
2 E+3
[4 E-2]
[9 E-l]
[6 E-2]
[6 E-l]
[5 E-2]
8
3
2
3
2
5
[3
2
9
1
7
3
6
1
1
1
2
2
2
[6
[2
[6
[2
[9
E+l
E+0
E+2
E+0
E-l
E-l
E-2]
E+l
E+0
E+0
E-l
E+0
E+0
E+l
E+0
E+2
E+0
E+l
E+2
E-3]
E-l]
E-2]
E-l]
E-3]
5 E-7
8 E-6
2 E-7
8 E-6
8 E-5
[5 E-5]
3 E-4
2 E-6
3 E-6
[4 E-5]
2 E-5
7 E-6
[2 £-4]
2 E-6
1 E-5
4 E-7
1 E-5
2 E-6
3 E-7
[6 E-5]
[5 E-4]
0 E-3]
[1 E-3]
[3 E-4]
5
7
2
8
7
[6
3
1
3
[1
2
7
[4
2
1
4
1
2
. 3
[2
3
6
[6
[3
E-6
E-5
E-6
E-5
E-4
E-4]
E-3
E-5
E-5
E-3]
E-4
E-5
E-4]
E-5
E-4
E-6
E-4
E-5
E-6
E-3]
E-l
E-3
E-4]
E-4]
1. Logarithm of the water/octanol partitioning coefficient (unitless)
2. The soil degraaation constant (1/s)
3. The soil/water distribution coefficient (mL/g)
4. The soil/plant bioaccumulation factor for vegetative plant parts
(unitless)
5. The soil/plant bioaccumulation factor for reproductive/storage
plant organs (unitless)
6. The milk transfer coefficient which relates daily intake to milk
concentration at equilibrium (d/kg)
7. The beef transfer coefficient which relates daily intake to beef
concentration at equilibrium or slaughter (d/kg)
8. Bracketed values have either been measured directly or are based
on experimental data reported in the literature
9. kead as 2 x 10"6.
-------
44
The distribution coefficient Kd is the ratio of solute
concentration in soil to that in water. This parameter is used to
predict leaching removal from root zone soil after the method of Baes
and Sharp (1983). More importantly, this parameter is used to predict
the soil/plant concentration factors B. and B. defined in Sect.
3.1. The prediction of K^ is based on the relationship given by
Briggs (1981),
log Kd = -0.99 + 0.53 (log KQW) . (3.1)
The estimation of B. is based on the relationship given by Baes
(1982) between Kd and Biy. Substitution of Eq. (3.1) into that
relationship gives
log Biv = 2.71 - 0.62 (log KQw) . (3.2)
The soil/plant CF for reproductive/storage plant organs B. is
assumed to be 0.1 Biv, based on the work of Baes et al. (1984b) on
inorganic compounds. This relationship also fits measured CFs for
benzo(a)pyrene (Kolar et al. 1975; Ellwardt 1977; Skcodich 1979).
Finally, the milk and beef transfer coefficients are derived by
incorporating into Kenega's (1980) relationship between bovine fat BCFs
and log K the average fat content of milk and beef (Spector 1956),
ow
and feed intake rates for milk cows and beef cattle (Shor et al.
1982). These substitutions yield for milk,
log Fm = -6.12 + 0.50 (log KQW) . (3.3)
and for beef,
log F = -5.15 + 0.50 (log K ) . (3.4)
I OW
-------
45
The soil degradation constant xd is assumed to be equal to the
air degradation constant if no measured values are available. This
approach is taken because no structure/activity relationships for soil
persistence have been published.
The workshop participants recognized many limitations in current
procedures for estimating transfer parameters. These limitations
include (1) the use of a single transfer coefficient to represent an
entire RAC, which may contain compounds having highly variable
transport properties (perhaps spanning four or five orders of
magnitude), (2) use of relationships derived from work done with
inorganic compounds for organic compounds, and (3) heavy dependence on
Io9 K«u, as tne measure of structure for the RAC. None of the
ow
participants was able to outline more appropriate procedures for
parameter determination, although there was general agreement on the
need to find better approaches. Also, there was a general consensus
that for a first-cut screening-level assessment, the parameter
estimation procedures were reasonable. Some participants suggested
that future assessments include uncertainty estimates based on either
analytical or numerical error propagation techniques, but the
estimation of probability density functions for the transport
parameters will be difficult in the absence of experimental data.
3.3 MAJOR LIMITATIONS ON EXISTING DATA AND METHODS
The workshop participants felt that one of the major shortcomings
of the current model and, indeed, the available data base is the lack
-------
46
of a food contamination pathway via foliar adsorption and translocation
to edible produce parts. This pathway would be important for both
direct deposition of atmospherically released pollutants and pollutants
dissolved in irrigation water. Once deposited, pollutants can remain
on leaf surfaces, be removed from the plant via weathering, or be
passed through the cuticle or stomata and be redistributed within the
plant. The first two processes are modeled in the current ORNL model,
but the latter is not.
Foliar absorption may be the primary route of entry for some
chemicals (e.g., herbicides and pesticides); however, information is
too sparse to develop predictive models based on structure/activity
relationships. Foliar interception of synfuels compounds, such as
gasses, particulates, or hydrosols, is primarily a function of leaf
surface morphology and structure. Adsorption through the cuticle can
be a significant route of entry for particulates and liquids and may be
related to lipid solubility, molecular size, and free energy. Stomatal
absorption occurs primarily with gasses and should be a function of
plant transpiration. The workshop participants recommended that the
foliar absorption/translocation pathway be included in the terrestrial
food-chain pathway model. It was suggested that absorption and
translocation be measured in experimental determinations of plant
uptake and distribution of synfuel compounds so that predictive models
based on the above relationships can be developed. Finally, it was
suggested that the existing data on pesticide foliar
absorption/translocation pathways be examined for structure/activity
relationships.
-------
47
A second area deemed important is the soil compartment. Here, two
areas were cited. First is the prediction of soil degradation kinetics
based on structure/activity relationships, and second is the
calculation of the synfuel compound concentration in the soil
solution. The former is considered an essential improvement of the
assessment methodology, and the second allows much closer communication
between modeler and experimentalist.
The current model assumes first-order kinetics for the degradation
of organics in soil. Data from field and laboratory studies of
organics (mostly pesticides) indicate that the assumption of
first-order kinetics is not always appropriate method in modeling
degradation in the soil. Rates of degradation in soil sometimes depend
on the concentrations in soil and on environmental variables such as
soil temperature, moisture, etc. Adaptation of soil microorganisms to
organics can increase degradation rates. In such cases, there may be a
significant departure from first-order kinetics.
An associated problem is the lack of a model for predicting the
soil degradation constants (x,). The current approach is to assume
the same value for x , as for x (the atmospheric degradation
Q a
constant). A good predictive model for soil degradation would
incorporate structure/activity relationships based on easily measured
chemical or structural properties (e.g., molecular size, functional
groups, and/or water solubility). The group recommended examination of
the pesticide data base for such structure/activity relationships.
-------
48'
The prediction of a soil solution concentration is an important
consideration from the standpoint of integrating results of
experimental plant uptake studies with modeling activities.
Experimental studies of root uptake of synfuel compounds must
necessarily be based on hydroponic solutions for practical reasons.
Constant measurable substrate concentrations must be maintained and
soil kinetics must be eliminated. Recognizing these requirements,
workshop participants felt that the model should be modified to
accommodate the experimental data, rather than vice versa.
Additionally, a soil solution submodel would allow the prediction of
the traditional soil/plant concentration factor from hydroponic data
and would also provide a means of assessing impact of synfuels
compounds on crops, because the majority of toxicological data is based
on hydroponic studies.
Another important area for further improvement in the model is the
consideration of the effects of food processing (especially cooking) on
human exposures. Exposure from leafy vegetables and exposed produce
would be expected to be significantly reduced by food processing,
including washing, trimming, and removal of outer exposed plant parts.
All of the attendees felt that, especially for organic compounds, the
effects of heating during cooking would tend to significantly reduce
human exposures because of thermal degradation. The EPA is currently
developing loss estimates for the various food preparation processes.
Preliminary results are highly variable and highly dependent on
individual preparation practices. Nevertheless, the unanimous opinion
-------
49
of the participants was that this consideration is an important one for
all meat and that most milk and vegetable pathways should be further
researched.
Other areas needing attention are (1) the inclusion of animal
products other than beef and milk into the model, (2) accounting for
differences in transfer coefficients resulting from livestock
management practice, (3) consideration of water and soil ingestion by
livestock (in addition to feed), (4) addition of irrigation water as a
source term to the terrestrial system, (5) capability to model acute
exposures and sensitive populations, and (6) estimation of uncertainty
associated with model predictions. These additional capabilities were
not unanimously considered to be essential by the workshop, although
one or more of the participants felt strongly about each. The reason
that these issues were not collectively considered to be as important
as the issues previously described is probably that their inclusion in
the model is not expected to change model results significantly, or, in
the case of the uncertainty issue, the uncertainty estimate could only
be tested through model validation, which could be difficult.
The food-chain model considers only cow's milk and beef pathways
in its current form. The workshop participants suggested that chicken,
pork, and egg pathways be considered as well. It was pointed out that
poultry and swine are significant components of the American diet.
Rupp (1980) found these two sources to constitute VI and 13% of the
total annual consumption of meat and other animal products,
respectively, for the over-18 age group. By contrast, milk (and milk
-------
50
products) and beef constitute 51 and 14%, respectively. A first
solution to account for poultry and swine would be to adjust the beef
consumption rates used in the risk assessment model upward to reflect
consumption of all meats. However, this approach assumes that beef
transfer coefficients are appropriate to poultry and swine. Eisele,
Traylor, and Schwarz (1983) demonstrate this not to be the case for
three organic compounds. In general, the ordering of transfer
coefficients was found to be poultry > swine > bovine. Therefore, a
better approach would be to include poultry and swine in the model.
The use of a single transfer coefficient for an animal product was
criticized because livestock management practice is thought to
significantly influence contaminant metabolism. Two examples are the
differences between broilers and egg-laying hens and between beef and
dairy cattle. Egg laying in poultry and milk production in cows can be
significant routes of depuration of ingested synfuels compounds because
these two excretory pathways are very important for these animals. In
the absence of these excretory pathways (broilers and beef cattle),
higher accumulations of organics in body tissues would be expected.
However, because this effect is not proven for synfuel compounds, it
was suggested that further research be done to examine this question.
If it is found that management practice does influence the measured
livestock transfer coefficients, the model should be modified
accordingly.
Also of concern in the modeling of livestock pathways is the
consumption of contaminated soil and water during grazing. It was
-------
51
pointed out that cattle consume roughly 0.5 to 1 kg of soil per day
while grazing pasture; poultry and swine, to a lesser extent, also
consume soil. If soil/plant CFs are low (as indeed they are for most
synfuel compounds as currently modeled), then soil may be an even more
important pathway than ingestion of feed. To a lesser extent,
contaminated water may also be considered an ingestion source. These
additional contamination sources may be more important in certain areas
of the United States than in others, depending on livestock management
practices. The workshop participants could not arrive at a consensus
as to whether incorporation of these pathways would significantly alter
model results. In the interest of conservatism, however, it was
generally agreed that provisions for these pathways should be examined
in the food-chain model.
Another recommended addition to the food-chain model is the
contribution of contamination of agricultural plants via irrigation
water. In many parts of the country, irrigation waters are derived
from groundwater supplies, and organics deposited on or in soil could
reach these reservoirs. In other parts of the country, surface waters
contaminated via atmospheric, aquatic, and surface runoff inputs, are
used for irrigation of crops. It was suggested that simple bounding
analyses be performed to determine the importance of this source term.
If this is found to be a critical pathway, data collection and model
modifications should be undertaken. Inclusion of this pathway is
appealing because it would link the terrestrial and aquatic transport
models. If irrigation proved significant, the addition of a
groundwater transport model would also be needed.
-------
52
The workshop addressed issues that went beyond the scope of the
food-chain model itself, including the capability to assess acute
exposures and sensitive populations and to estimate uncertainties. The
current assessment strategy addresses chronic population exposures
after 35 years of synfuel plant operation. Because these facilities
have a projected lifetime of 30 years, the assessment scenario now used
should provide maximum average exposure estimates. However,
participants recognized the need to address acute exposures from
intermittent pollutant releases and exposures to sensitive populations
(nursing mothers, children, vegetarians, individuals growing their own
food, etc.). Acute exposures and exposures to sensitive populations
should eventually be addressed in both the transport modeling and in
the determination of risk estimates. Currently, simple bounding
estimates to address these needs should be performed in both the
transport and risk methodologies.
Finally, although sensitivity analyses have been performed on the
food-chain model, numeric determination of model uncertainties have not
been performed. Stochastic variability could be examined using
available numerical and analytic tools [e.g., ORNL's Monte Carlo
techniques (Gardner and O'Neill 1983) and Carnegie-Mellon University's
DEMOS model (Henrion and Morgan, in press)]. Model results would then
be presented as distributions rather than as single-value estimates.
These techniques allow important contributors to overall model
uncertainty to be specified as a guide to further research. Also,
methods for examining systematic error need to be developed. Of
-------
53
course, the best method to address systematic error would be model
validation. The workshop concluded that validation is really the only
method not only to ensure that the assessment model is both appropriate
and accurate but also to specify definitively the uncertainty
associated with model predictions. Validation would also determine
whether the various concerns about the current model expressed above
are important.
-------
54
4. REFERENCES
Baes, C. F., III. 1982. "Prediction of radionuclide Kd values from
soil-plant concentration ratios." Trans. Amer. Nucl. Soc.
41:53-54.
Baes, C. F., Ill and R. D. Sharp. 1983. "A proposal for estimation of
soil leaching and leaching constants for use in assessment
models." J. Environ. Qual. 12(l):17-28.
Baes, C. F., Ill, r. D. Sharp, A. L. Sjoreen, and 0. W. Hermann.
1984a. TERRA: A Computer Code for Calculation of the Transport
of Environmentally Released Radionuclides Through Agriculture.
ORNL-5785. Martin Marietta Energy Systems, Inc., Oak Ridge
National Laboratory.
Baes, C. F., Ill, R. D. Sharp, A. L. Sojoreen, and R. W. Shor. 1984b.
A Review and Analysis of Parameters for Assessing Transport of
Environmentally Released Radionuclides Through Agriculture.
ORNL-5786. Martin Marietta Energy Systems, Inc., Oak Ridge
National Laboratory.
Baes, C. F., Ill, M. A. Bell, P. A. Mason, and R. D. Sharp. in
preparation. TERREX: A Computer Code for Calculating Foodchain
Transport of Exotic Contaminants. ORNL-5832. Martin Marietta
Energy Systems, Inc., Oak Ridge National Laboratory.
Breck, J. E. In preparation. "Modeling bioaccumulation in fishes:
A comparison of synfuels and long-lived environmental
contaminants." Environ. Int.
-------
55
Briggs, 6. G. 1981. "Theoretical and experimental relationships
between soil adsorption, octanol-water partition coefficients,
water solubilities, bioconcentration factors, and the parachor."
J. Agric. Food Chem. 29:1050-1059.
Brown, M. P., 0. J. A. Mclaughlin, 0. M. O'Connor, and K. Wyman.
1982. "A mathematical model of PCB bioaccumulation in plankton."
Ecol. Modelling 15:29-47.
Bruggeman, W. A., L. B. J. M. Martron, D. Kooiman, and 0. Hutzinger.
1981. "Accumulation and elimination kinetics of di-, tri-, and
tetra chlorobiphenyls by goldfish after dietary and aqueous
exposure." Chemosphere 10:811-832.
Connolly, J. P. and R. Tonelli. Modeling kepone in the striped bass
food chain of the James River estuary. Estuarine, Coastal and
Shelf Science (in press).
Connolly, J. P., and R. Tonelli. 1984.
Donigian, A. S., Jr., and S. M. Brown. 1983. A workshop on water
modeling needs and available techniques for synfuels risk
assessment. Prepared for Georgia Institute of Technology,
Subcontract No. l-E-26-630. Anderson-Nichols & Co., Inc., Palo
Alto, California.
Eisele, G. R., T. D. Traylor, and 0. J. Schwarz. 1982. Food Chain
Transport of Synfuels Progress Report. Oak Ridge Associated
Universities, Oak Ridge, Tennessee (April, 1982).
Ellgehausen, H., J. A. Guth, and H. 0. Esser. 1980. Factors
determining the bioaccumulation potential of pesticides in the
individual compartments of aquatic food chains. Ecotoxicology and
Environmental Safety 4:134-157.
-------
56
Ellwardt, P. C. 1977. "Variation in content of polycyclic aromatic
hydrocarbons in soil and plants by using municipal waste composts
in agriculture." IN Soil Organic Matter Studies II.
IAEA-SM-211/31. pp. 291-298.
Fagerstrom, T., and B. Asell. 1973. Methyl mercury accumulation in an
aquatic food chain. A model and some implications for planning.
Ambio 2:164-171.
Fagerstrom, T., B. Asell, and A. Jernelov. 1974. Model for
accumulation of methyl mercury in northern pike Esox lucius.
OIKOS 26:14-20.
Farrington, J. w., E. D. Goldberg, K. W. Risebrough, J. H. Martin, and
V. T. Bowen. 1983. U.S. "Mussel Watch" 1976-1978: An overview
of the trace-metal, DDE, PCB, hydrocarbon, and artificial
radionuclide data. Environmental Science and Technology
17:490-496.
Gardner, k. H. and R. V. O'Neill. 1983. "Parameter uncertainty and
model predictions: A review of Monte Carlo results."
pp. 245-257, In M. B. Beck and G. van Straten (eds.) Uncertainty
and Forecasting of Water Quality, Springer-Verlag, Berlin.
Goldstein, A., L. Aronow, and S. M. Kalmon. 1974. Principals of drug
action: The basis of Pharmacology. Second Ed. John Wiley &
Sons, New York.
Guiney, P. D., R. E. Peterson, M. J. Melancon, Jr., and J. J. Lech.
1977. "The distribution and elimination of 2, 5, 2', 5'-[14C]
tetrachlorobiphenyl in rainbow trout (Salmo gairdneri)." Toxicol.
Appl. Pharmacol. 39:329-338.
-------
57
Guiney, P. D., M. J. Melancon, Jr., J. J. Lech, and R. E. Peterson.
1979. "Effects of egg and sperm maturation and spawning on the
distribution and elimination of a polychlorinated biphenyl in
rainbow trout (Salmo gairdneri)." Toxicol. Appl. Pharmacol.
47:261-272.
Henrion, M. and M. G. Morgan. A Computer Aid for Risk and other Policy
Analysis. J. Risk Analysis (in press).
Kenaga, E. E. 1980. "Correlation of Bioconcentration factors of
chemicals in aquatic and terrestrial organisms with their physical
and chemical properties." Environ. Sci. Tech. 14(5):553-556.
Kenaga, E. £., and C. A. I. Goring. 1980. "Relationship between water
solubility, soil sorption, octanol-water partitioning, and
concentration of chemicals in biota." pp. 78-115. IN
d. E. Eaton, P. R. Parrish, and A. C. Hendricks (eds.), Aquatic
Toxicology. ASTM STP 707. American Society of Testing and
Materials, Philadelphia.
Kinsella, J. E., J. L. Shimp, J. Mai, and J. Weihrauch. 1977. Fatty
acid content and composition of freshwater finfish. J. Amer. Oil
Chem. Soc. 54:424-429.
Kolar, L., R. Ledvina, J. Ticha, and F. Hanus. 1975. "Contamination
of soil, agricultural crops, and vegetables by 3,4-benzpyrene in
the vicinity of Ceske Budejovice." Ceskosolvenska Hygiena
20(3):135-139.
Mackay, D. 1982. "Correlation of bioconcentration factors." Environ.
Sci. Technol. 16:274-278.
-------
58
Malins, D. C., T. K. Collier, and H. R. Sanborn. 1979. "Disposition
and metabolism of aromatic hydrocarbons in marine organisms." pp.
57-75. IN Pesticide and Xenobiotic Metabolism in Aquatic
Organisms. American Chemical Society.
Moghissi, A. A., and G. J. Foley. 1982. Assessment of environmental
impact of coal-conversion and oil-shale technologies. pp.
665-673. IN On Health Impacts of Different Sources of Energy.
International Atomic Energy Agency, Vienna, Austria.
Moskowitz, P. D., S. C. Morris, H. Fischer, and H. C. Thode, Jr.
November 1983. Synthetic fuel plants: potential cancer risks to
public health. Biomedical and Environmental Assessment Division,
Department of Applied Science, Brookhaven National Laboratory,
Associated Universities, Inc.
Norstrom, R. J., A. E. McKinnon, and A. S. W. deFreitas. 1976. "A
bioenergetics-based model for pollutant accumulation by fish.
Simulation of PCB and methyl mercury residue levels in Ottawa
River yellow perch (Perca flavescens)." J. Fish. Res. Board Can.
33:248-267.
Oliver, B. G., and A. J. Niimi. 1983. "Bioconcentration of
chlorobenzenes from water by rainbow trout: correlations with
partition coefficients and environmental residues." Environ. Sci.
Technol. 17:287-291.
Olsen, R. J., C. J. Emerson, and M. K. Nungesser. 1980. GEOECOLOGY:
A County-Level Environmental Data Base for the Conterminous United
States. ORNL/TM-7351. Environmental Sciences Division
Publication No. 1537. Oak Ridge National Laboratory. September,
1980.
-------
59
Reichert, W. L., and D. C. Malins. 1974. "Interaction of mercurials
with salmon serum lipoproteins." Nature 247:569-570.
Rottiers, D. V., and R. M. Tucker. 1982. Proximate composition and
caloric content of eight Lake Michigan fishes. Fish and Wildlife
Service. Tech. Pap. 108, U.S. Department of the Interior,
Washington, D.C.
Ruffner, J. A. 1978. Climates of the States, vols. 1 and 2. Gale
Research Company. Book Tower, Detroit, Michigan.
Rupp, E. M. 1980. "Age dependent values of dietary intake for
assessing human exposures to environmental pollutants." Health
Phys. 39(2):151-163.
Saarikoski, J., and M. Viluksela. 1982. "Relation between
physicochemical properties of phenols and their toxicity and
accumulation in fish." Ecotoxicology and Environmental Safety
6:501-512.
Schanker, L. S. 1960. "On the mechanism of absorption of drugs from
the gastrointestinal tract." Journal of Medicinal and
Pharmaceutical Chemistry 2(4):343-359.
Shkodich, P. E. and Y. A. Litvinov. 1979. "Content of polycyclic
aromatic hydrocarbons in potato tubers and head cabbage grown in a
zone affected by organic synthesis industry." Rast. Kyim.
Kantserogeny (Simp.), 1st, 1979, pp. 142-144.
Shor, R. W., Baes, C. F., Ill, and R. D. Sharp. 1982. Agricultural
Production in the United States by County: A Compilation of
Information from the 1974 Census of Agriculture for Use in
Terrestrial Food Chain Transport and Assessment Models. ORNL-5768.
-------
60
Sidwell, V. D., P. R. Foncannon, N. S. Moore, and J. C. Bonnet. 1974.
Composition of the edible portion of raw (fresh or frozen)
crustaceans, finfish, and mollusks. I. Protein, fat, moisture,
ash, carbohydrate, energy value, and cholesterol. Mar. Fish. Rev.
36:21-35.
Southworth, G. R., C. C. Keffer, and J. J. Beauchamp. 1980. Potential
and realized bioconcentration. A comparison of observed and
predicted bioconcentration of azaarenes in the fathead minnow
(Pimephales promelas). Environmental Sciences and Technology
14:1529-1531.
Spacie, A., P. F. Landrum, and G. J. Leversee. 1983. "Uptake,
depuration, and biotransformation of anthracene and benzo(a)pyrene
in bluegill sunfish." Ecotoxicology and Environmental Safety
7:330-341.
Spacie, A., and J. L. Hamelink. 1982. "Alternative models for
describing the bioconcentration of organics in fish."
Environmental Toxicology and Chemistry 1:309-320.
Spector, W. S. 1956. Handbook of Biological Data. W. B. Saunders
Co., Philadelphia, p. 187.
Thomann, R. V. 1981. "Equilibrium model of fate of microcontaminants
in diverse aquatic food chains." Can. J. Fish. Aquat. Sci.
38:280-296.
Thomann, R. V., and J. P. Connolly. 1984. "Model of PCB in the Lake
Michigan lake trout food chain." Environ. Sci. Tech. 18:65-71.
Thomann, R. V., and J. P. St. John. 1979. "The fate of PCBs in the
Hudson River ecosystem." Ann. New York Acad. Sci. 320:610-629.
-------
61
Trabalka, J. R., and C. T. Garten, Jr. 1982. Development of
predictive models for xenobiotic bioaccumulation in terrestrial
ecosystems. ORNL-5869. Oak Ridge National Laboratory, Oak Ridge,
Tennessee. 249 pp.
Travis, C. C., C. F. Baes III, L. W. Barnthouse, E. L. Etnier,
G. A. Holton, B. D. Murphy, G. P. Thompson, G. W. Suter II, and
A. P. Watson. 1983. Exposure Assessment Methodology and
Reference Environments for Synfuel Risk Analysis. ORNL/TM-8672.
Oak Ridge National Laboratory, Oak Ridge, Tennessee.
U.S. Department of Commerce, Bureau of the Census. 1977. The 1974
Census of Agriculture, United States Summary and State Data. Vol.
1, Part 51. U.S. Government Printing Office, Washington, D.C.
U.S. Department of Commerce. 1968. Climatic Atlas of the United
States. Environmental Science Services Administration,
Environmental Data Service. Reprinted by the National Oceanic and
Atmospheric Administration. 1977.
U.S. Environmental Protection Agency (U.S. EPA). 1980. Water Quality
Criteria Documents; Availability. Federal ; Register
45(231):79318-79379.
U.S. Department of Commerce. 1979. Comparative Climatic Data for the
United States through 1978. Environmental Data and Information
Service, National Climatic Center, Asheville, N.C.
U.S. Nuclear Regulatory Commission. 1977. Regulatory Guide 1.109,
Calculation of Annual Doses to Man from Routine Releases of
Reactor Effluents for the Purpose of Evaluating Compliance with 10
CFR Part 50, Appendix I (Revision 1). Office of Standards
Development, U.S. Nuclear Regulatory Commission, Washington, D.C.
-------
62
Vanderploeg, H. A., 0. C. Parzyck, W. H. Wilcox, J. R. Kercher, and
S. V. Kaye. 1975. Bioaccumulation factors for radionuclides in
freshwater biota. ORNL-5002. Oak Ridge National Laboratory, Oak
Ridge, Tennessee. 222 p.
Veith, G. D., K. J. Macek, S. R. Petrocelli, and J. Carroll. 1980.
"An evaluation of using partition coefficients and water
solubility to estimate bioconcentration factors for organic
chemicals in fish." pp. 116-129. IN J. G. Eaton, P. R. Parrish,
and A. C. Hendricks (eds.), Aquatic Toxicology, ASTM STP 707.
American Society for Testing and Materials, Philadelphia,
Pennsylvania.
Weininger, D. 1978. Accumulation of PCBs in lake trout in Lake
Michigan. Ph.D. dissertation (Water Chemistry), University of
Wisconsin-Madison, Madison, Wisconsin. 232 p.
-------
63
APPENDIX A. AGENDA
-------
65
WORKSHOP ON FOOD CHAIN MODELING FOR RISK ANALYSIS
Capitol Hill Hotel - Washington, D.C.
March 22-24, 1983
AGENDA
Tuesday, March 22, 1983 - All Day
Welcome: Mel Carter, Georgia Inst. of Tech.
Introduction to the Workshop: C. Fred Baes III, ORNL
Charge to Workshop Participants: Alan Moghissi, EPA
Presentations:
C. Fred Baes III, ORNL
James Breck, ORNL
Jim Falco, EPA
Jerry Eisele, ORAU/CARL
Craig McFarlane, EPA
Chuck Garten, ORNL
John Connolly, Manhattan College
John Nagy, Brookhaven National Laboratory
Paul Moskowitz, Brookhaven National Laboratory
F. Owen Hoffman, ORNL
Break into working groups: Aquatic and Terrestrial
-------
66
Wednesday, March 23 - All Day
Working group discussions and report writing
Topics discussed by each group:
1. Significance of potential uptake pathways
2. Lessons from radionuclide assessment
3. Extrapolation from laboratory to field
4. Major uncertainties in food chain assessment
Thursday, March 24 - A.M.
Presentation, discussion, and review of draft recommendations
1. Footi chain models best suited to synfuels risk analysis
2. Data sources and parameter estimation methods best suited
to synfuels risk analysis
3. Major limitations on existing data ano methods
Noon: Adjournment
-------
67
APPENDIX B. LIST OF PARTICIPANTS
-------
69
WORKSHOP ON FOOD CHAIN MODELING FOR RISK ANALYSIS
Capitol Hill Hotel - Washington, D.C.
March 22-24, 1983
LIST OF PARTICIPANTS
AQUATIC
Jim Breck
Environmental Sciences Division
Oak Ridge National Laboratory
P. 0. Box X
Oak Ridge, TN 37831
(616) 574-7263/FTS 624-7263
Paul Cho
Health £ Environmental Risk Analysis
Program
U.S. Department of Energy
Washington, DC 20545
(202) 353-5897/FTS 233-5897
John Connolly
Environmental Engineering & Science
Program
Manhattan College
4513 Manhattan College Parkway
Riverdale, NY 10471
(212) 920-0276
Anthony S. W. deFreitas
National Research Council of Canada
Atlantic Research Laboratory
1411 Oxford Street
Canada B3H 3£1
(902) 426-8263
Jim Falco
Environmental Protection Agency
ORD
401 M. Street
Washington, DC 20460
(202) 382-7327/FTS 382-7327
F. 0. Hoffman
Health £ Environmental Research
Division
Oak Ridge National Laboratory
P. 0. Box X
Oak Ridge, TN 37831
(615) 576-2118/FTS 626-2118
John Nagy
Brookhaven National Laboratory
Associated Universities, Inc.
Upton, NY 11973
(516) 282-2667/FTS 666-2667
Anne Spacie
Department of Forestry & Natural
Resources
Puraue University
West Lafayette, IN 47907
(317) 494-3621
-------
70
WORKSHOP ON FOOD CHAIN MODELING FOR RISK ANALYSIS
Capitol Hill Hotel - Washington, D.C.
March 22-24, 1983
LIST OF PARTICIPANTS
TERRESTRIAL
C. F. Baes, III
Environmental Sciences Division
Oak Ridge National Laboratory
P. 0. Box X
Oak Ridge, TN 37831
(615) 576-2137/FTS 626-2137
Jerry Eisele
Oak Ridge Associated Universities
Comparative Animal Research
Laboratory
Oak Ridge, TN 37831
(615) 576-4081/FTS 626-4081
Chuck Garten
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, TN 37831
(615) 574-7355/FTS 624-7355
Russell Kinerson
Exposure Evaluation Division (TS798)
Office of Toxic Substances
U.S. Environmental Protection Agency
401 M. Street, SW
Washington, DC 20460
(202) 382-3929/FTS 382-3929
Craig McFarlane
Corvallis Environmental Research
Laboratory
200 SW 35th Street
Corvallis, OR 97333
(505) 757-4670/FTS 420-4670
Paul D. Moskowitz
Brookhaven National Laboratory
Associated Universities, Inc.
Upton, NY 11973
(516) 282-2017/FTS 666-2017
Curtis Travis
Health & Safety Research Division
Oak Ridge National Laboratory
P. 0. Box X
Oak Ridge, TN 37831
(615)-576-2107/FTS 626-2107
Melvin W. Carter
School of Nuclear Engineering
and Health Physics
Georgia Institute of Technology
Atlanta, GA 30332
(404) 894-3745
-------
71
ORNL-6051
INTERNAL DISTRIBUTION
1.
2-6.
3.
4.
5.
6-10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
S. I.
C. F.
L. W.
S. M.
M. A.
J. E.
R. 0.
C. C.
K. E.
W. F.
R. H.
C. T.
C. W.
F. 0.
S. V.
D. C.
Auerbach
Baes III
Barnthouse
Bartell
Bell
Breck
Chester
Coutant
Cowser
Furth
Gardner
Garten
Gehrs
Hoffman
Kaye
Kocher
21. P. R. O'Donnell
22. R. V. O'Neill
23. D. C. Parzyck
24. D. E. Reichle
25. G. W. Suter II
26. C. C. Travis
27. P. J. Walsh
28. A. P. Watson
29. J. P. Witherspoon
30. H. E. Zittel
31. Central Research Library
32-41. ESD Library
42-43. Laboratory Records Dept.
44. Laboratory Records, ORNL-RC
45. ORNL Y-12 Technical Library
46. ORNL Patent Office
EXTERNAL DISTRIBUTION
47. J. Frances Allen, Science Advisory Board, Environmental
Protection Agency, Washington, DC 20460
48. Richard Balcomb, TS-769, Office of Pesticide Programs,
Environmental Protection Agency, 401 M Street, SW,
Washington, DC 20460
49. Nathaniel F. Barr, Office of Health and Environmental
Research, Department of Energy, Washington, DC 20545
50. Colonel Johan Bayer, USAF OHEL, Brook AFB, TX 78235
51. Frank Benenati, Office of Toxic Substances, Environmental
Protection Agency, 401 M Street, SW, Washington, DC 20460
52. K. Biesinger, Environmental Protection Agency, National Water
Quality Laboratory, 6201 Congdon Boulevard, Duluth, MN 55804
53. J. D. Buffington, Director, Office of Biological Services,
U.S. Fish and Wildlife Services, 1730 K Street, NW,
Washington, DC 20240
54. J. Cairns, Center for Environmental Studies, Virginia
Polytechnic Institute and State University, Blacksburg,
VA 24061
55. J. Thomas Callahan, Associate Director, Ecosystem Studies
Program, Room 336, 1800 G Street, NW, National Science
Foundation, Washington, DC 20550
56. Melvin W. Carter, Georgia Institute of Technology, School of
Nuclear Engineering and Health Physics, Atlanta, GA 30332
57. Paul Cho, Health and Environmental Risk Analysis Program,
HHAD/OHER/ER, Department of Energy, Washington, DC 20545
58. John Connolly, Environmental Engineering & Science Program,
Manhattan College, 4513 Manhattan College Parkway, Riverdale,
NY 10471
-------
72
59. C. E. Cashing, Ecosystems Department, Battelle-Northwest
Laboratories, Richland, WA 99352
60. R. C. Dahlman, Carbon Cycle Program Manager, Carbon Dioxide
Research Division, Office of Energy Research, Room J-311,
Ek-12, Department of Energy, Washington, DC 20545
61. Anthony S. W. deFreitas, National Research Council of Canada,
Atlantic Research Laboratory, 1411 Oxford Street, Canada B3H
3Z1
62. Sidney Draggan, Ecologist-Policy Analyst, Division of Policy
Research and Analysis, National Science Foundation,
Washington, DC 20550
63. Charles W. Edington, Office of Health and Environmental
Research, Department of Energy, Washington, DC 20545
64. Gerhard R. Eisele, Comparative Animal Research Laboratory,
1299 Bethel Valley Roaci, Oak Ridge, TN 37830
65. Jim Falco, Environmental Protection Agency, ORU, 401 M.
Street, Washington, DC 20460
66. David Flemar, Environmental Protection Agency, Washington,
DC 20460
67. G. J. Foley, Office of Environmental Process and Effects
Research, U.S. Environmental Protection Agency, 401 M Street,
SW, RD-682, Washington, DC 20460
68. Ralph Franklin, Office of Health and Environmental Research,
Department of Energy, Washington, DC 20545
69. David Friedman, Hazardous Waste Management Division (WH-565),
Office of Solid Waste, Environmental Protection Agency,
401 M Street, SW, Washington, DC 20460
70. Norman R. Glass, National Ecological Research Laboratory,
Environmental Protection Agency, 200 SW 35th Street,
Corvallis, OR 97330
71. Leonard Hamilton, Department of Energy and Environment,
Brookhaven National Laboratory, Upton, NY 11973
72. J. W. Huckabee, Project Manager, Environmental Assessment
Department, Electric Power Research Institute, 3412 Hi 11 view
Avenue, P.O. Box 10412, Palo Alto, CA 94303
73. Norbert Jaworski, Environmental Research Laboratory-Duluth,
6201 Congdon Boulevard, Duluth, NM 55804
74. Donald Johnson, Gas Research Institute, 8600 West Bryn Mawr
Avenue, Chicago, IL 60631
75. George Y. Jordy, Director, Office of Program Analysis, Office
of Energy Research, ER-30, G-226, U.S. Department of Energy,
Washington, DC 20545
76. Russell Kinerson, Exposure Evaluation Division (TS798),
Office of Toxic Substances, U.S. Environmental Protection
Agency, 401 M Street, SW, Washington, DC 20460
77. Library, Bureau of Sport Fisheries and Wildlife, Department
of the Interior, Washington, DC 20240
78. Library, Food and Agriculture, Organization of the United
Nations, Fishery Resources and Environment Division, via
delle Termi di Caracal la 001000, Rome, Italy
-------
73
79. Library, Western Fish Toxicology Laboratory, Environmental
Protection Agency, Corvallis, OR 97330
80. Ronald R. Loose, Department of Energy, Washington, DC 20545
81. Helen McCammon, Director, Ecological Research Division, Office
of Health and Environmental Research, Office of Energy
Research, MS-E201, ER-75, Room E-233, Department of Energy,
Washington, DC 20545
82. Craig McFarlane, Corvallis Environmental Research Laboratory,
200 SW 35th Street, Corvallis, OR 97333
83-132. Alan Moghissi, Environmental Protection Agency, MC RD-682,
401 M Street, SW, Washington, DC 20460
133. Dario M. Monti, Division of Technology Overview, Department of
Energy, Washington, DC 20545
134. Harold A. Mooney, Department of Biological Sciences, Stanford
University, Stanford, CA 94305
135. Sam Morris, Brookhaven National Laboratory, Associated
Universities, Inc., Upton, NY 11973
136. Paul D. Moskowitz, Brookhaven National Laboratory, Associated
Universities, Inc., Upton, NY 11973
137. J. Vincent Nabholz, Health and Environmental Review Division,
Office of Toxic Substances, Environmental Protection Agency,
401 M Street, SW, Washington, DC 20460
138. John Nagy, Brookhaven National Laboratory, Associated
Universities, Inc., Upton, NY 11973
139. Barry E. North, Engineering-Science, 10 Lakeside Lane, Denver,
CO 80212
140. Goetz Oertel, Waste Management Division, Department of Energy,
Washington, DC 20545
141. Office of Toxic Substances, Environmental Protection Agency,
401 M Street, SW, Washington, DC 20460
142. F. L. Parker, College of Engineering, Vanderbilt University,
Nashville, TN 37235
143. G. P. Patil, Statistics Department, 318 Pond Laboratory,
Pennsylvania State Universtiy, University Park, PA 16802
144. Ralph Perhac, Electric Power Research Institute, 3412 Hillview
Avenue, P.O. Box 10412, Palo Alto, CA 94304
145. J. C. Randolph, School of Public and Environmental Affairs,
Indiana University, Bloomington, IN 47405
146. Irwin Remson, Department of Applied Earth Sciences, Stanford
University, Stanford, CA 94305
147. Abe Silvers, Electric Power Research Institute, P.O. Box
10412, Palo Alto, CA 94303
148. David Slade, Office of Health and Environmental Research,
Department of Energy, Washington, DC 10545
149. Anne Spacie, Department of Forestry & Natural Resources,
Purdue University, West Lafayette, IN 47907
150. R. J. Stern, Director, Office of Environmental Compliance,
MS PE-25, FORRESTAL, U.S. Department of Energy, 1000
Independence Avenue, SW, Washington, DC 20585
-------
74
151. Frank Swanberg, Jr., U.S. Nuclear Regulatory Commission,
Washington, DC 20555
152. The Institute of Ecology, 1401 Wilson Blvd., Box 9197,
Arlington, VA 22209
153. Burt Vaughan, Battelle-Pacific Northwest Laboratory,
Richland, WA 99352
154. John Walker, Assessment Division, TS 778, Office of Toxic
Substances, U.S. Environmental Protection Agency, 401 M
Street, SW, Washington, DC 20460
155. Robert L. Watters, Ecological Research Division, Office of
Health and Environmental Research, Office of Energy Research,
MS-E201, ER-75, Room F-226, Department of Energy, Washington,
DC 20545
156. D. E. Weber, Office of Energy, Minerals, and Industry,
Environmental Protection Agency, Washington, DC 20460
157. A. M. Weinberg, Institute of Energy Analysis, Oak Ridge
Associated Universities, Oak Ridge, TN 37830
158. Raymond G. Wilhour, Chief, Air Pollution Effects Branch,
Corvallis Environmental Research Laboratory, U.S.
Environmental Protection Agency, 200 SW 35th Street,
Corvallis, OR 97330
159. Ted Williams, Division of Policy Analysis, Department of
Energy, Washington, DC 20545
160. Frank J. Wobber, Ecological Research Division, Office of
Health and Environmental Research, Office of Energy Research,
MS-E201, Department of Energy, Washington, DC 20545
161. Gordon M. Wolman, The Johns Hopkins University, Department of
Geography and Environmental Engineering, Baltimore, MD 21218
162. Bill Wood, TS-798, U.S. Environmental Protection Agency,
401 M Street, SW, Washington, DC 20460
163. Robert W. Wood, Director, Division of Pollutant
Characterization and Safety Research, Department of Energy,
Washington, DC 20545
164. R. Wyzga, Manager, Health and Environmental Risk Department,
Electric Power Research Institute, P.O. Box 10412, Palo Alto,
CA 94303
165. Office of Assistant Manager for Energy Research and
Development, Oak Ridge Operations, P. 0. Box E, U.S.
Department of Energy, Oak Ridge, TN 37831
166-192. Technical Information Center, Oak Ridge, TN 37831
------- |