Ecotoxicological Test Systems Proceedings Of A Series Of Workshops


ORNL
Operated by
Union Carbide Corporation
for the Department of Energy
Oak Ridge, Tennessee 37830
OR NL-5709
EPA
           United States
           Environmental Protection
           Agency-
Office of Toxic Substances
Washington DC 20460
           Toxic Substances
EPA 560/6-81 004
Jurie 1981
           Ecotoxicological
           Test Systems

           Proceedings of a Series
           of Workshops

-------
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 A15 Microfiche A01
This report was prepared as an account of work sponsored Dy an agency of the
United States Government Neither the United States Government nor any agency
•hereof nor any of their employees makes any warranty, express or implied, or
assumes ary iega> liability or -esponsib'lity 'or the accuracy completeness, or
usefuTiess of any information apparatus product or process disclosed or
rep'ebei'ts that its jse would not infringe privately owned rights Reference herein
to any speC'ficcon.mercia, pro, UK,! ',.<'•( tSii, or service bv trade name trademaiK,
o'herw :,e jotjs not necessan.y constitute or imply ts
mme'idahO'i 01 ta/onnq b/ the United States Government :u
' *hereot i he nf-AS anu op'0
-------
ORNL-5709
EPA-560/6-81-004
June 1981
Contract No. W-7405-eng-26

ECOTOXICOLOGICAL TEST SYSTEMS

PROCEEDINGS OF A SERIES OF WORKSHOPS

Edited by

Anna S. Mammons

Contributors

J. M. Giddings
G. W. Suter, II
L. W. Barnthouse

Environmental Sciences Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee 37830

ESD Publication No. 1778

Date Published: July 1981

Interagency Agreement No. EPA 78-D-X0387

Project Officer

J. Vincent Nabholz
Health and Environmental Review Division
Office of Toxic Substances
Washington, D.C. 20460
Prepared for

Office of Toxic Substances
U.S. Environmental Protection Agency
Washington, D.C. 20460
OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee 37830
Operated by
UNION CARBIDE CORPORATION
for the
DEPARTMENT OF ENERGY
-------
DISCLAIMER
This document has been reviewed and approved for publication by
the Office of Toxic Substances, U.S. Environmental Protection Agency.
Approval does not signify that the contents necessarily reflect the
views and policies of the Environmental Protection Agency, nor does
the mention of trade names or commercial products constitute
endorsement or recommendation for use.
-------
FOREWORD
The scientific disciplines of ecology and environmental
toxicology have not been communicating adequately with each
other, to the detriment of both. Ecologists are often
falling short when it comes to applying the theory and
findings of their relatively young science in useful practice
to meet society's needs for assessment of the environmental
impacts of toxic pollutants. Environmental toxicologists
are increasingly having difficulty in trying to convince
society's decision makers what the results of their test
methodologies in simple systems really mean in a complex,
highly interactive ecological world.

These workshops take a step toward marrying some of the
concepts of these two scientific disciplines. At the request
of the Environmental Protection Agency's Office of Toxic
Substances, the Environmental Sciences Division of Oak Ridge
National Laboratory has convened this series of workshops to
review and evaluate potential techniques for studying ecological
effects of toxic chemicals in systems that transcend the
practicable but oversimplified conditions of most currently
used toxicological test systems.

EPA intends to use this study, and companion efforts,
to help guide our future attempts to bring about better
synergy between ecology and environmental toxicology in our
implementation of the Toxic Substances Control Act.
James J. Reisa, Ph.D.
Associate Deputy Assistant Administrator
for Toxic Substances
U.S. Environmental Protection Agency
m
-------
ABSTRACT
A series of six workshops was conducted by the Environmental
Sciences Division, Oak Ridge National Laboratory, to identify
laboratory methods and data evaluation techniques for predicting the
environmental effects of chemical substances. Methods were evaluated
for their potential for standardization and for use in the ecological
hazard and risk assessment processes under the Toxic Substances
Control Act. The workshops addressed assessment and policy
requirements of multispecies toxicology test procedures, mathematical
models useful in hazard and risk assessments, and methods for
measuring effects of chemicals on terrestrial and aquatic population
interactions and ecosystem properties. The workshops were primarily
used as a mechanism to gather information about research in progress.
This information was part of the data base used to prepare a critical
review of laboratory methods for ecological toxicology.

This report was submitted in partial fulfillment of Interagency
Agreement No. EPA 78-D-X0387 between the Department of Energy and the
U.S. Environmental Protection Agency.
-------
CONTENTS

Page

FOREWORD 111

ABSTRACT v

TABLES xi

FIGURES xii

ACKNOWLEDGMENTS xiii

1. INTRODUCTION 1

2. CONCLUSIONS AND RECOMMENDATIONS 5

2.1 Introduction 5
2.2 Terrestrial Test Systems 6
2.2.1 Population Interactions 6
2.2.2 Ecosystem Properties 6
2.3 Aquatic Test Systems 7
2.3.1 Population Interactions 7
2.3.2 Model Ecosystems 8
2.4 Mathematical Models 9

3. ASSESSMENT AND POLICY REQUIREMENTS OF MULTISPECIES
TOXICOLOGY TESTING PROCEDURES 11

3.1 Introduction 13
3.2 Results and Discussion 13
3.3 Conclusions 15

4. MATHEMATICAL MODELS USEFUL IN TOXICITY ASSESSMENT 19

4.1 Introduction 21
4.2 Results and Discussion 22
4.2.1 Terrestrial Simulation Models 25
(1) Regional Biogeochemical Models 29
(2) Radiological Cycling and Transport
Models 29
(3) Global Biogeochemical Models 29
(4) Forest Succession Models 30
(5) Theoretical Ecosystem Model 31
4.2.2 Aquatic Simulation Models 31
(1) Ecosystem Models 31
(2) Fate Models 33
(3) Fate-and-Effects Models 33
VI1
-------
CONTENTS (continued)
4.2.3 Generalized Multipopulation Models 34
4.2.4 Alternative Methodologies 34
(1) Loop Analysis 34
(2) Time Averaging 36
(3) Input-Output Analysis 36
(4) Natural Selection Models 37
4.3 Conclusions 37
4.3.1 Criteria for Evaluating and Selecting
Models 37
4.3.2 Research Priorities 40
4.4 References 42

5. METHODS FOR MEASURING EFFECTS OF CHEMICALS ON
TERRESTRIAL ECOSYSTEM PROPERTIES 49

5.1 Introduction 51
5.2 Results and Discussion 52
5.2.1 Microbial Processes 52
(1) Identification and Evaluation of
Test Systems 53
(a) Cellulose Decomposition 53
(b) Nitrogen Transformation 54
(c) Sulfur Transformation 54
(d) Other Existing Test Systems 55
(e) Proposed Alternate Tests 59
(2) Protocol Development 63
5.2.2 Identification of Model Ecosystems 65
(1) Test Description 65
(a) Synthetic Systems 65
(b) Excised Systems 65
(2) Measurable Parameters 66
5.2.3 Evaluation Model Ecosystems 68
(1) Synthetic Systems 68
(2) Excised Systems 69
(3) Uncertainties Concerning All
Model Ecosystems 69
5.2.4 Protocol Development for Model Ecosystems ... 70
5.3 Conclusions 71
5.3.1 Microbial Processes 71
5.. 3.2 Model Ecosystems 72
5.3.3 General Discussion 72
5.4 References 74

6. METHODS FOR MEASURING EFFECTS OF CHEMICALS ON AQUATIC
ECOSYSTEM PROPERTIES 79

6.1 Introduction 81
vm
-------
CONTENTS (continued)
6.2 Results and Discussion 83
6.2.1 Test Descriptions and Measurable
Ecosystem-Level Properties 83
(1) Lentic Working Group 83
(2) Lotic Working Group 87
6.2.2 Evaluation of Test Systems 87
(1) Lentic Working Group 87
(2) Lotic Working Group 93
6.2.3 Protocol Development 96
(1) Sediment Core Effects Test Procedure ... 96
(2) Pelagic Ecosystem Effects Test
Procedure 98
(3) Model Stream Effects Test Procedure .... 99
6.3 Conclusions 100
6.4 References 101

7. METHODS FOR MEASURING EFFECTS OF CHEMICALS ON
TERRESTRIAL POPULATION INTERACTIONS 105

7.1 Introduction 107
7.2 Results and Discussion - Microbial Populations .... 109
7.2.1 Population Interactions 109
(1) Predation and Parasitism 109
(2) Competition Ill
(3) Mutualism Ill
(a) Lichens Ill
(b) Methanogenesis Ill
(4) Antagonism 112
(a) Test Procedure 113
(b) Analysis of Results 113
(c) Development Needs 115
7.2.2 Community Properties 115
(1) Survival Test 115
(2) Fungistasis 117
(3) Population Levels 118
(4) Succession 119
(5) Diversity 119
7.3 Results and Discussions - Plant and Microbe
Populations 120
7.3.1 Interference 120
7.3.2 Mycorrhizae - Plant Interactions 123
(1) Endomycorrhizae - Grass 123
(2) Ectomycorrhizae - Conifer 124
7.3.3 Rhizobium - Legume Interaction 124
7.3.4 Wheat - Wheat Rust 125
7.3.5 Carrot - Crown Gall 125
7.3.6 Plant - Nematode Interaction 125
7.3.7 Agricultural Soil Microcosm 125
IX
-------
CONTENTS (continued)
7.4 Results and Discussion - Arthropod Interactions .... 126
7.4.1 Proposed Test Systems 126
(1) Plant - Whitefly - Parasitoid 129
(2) Corn - Earworm - Exploiters 129
(3) Alfalfa - Aphid - Parasitoid 129
(4) Plant - Brown Scale - Exploiters 129
(5) Housefly - Blowfly - Parasitoid 130
(6) Flour Beetle Competition 130
7.4.2 Promising Systems 130
(1) Plant - Herbivore - Exploiters 130
(2) Predator Competition 131
(3) Mutualism 131
(4) Plant Competition Mediated by Insects . . . 131
(5) Insect - Pathogen 131
(6) Interspecific Competition 132
7.4.3 Protocol Development (Tribolium Competition) . . 132
(1) Test Description 132
(2) Analysis of Results 133
7.5 Conclusions 134
7.6 References 136

8. METHODS FOR MEASURING EFFECTS OF CHEMICALS ON AQUATIC
POPULATION INTERACTIONS 141

8.1 Introduction 143
8.2 Results and Discussion 143
8.2.1 Evaluations of Test Methods 143
8.2.2 Group Discussion 147
(1) Predation Experiments 148
(2) Competition Experiments and Multi-
species Cultures 152
8.3 Conclusions 155
8.4 References 157

APPENDICES

A. EVALUATIONS OF SELECTED TESTS FOR EFFECTS ON
AQUATIC POPULATION INTERACTIONS 161

B. ALPHABETICAL LIST OF PARTICIPANTS 175
-------
TABLES

Page

4.1 Terrestrial Simulation Models 26
4.2 Forest Succession Models 30
4.3 Aquatic Simulation Models 32
4.4 Examples of Generalized Multipopulation Models 35
5.1 Evaluation of Major Test Systems for Evaluating
Effects of Microbial Processes 56
5.2 Characteristics of Proposed Substrates 61
5.3 Number of C02 Measurements Per Test 62
5.4 Schedule for Sequential Sampling: Aerobic System 62
5.5 Parameters Measurable in Terrestrial Model Ecosystems ... 67
6.1 Model Ecosystems (Lentic Working Group) 84
6.2 Measurable Ecosystem Properties (Lentic Working
Group) 85
6.3 Evaluation of Model Ecosystems for Measuring
Ecosystem Properties 86
6.4 Model Ecosystems (Lotic Working Group) 88
6.5 Measurable Ecosystem Properties (Lotic Working Group) ... 89
6.6 Evaluation of Model Ecosystems for Measuring
Ecosystem Properties (Lotic Working Group) 90
6.7 Evaluation of Test Systems (Lentic Working Group) 91
6.8 Evaluation of Test Systems (Lotic Working Group) 94
7.1 Rating and Evaluation of Test Systems for Microbial
Population Interactions 110
7.2 Rating and Evaluation of Test Systems for Plant-
Microbe and Plant-Plant Interactions 121
7.3 Ranking and Evaluation of the Proposed Test Systems .... 127
7.4 Rating of Arthropod Population Interactions 128
8.1 Evaluation of Predation Tests 144
8.2 Evaluation of Competition Tests 145
8.3 Evaluation of Multispecies Culture Systems 146
XI
-------
FIGURES
4.1 Hypothetical Scheme for Selecting Appropriate
Models for Use in Hazard Assessments 39
7.1 Typical Survival Curve 116
xi i
-------
ACKNOWLEDGMENTS
We wish to express our appreciation to the workshop participants
who graciously tackled the problems presented to them on a very short
schedule. We are also grateful to J. Vincent Nabholz, Project Officer,
Environmental Protection Agency (EPA), James J. Reisa, Associate Deputy
Assistant Administrator for the Office of Toxic Substances, and
David E. Reichle, Environmental Sciences Division Associate Director,
Oak Ridge National Laboratory, for their advice and continuing support.

Gratitude is also expressed to members of the Technical Information
Department, Science Applications, Inc., Oak Ridge, for their help in
arranging the workshops and preparing the manuscript for publication.
Donna Reichle, Judy Mason, and Bonnie Winsbro are especially acknowledged.
xm
-------
SECTION 1

INTRODUCTION
This series of six workshops was conducted by the Environmental
Sciences Division (ESD), Oak Ridge National Laboratory (ORNL) and
sponsored by the Office of Toxic Substances, U.S. Environmental
Protection Agency (EPA). The workshops were designed to identify
laboratory methods for measuring the ecological effects of chemical
substances and to evaluate those methods for their potential utility
to the hazard and risk assessment processes of the Toxic Substances
Control Act (TSCA). TSCA is comprehensive legislation that subjects
the chemical industry in the United States to federal regulation that
broadly protects human health and the environment from unreasonable
risks resulting from the manufacture, processing, distribution, use,
and disposal of a chemical substance. Under TSCA, EPA is responsible
for identifying and prescribing test standards to be used in
developing the data necessary to predict the risks associated with
exposure to chemical substances. Responsibility for implementation of
TSCA resides with the Office of Toxic Substances, EPA.

Results from the workshops were used by ESD staff in preparing a
critical review of Methods for Ecological Toxicology*. This review
was prepared to aid EPA in investigating the potential for developing
test protocols that predict the effects of chemical substances on
selected ecological parameters that are indicative of interspecific
interactions, community dynamics, and ecosystem functions.
Streamlined protocols are necessary if consistent results are to be
expected among different laboratories. The workshops were primarily
used as a mechanism to collect information about research in progress
by bringing together investigators presently working with aquatic or
terrestrial laboratory test systems.

The workshops were designed under the assumption that a
tiered-testing scheme would be the basis for EPA's environmental
hazard assessment process. Such a scheme provides for different
levels of testing ranging from simple, inexpensive screening tests to
higher levels of increasingly complex, definitive tests. Positive
results at one level of testing indicate the need to proceed to the
next higher level. Participants were asked to identify tests for
measuring the ecological effects of chemical substances and to
evaluate those tests in terms of their potential usefulness as
predictive tools in hazard assessment. In addition, one of the
*Hammons, A. S. 1981. Methods for Ecological Toxicology. A
Critical Review of Laboratory Multispecies Tests. Oak Ridge National
Laboratory, Oak Ridge, Tennessee. EPA 560/11-80-026; ORNL 5708.
-------
workshops addressed the role of mathematical modeling in ecological
hazard assessment and another addressed the major problems associated
with assessment and policy requirements of ecological toxicology
testing under TSCA.

This report represents a summary of the results of all six
workshops. It is obvious that even though an attempt was made to
design the workshops as consistently as possible considering the
different topics, assigned tasks were, nevertheless, handled
differently by the different groups of participants. Workshop results
are presented without argument or attempts to include information that
was not actually discussed during the workshops.

The criteria that were used in evaluating identified tests were
defined as follows:

Cost Per Test. The total cost of completing a test for a single
chemical assuming that the facilities are already available.

Documentation. The extent to which the behavior of a laboratory
system (not necessarily toxicological) has been investigated
and reported.

Generality. The usefulness of the test in predicting the
responses of a variety of interspecific interactions or
ecosystems and their major components.

Rapidity. The total amount of time required to complete a test
assuming that facilities already exist.

Realism. The ability to unambiguously interpret the response of
the test system in terms of responses of real ecosystems.

Rejection Standards. Defined criteria for rejecting test results
that range from informal or common sense criteria (e.g.,
many controls die) to a complete and well-defined set of
criteria (e.g., more than 10% of controls fail to achieve a
weight of 20g).

Replicability. The variance in response within an experiment
among individual units of a test system.

Reproducibility. The ability of a test to produce common results
in different laboratories.

Sensitivity. The ability of the test to produce measurable
responses at low doses of test chemicals.

Social Relevance. The value to society, direct or indirect, of
the response measured. The value may be economic,
aesthetic, or indirectly related to human health.
-------
Standardization. The definition of conditions and components of
a test system to allow different laboratories to obtain
similar results from a test.

Statistical Basis. Accepted statistical criteria for detecting
and interpreting responses of the test system.

Training, Expertise Requ irements. The extent to which use of a
test may be limited by requirements for higher education,
specialized training, or expertise.

Validity. The extent to which the responses of a test system are
known to reflect responses in the field.
-------
SECTION 2

CONCLUSIONS AND RECOMMENDATIONS
2.1 Introduction

No laboratory systems or mathematical models are presently ready
for use as predictive tools in environmental hazard assessment.
Nevertheless, there are several promising methods that are recommended
for further development. For example, protocols are suggested for
testing chemical effects on sediment cores, mixed microbial cultures,
model streams, Tribolium (flour beetle) competition, and carbon and
nitrogen mineralizations. Further experimentation is needed to adapt
many of the systems to chemical testing and to standardize and
validate all of the proposed protocols.

Selection of appropriate multispecies tests is not easy. Many
factors must be considered which require further research before
proper choices will be clear. For example, which types of systems
will yield the most useful, generalizable information when tested?
Which properties are most critical to the functioning of the system
and most sensitive to chemical stress? What magnitude of effect is
significant? What are the criteria for validating these systems in
the field?

Perhaps the major problem to be resolved before interspecific
interactions can be useful in hazard assessment is extrapolation or
generalization of experimental results to predict effects in natural
ecosystems. The degree to which chemical effects may be distorted by
the necessary simplification of laboratory test systems is not known.
Research is needed to (1) compare the sensitivity of laboratory
systems to that of natural ecosystems, (2) relate the ecological
complexity of laboratory systems (number of taxa or number of
functional groups) to their responses to chemicals, and (3) develop
models or other analytical approaches to link laboratory results to
predictions about chemical effects in nature.

Although many questions remain unanswered about the proper use of
multispecies tests in hazard assessment, certain generalizations can
be made. Because these tests generally are more complex than single
species tests and the results are more difficult to interpret, they
probably will be of most practical use in the later stages of a
tiered-testing scheme. As a result, relatively few chemicals may ever
be tested in these systems. Chemicals probably will reach the higher
levels of testing only if their economic or commercial potential is
great enough to justify additional expensive tests when a probable
hazard has already been indicated. A testing sequence should begin
with single-species screening tests. These tests are needed to (1)
flag potential problems, (2) help select additional tests and test
organisms, and (3) aid in the interpretation of multispecies test
-------
results. Despite the uncertainties associated with the use of
multispecies tests, they may be necessary to determine ecological
hazards if simpler tests indicate potential problems. Multispecies
tests will be especially important if the questionable chemical will
be either persistent in the environment or continuously released into
the environment. The final step of an entire testing scheme should be
field validation.

The main conclusions from this series of workshops are briefly
outlined in the following sections.

2.2 Terrestrial Test Systems

2.2.1 Population Interactions

1. The clover-fescue interference system received the highest
rating among plant and microbe systems because it combines
interactions between plant populations with interactions between plant
and microbial symbionts.

2. Other plant microbe test systems proposed for further
development include: mycorrhizae-plant; Rhizobiurn-1egume; wheat-wheat
rust; carrot-crown gall; plant-nematode; and agricultural soil
microcosm.

3. There is no clear perception that one or a few particular
types of arthropod interactions are superior to the others.

4. Tests for chemical effects on arthropod population
interactions with the greatest potential for use in hazard assessment
are: plant-white fly-parasitoid; corn-earworm-exploiters; alfalfa-
aphid-parasitoid; plant-brown scale-exploiters; housefly-blow fly-
parasitoid; and flour beetle competition.

5. The potential test systems for microbial population
interactions and community properties are not highly recommended.

2.2.2 Ecosystem Properties

1. Tests for predicting chemical effects on microorganisms
should be as close to the natural system as reasonably possible.

2. Kinetic (sequential) testing should be done.

3. Measurements of effects of general metabolic processes
occurring in the whole population or community (e.g., C02 formation,
02 consumption) are more meaningful than results from more selective
tests based on a single enzymatic criterion (e.g., sulfatase,
phosphotase, amylase).
-------
4. A test system is proposed to measure the effects of
chemicals on carbon and nitrogen mineralizations simultaneously using
environmentally relevant high nitrogen substrates and mixed microbe
populations that can be manipulated easily by technicians with minimal
training.

5. It is unknown whether microorganisms in the soil are
sensitive indicators of the effects of chemicals.

6. No model ecosystem, synthetic or excised, is considered
ready to serve as a test protocol.

7. Both defined (gnotobiotic) systems and intermediate-sized
grassland microcosms are recommended for further development.

8. The relevance of measured parameters to major ecosystem
processes should be determined.

9. Encasement materials should be evaluated in terms of
Teachability, absorptive capacity, optical properties, and durability.

10. The effect of variation in the chemical, physical, and
microbiological properties of soil on responses to chemicals should
be determined.

11. Round-robin evaluation is needed for all tests.

12. Field validation is necessary for all tests.

2.3 Aquatic Test Systems

2.3.1 Population Interactions

1. Laboratory systems involving predation, competition, and
multiple population interactions are available for development as
hazard assessment protocols, but few systems have been used for
chemical testing.

2. Comparison of simple and complex laboratory systems with
natural systems is a major research priority.

3. Many tests require special facilities or skills.

4. Reproducibility is virtually unknown for all systems
evaluated.

5. Relative sensitivities of different laboratory systems is
an important research problem.
-------
6. Few tests are rated highly expensive, but absolute cost per
test is generally not known.

7. Predator-prey tests appear to be more rapid, more
replicable, more advanced, and more readily standardized than
competition tests or multispecies culture systems.

8. Predator-prey systems have been used very little to test
effects of organic chemicals.

9. Predator-prey tests are more sensitive to chemicals than
acute single-species bioassays in many cases.

10. Multispecies tests are more useful in the intermediate
stages of hazard assessment.

11. Criteria that should be used in setting up the optimal
system for testing predator-prey relationships include criteria for
(1) the test organism, (2) the test systems, and (3) the test
protocols themselves. (These are outlined in Section 8.2.)

2.3.2 Model Ecosystems

1. Extrapolating from laboratory tests to natural systems is
the major problem in model ecosystem research.

2. Research is needed to identify the responses most sensitive
to chemical stress.

3. A predictive or mimicking model ecosystem is probably most
useful in later stages of hazard assessment.

4. A generic ecosystem (e.g., mixed-flask culture) is probably
most useful earlier in the testing sequence to screen chemicals for
their ability to disrupt ecosystem processes.

5. Smaller systems are more replicable and more easily
standardized among laboratories.

6. Statistical analysis of results is easiest with small
systems.

7. Standards for rejection are less likely for larger models.

8. Interpretation of chemical effects is more difficult for
larger systems.

9. Systems meeting the operational criteria for a screening
test are least generalizable to natural ecosystems.
-------
10. Protocols were developed for chemical effects on model
streams, mixed microbial cultures, and sediment cores.

11. Each proposed protocol needs extensive refinement and
validation.

2.4 Mathematical Models

1. Available mathematical models appear to be best suited for
use as relatively inexpensive and rapid qualitative tools for
preliminary screening to explore the possible effects of chemicals.

2. Considerable development and testing will be required before
mathematical models can be reliably used for predicting effects of
chemical substances on ecosystems.

3. An overall strategy for selecting and applying models is
required before models can be used productively in hazard assessment.
-------
ASSESSMENT AND POLICY REQUIREMENTS OF
MULTISPECIES TOXICOLOGY TESTING PROCEDURES
November 13 and 14, 1979
Jeffrey M. Giddings

Environmental Sciences Division
Oak Ridge National Laboratory
-------
12
PARTICIPANTS

J. M. Giddings, Chairman
Oak Ridge National Laboratory

Willard Chappell A. S. Mammons
University of Colorado Oak Ridge National Laboratory

C. F. Cooper Eugene Kenaga
San Diego State University Dow Chemical Company

Sidney Draggan J. V. Nabholz
National Science Foundation U.S. Environmental Protection Agency

Farley Fisher J. C. Randolph
National Science Foundation Indiana University

J. W. Gillett J. J. Reisa
U.S. Environmental Protection U.S. Environmental Protection Agency
Agency
-------
13
SECTION 3

ASSESSMENT AND POLICY REQUIREMENTS OF
MULTISPECIES TOXICOLOGY TESTING PROCEDURES

3.1 Introduction

This workshop was designed to help establish guidelines for
evaluation of laboratory tests for use in predicting the ecological
effects of chemical substances above the population level of
biological organization. Three main topics were selected by the
Environmental Sciences Division (ESD) staff to be addressed in
roundtable discussions during the two-day workshop:

1. What properties and functions of communities and ecosystems
should be addressed in evaluating the ecological hazard of a
chemical? Consider: Ecological significance, System
specificity, Natural variability, Sensitivity to chemical
disturbance, Ability to be measured, Ability to be simulated
in the laboratory.

2. How should these properties and functions be used in a
hazard evaluation process? Consider: Utility in
preliminary screening, Utility in predictive modeling, At
what stages particular types of information are needed.

3. Identify criteria that are important in evaluating the
usefulness of any test. Examples might include:
Replicability of test, Sensitivity of test, Statistical
basis for interpreting results, Standards for rejecting test
results, Frequency of failure, Time required, Cost per
chemical.

The following sections summarize the workshop discussions and
conclusions. Comments and suggestions that were considered by ESD
staff to be most relevant to the goals of the workshop are included.
No attempt was made to develop a consensus report.

3.2 Results and Discussion

Much remains unknown about the use of multispecies laboratory
tests for predicting chemical effects on interspecific interactions
and ecosystems. The significance of measurable effects to the health
of these systems has generally not been determined. For this reason,
emphasis was placed on the basic need for developing generally
accepted hypotheses that would allow the identification of significant
effects on interspecific interactions or ecosystem properties and that
would enable tests to be statistically designed to either validate or
disprove previously formulated testable hypotheses. It is important
to understand perturbed or stressed systems well enough to select the
most important responses to test. So that test results can be
-------
14
properly evaluated, it is equally important to understand the
generalities of the mechanisms of processes in non-stressed systems
and the ecological significance of variations in measurements. Until
such knowledge is available, the safest objective of an ecological
hazard evaluation process might be to maintain viable ecosystems close
to their present point of balance.

Development of a data base on studies of notable "natural
experiments"--ecosystems that have already been impacted by toxic
substances released into the environment (e.g., smelters or accidental
spills) may be a way to help identify appropriate parameters to test.
Such an exercise would be particularly useful if assessment of such
studies uncovered patterns of "typical" effects on "typical"
ecosytems.

No single species, combination of species, or any ecosystem can
be representative of all species or all ecosystems. As a result, the
choice of species or systems for a test will necessarily vary and
depend heavily on the uses, residues, and resulting exposure potential
expected for each chemical under consideration. Some participants
recommended that the most sensitive and the most likely to be exposed
species or systems would be the best choices to test. Obviously,
determining the most sensitive species or system would not be an easy
task. In addition, if an ecosystem that would likely be exposed to a
questionable chemical contains species of particular economic or
aesthetic value, special attention must be given to the possible
effects on those particular species. Special attention was also
suggested for multispecies interactions associated with pathogens and
parasites. For example, hosts weakened by exposure to toxic
substances may be more prone to succumb to disease or predatory
attack. There was a suggestion that species should also be tested
separately to determine whether either might be affected
independently. Participants cautioned (1) against assuming that
protection of an ecosystem, as determined by gross parameters, would
protect all of the individual components of the system, (2) against
substituting multispecies tests for single-species tests, and
(3) against excluding "special" ecosystems or worst-case systems for
consideration in a hazard evaluation process.

Several participants speculated that comparatively few chemicals
would ever be tested at the highest level in an ecological hazard
evaluation scheme where multispecies tests would perhaps be most
appropriate. A chemical will reach the higher levels of testing only
if its economical or commercial potential is great enough to justify
additional expensive testing even after earlier tests indicate
probable hazard. Because tests at this level (i.e., most multispecies
tests) will be used on relatively few chemicals, they can generally be
more complex, sophisticated, time-consuming, expensive, system-
specific, etc., than screening tests. Except, perhaps where tests
that require purchase of costly, highly-specialized equipment would
not be acceptable for limited use. Although the workshop consensus
-------
15
seemed to be that effects on interspecific interactions and ecosystem
processes would probably be tested only at the higher levels of a
tiered-testing scheme, this does not prevent the development of
multispecies tests for screening purposes.

There appeared to be some agreement that, for purposes of the
Toxic Substances Control Act (TSCA), aquatic ecosystems may prove to
be generally more relevant to test than terrestrial ecosystems.
Aquatic systems, especially those lacking sediments, are likely to be
more sensitive than terrestrial systems, which are buffered by soil.
Consequently, results from tests on terrestrial systems might cause
the hazard to aquatic species to be underestimated.

Despite the numerous problems and uncertainties associated with
such an effort, tests for chemical effects on ecosystems and/or
interspecific interactions will be necessary to determine ecological
hazards if simpler tests indicate potential problems. These tests
will be necessary especially if the questionable chemical will be
either persistent in the environment or continuously released into the
environment. Multispecies tests should be used to enhance our
knowledge and predictive capabilities.

3.3 Conclusions

What properties and functions of communities and ecosystems
should be addressed in evaluating the ecological hazard of a chemical?

Integrative parameters, such as diversity or total primary
production, were generally considered less useful in hazard assessment
than information on the presence or absence of particular species in
the community. Species composition and interspecific interactions
(competition, predation, herbivory, and parasitism) are probably more
sensitive to chemicals than gross ecosystem functions (primary
production, secondary production, and decomposition) because changes
in the latter functions can be compensated by shifts in community
composition. On the other hand, changes in species interactions and
community composition are probably more ecosystem-specific (i.e., less
generalizable) than effects on gross ecosystem functions.

Among the emergent ecosystem properties, nutrient cycling and
resistance to additional stress were considered to be currently the
most useful to hazard assessment.

The opinion of one participant was that predator-prey
interactions are easier to study in the laboratory than in the field,
whereas the reverse is true of plant-herbivore interactions.

The following comments are presented as examples of individual
thoughts concerning this question.
-------
16
J. W. GILLETT: Only those functions need to be tested that are likely
to be more sensitive than any single-species response.

J. C. RANDOLPH: Production (P), respiration (R), and P/R ratios are
functional attributes that have been shown to have a high level of
ecological "significance." P/R data have a certain level of system
specificity, mostly between terrestrial and aquatic systems; however,
these characteristics are not unique to a wide variety of system
types. Sensitivity to chemical disturbance is likely to be extremely
variable. Although technically somewhat tedious in some cases, there
seem to be neither conceptual nor technological constraints on our
ability to measure community respiration and production.

C. F. COOPER: In the development of multispecies protocols, it may be
advisable to start slowly (i.e., to begin with combinations of two
species), perhaps along an induced environmental gradient. One could
determine how the two species sort themselves out along that gradient
(habitat preference; optimal growth under competition) and then see if
the equilibrium is shifted when the system is challenged with the
chemical in question.

W. CHAPPELL: For the longer term there are areas where research needs
to be done to establish whether meaningful measurements could be
developed for diversity, connectivity, competition coefficients, etc.

How should these properties and functions be used in a hazard
evaluation process?

Most of the information already discussed will be of secondary
importance in hazard assessment and need not be tested unless earlier,
simpler tests (such as single-species bioassays) indicate potential
problems. Effects on some interspecific interactions might be
predictable (via mathematical models) from single-species tests.
Gross ecosystem function, nutrient cycling, and resistance to stress
could be tested after the initial screening tests if the exposure
assessment and screening tests indicate a need for further testing.
If ecosystem functions are more sensitive than selected single-species
parameters, it may be necessary to develop suitable screening tests
for effects on ecosystems.

To a large extent, the sequence of tests will be determined by
the availability of simple, inexpensive protocols. However, some
aspects of the sequence can be based on scientific reasoning. For
example:

1. If a chemical is expected to persist in soils or sediments,
tests on nutrient cycling and decomposition are warranted.

2. If different species on a single trophic level vary greatly
in their sensitivity to a chemical, effects on interspecific
competition should be tested.
-------
17
3. If a chemical affects plants and animals differently,
plant-herbivore interactions should be tested.

The final step of an entire hazard assessment scheme should be
field experimentation to validate previous tests and confirm
predictions.

We agreed that no tests are ready for use as screening tools in
hazard assessment. All tests are presently too complex, too
expensive, and too difficult to interpret, and none have been
validated. The Oak Ridge National Laboratory (ORNL) soil core
microcosms now being tested at the Environmental Protection Agency's
(EPA's) Corvallis laboratory are perhaps closest to standardization
for routine use. Selected individual comments are presented below.

J. C. RANDOLPH: If, at the community/ecosystem level, some
predictable input/output (or dose-response) relationship could be
established between exposures to hazardous chemicals and ecosystem
properties such as production and respiration, there exists the
possibility of using these relationships in routine preliminary
screening tests. If we wish to investigate the mechanisms of the
ecosystem responses to exposure to hazardous chemicals, much
additional research is needed. There seems to be little justification
for routinely attempting to monitor for complex ecosystem properties,
such as nutrient cycling, when we presently know relatively little
about the generalities of the mechanisms of nutrient cycling in
nonstressed ecosystems. There is a very wide conceptual and
technological gap between single-species acute and chronic tests done
under carefully controlled laboratory conditions and truly
process-oriented ecosystem analysis that by its very nature must be
within the particular ecosystem of interest. Thus, it would seem
desirable to go to a third-tier level of hazard evaluation for
ecosystem processes only after screening and longer-term,
single-species tests have indicated a high probability for some effect
on some specific ecosystem property.

C. F. COOPER: The test should be statistically designed to validate
or disprove previously formulated testable hypotheses about not just
the occurrence but the nature of an effect. Preparation of these
hypotheses, based on knowledge about the behavior of similar
compounds, behaviors in single-species tests, etc., are likely to be
the most important and the most difficult part of the program.

W. CHAPPELL: If one could obtain a field sample (e.g., lake water
and/or sediment) or a suitable model and subject it to a dose of the
chemical of interest, it might be possible to identify the most
sensitive species for later single- and multispecies testing under
more controlled conditions. This may or may not work, but it seems
worth a try.
-------
18
F. FISHER: The position of a test in a screening system must depend
on the difficulty of the actual test procedure, which is difficult to
assess at this time. Field testing may be appropriate at the higher
tier.

Identify criteria that are important in evaluating the usefulness
of any test.

The following criteria (in alphabetical order) are considered
important in evaluating the usefulness of any multispecies test for
hazard assessment:

Ability to interpret results unequivocally

Cost of capital equipment

Cost per chemical

Existence of standards for rejecting experimental results
(e.g., death of controls)

Frequency of rejection of results

Replicability

Sensitivity (i.e., effects are seen at low chemical concen-
trations, compared to single-species tests)

Skill or expertise required

Statistical basis for detecting effects

Time required

Wide range of sensitivity to different classes of chemicals
and a range of sensitivities within classes.
-------
MATHEMATICAL MODELS USEFUL IN TOXICITY ASSESSMENT
January 8 and 9, 1980
L. W. Barnthouse
Environmental Sciences Division
Oak Ridge National Laboratory
-------
20
PARTICIPANTS

L. W. Barnthouse, Chairman
Oak Ridge National Laboratory
D. L. DeAngelis
Oak Ridge National Laboratory

W. R. Emanuel
Oak Ridge National Laboratory

R. H. Gardner
Oak Ridge National Laboratory

Daniel Goodman
Scripps Institute of Oceanography

T. G. Hal lam
University of Tennessee

James Hill
U.S. Environmental Protection
Agency

Richard Levins
Harvard School of Public Health

Marc Lorenzen
Tetra Tech, Inc.
R. J. Mulholland
Oklahoma State University

R. V. O'Neill
Oak Ridge National Laboratory

Richard Park
Rensselaer Polytechnic Institute

J. E. Richey
University of Washington

B. W. Rust
Oak Ridge National Laboratory

H. H. Shugart
Oak Ridge National Laboratory
G. L. Swartzman
University of Washington
OBSERVERS
J. V. Nabholz
U.S. Environmental Protection
Agency

A. S. Hammons
Oak Ridge National Laboratory
D. A. Mauriello
U.S. Environmental Protection
Agency
-------
21
SECTION 4

MATHEMATICAL MODELS USEFUL IN TOXICITY ASSESSMENT

4.1 Introduction

Four and seven years ago, our Father built Him a model,
And He built it out of the stuff of the computer,
And He nurtured it with the stuff of His own soul
And He breathed life into it--and He thought that it was good.
And He rested.
And all the models on the face of the earth—they were fruitful
And they multipled.
And now—in this great judgement hall we are called
Together to evaluate these models—for didn't He say that
There shall come time when the protection agency shall
Arise among us and the ghosts of dead models shall
Congregate in the halls of judgement awaiting redemption.
Yea, all the models, the lame and the sick, even the
Poorly documented shall come . . .
And they shall be judged.
Be this our task? I beg of you LORD, give us strength
To know the fat from the lean.

—Invocation by Gordon L. Swartzman
January 9, 1980

The purpose of this workshop was to identify specific
mathematical models and general modeling techniques that could be
useful for predicting the effects of toxic substances on ecosystems or
multispecies assemblages. The workshop sessions were organized around
three major topics:

1. Identification and documentation of models and modeling
methods potentially useful for prediction of toxic
effects.

2. Development of criteria for evaluating the usefulness
of models.

3. Identification of research priorities.

Because of the large number of extant mathematical models, the
workshop focused on general types of models, categorized as follows:

1. Terrestrial simulation models

2. Aquatic simulation models
-------
22
3. Generalized multipopulation models
4. Alternative methodologies
Participants were divided into four groups, one for each of these
categories. To stimulate discussion, the compositions of the working
groups were varied among the three sessions. The criteria and
research priorities proposed by the working groups were then presented
and discussed at general roundtable sessions. Group participants are
identified for each session as follows:
Group 1
Mulhoi land
Shugart
Emanuel
Gardner
SESSION I (Model Description)
Group 2
Lorenzen
Hill
Lassiter
Park
Group 3
DeAngelis
Swartzman
Hall am
O'Neill
Group 4
Richey
Levins
Goodman
Rust
SESSION II (Model Evaluation)
Group 1
Mulholland
Gardner
Swartzman
Goodman
Group 2
Lassiter
Park
Richey
O'Neill
Group 3
DeAngelis
Hal lam
Emanuel
Lorenzen
Group 4
Levins
Hill
Shugart
Rust
SESSION III (Research Priorities)
Same as Session I
4.2 Results and Discussion
We agreed that, although no specific model has proven valuable in
predicting the effects of chemical substances on ecosystems, a great
many models have potential value.
Ten distinct types of terrestrial simulation models were
identified. These types of models ranged in scale from models of
single plants (suitable for coupling to more complex models) to models
of regional and global biogeochemical cycling. The majority of these
models were developed to simulate material transport and cycling and
are, therefore, more suitable for predicting transport and
accumulation of chemical substances than for predicting their effects
on ecosystems. Most could, however, be modified (with varying degrees
of difficulty) to incorporate toxic effects.
-------
23
The many extant aquatic simulation models were divided into three
basic types: fate, effects, and traditional ecosystem models. The
fate models simulate biotic and abiotic transport phenomena; the
effects and traditional ecosystem models simulate ecosystem dynamics.
The prediction of effects of toxic substances on ecosystems requires
the coupling of fate models to ecosystem models; effects models are
ecosystem models expressly designed for this purpose. Like
terrestrial models, aquatic simulation models vary widely in scale and
complexity. Models have been developed for most types of aquatic
ecosystems, including lakes, rivers, estuaries, and seas.

Generalized multispecies models are much less complex than
terrestrial and aquatic simulation models. They are not site specific
or even ecosystem specific. The structures of these models (i.e.,
identification of variables, functional complexity, and environmental
coupling) can be tailored to suit the objectives of the modeler.

The alternative methodologies category includes all modeling
techniques that cannot be placed into one of the other categories.
Four alternative methodologies were identified, all distinctly
different in approach and application than the ecosystem simulation
and generalized multispecies models. These alternative methodologies
can be used for such purposes as predicting the effects of stress on
the stability of systems, predicting the direction of change of
arbitrarily defined variables (e.g., population sizes or production
rates) in response to stress, predicting changes in patterns of
nutrient cycling, and predicting the evolution of populations
subjected to stress. None of these methodologies have yet been
applied in chemical effects studies, but all have potential
applications.

A variety of criteria were proposed for evaluating the usefulness
of models. Some are very general and apply to all model types. Among
these criteria are the match between the properties of the model and
the objectives of the assessment, the generality and ease of
validation of the model, temporal scale and resolution, whether the
model makes socially relevant predictions, and whether the model
suggests practical monitoring protocols.

Research priorities for both model development and testing were
suggested. Development of "standard" models and "standard"
environments was proposed for both aquatic and terrestrial simulation
models. Development of parameter structure handbooks and flowchart
decision trees were recommended as aids in the development of new
models and as guides for the selection and use of models in chemical
hazard assessment. Theoretical studies were also recommended to
delimit the possible effects of chemicals on ecosystems. More
generally, in addition to the development of specific models, an
overall strategy for using models as part of hazard assessments should
be developed. This strategy should include modification of current
laboratory protocols to provide appropriate data for model input, the
-------
24
development of evaluation criteria and benchmark test data, and
verification of model predictions by comparison to field monitoring
data.

Models and modeling methods identified by the workshop
participants as being potentially useful in chemical hazard assessment
are briefly described in the following sections. These discussions
focus on the properties that are most relevant for evaluating the
usefulness of the models in hazard assessment. Among these are:

1. Spatial and temporal scale. Spatial scale relates to the
size of the region being modeled (e.g., local site,
watershed, state, or whole planet). Temporal scale relates
to the time period being modeled (days, weeks, years, or
decades). It is important that the spatial and temporal
scales of the model be commensurate with the spatial and
temporal scales of the expected impact.

2. Trophic complexity. In most ecosystem models, the hundreds
or thousands of individual species of organisms are
aggregated into groups of species with similar properties
(e.g., trophic levels or functional groups). Alternatively,
in some models the organisms of interest (e.g., tree species
in forest succession models) are modeled as individual
species, and the remainder are aggregated into broad
groupings or ignored completely. Clearly, the way in which
species are aggregated affects the purposes for which a
model can be used.

3. Mathematical formalism and computer implementation. What
kind of equations are used (e.g., differential equations or
finite difference equations; deterministic or stochastic
equations; linear or non-linear equations)? What
programming language is used to translate the equations into
a computer program? Is the program documented so that it
can be understood and used by persons other than the program
developers? Are program modifications required to run the
program on different computers? The mathematical formalism
used in a model affects the kinds of purposes for which it
can be used and, secondarily, the cost and difficulty of
using it. In addition, programming considerations can place
severe constraints on the usefulness of complex simulation
models. An otherwise suitable model can be virtually
useless if it can be understood and used only by its
creators or if it can be run on only one kind of computer.
The ideal simulation model should be documented in a user's
manual that describes the computer program and explains how
to use it. The program should be written in a programming
language, such as FORTRAN, that is available on most
computers.
-------
25
4. Kinds of effects predicted. These can include changes in
the abundance of populations or of groups of populations,
changes in yield of economically important species or
changes in the stability of ecosystems.

5. Validation. All models are abstractions and simplifications
of reality. Therefore, it is necessary to investigate the
correspondence between the properties of the model and the
properties of the real system being modeled. Validation
must not be confused with verification (i.e., the
demonstration that the computer program is an accurate
translation of the model's equations) or calibration (i.e.,
the adjustment of model parameters so that the output of the
model matches a data set). Comparisons between model
predictions and empirical data (especially experimental
data) are particularly valuable for assessing the validity
of models to be used in chemical hazard assessment.

6. Original purpose of the model. The purpose for which a
model is developed inevitably affects the structure of the
model and constrains the ways that it can be used. A model
specifically designed for predicting effects of chemical
substances will generally be more useful for chemical hazard
assessment than will a similar model designed for some other
purpose.

7. Modifications needed to predict effects of chemical
substances. Models developed for purposes other than the
prediction of effects of chemical substances may require
substantial modification to be useful in chemical hazard
assessment. For example, models designed to predict the
bioaccumulation of pesticides or radionuclides generally
employ extremely simplistic representations of biological
interactions that must be made more realistic to predict
biological effects of chemical stress.

4.2.1 Terrestrial Simulation Models

The working group on terrestrial simulation models identified ten
types of such models and prepared a table (Table 4.1) presenting
summary descriptions of each type. In addition, the group developed
more detailed descriptions of the five types with which they were most
familiar and evaluated each one of these with respect to the criteria
presented in Section 4.4.1. These descriptions, along with citations
to specific examples of each type of model, are presented below.
Readers who are unfamiliar with the principles of simulation modeling
and the somewhat specialized language used by modelers may wish to
consult the excellent non-technical discussion of modeling written by
Walters (1971).
-------
26

2
Lbl
O

1
vo
_J
I/}
UJ
ce
o:
UJ
h-

i-H
UJ
1

m
re •—
m re
o> u
m »^-
3 w

|S
c -o
QJ 01
CL OJ
C

5
m
u

C
.^

re
•o

IA
at
i
re
X
UJ

re
u o
•— L.

11
+* *J

o
t»
QJ QJ

•— u

S C - O -3
•— E^i"Oi" *— C in

X: Oi — » i- i- ro re >
oca> . o e u "c m c
3>^ 'f— ^) »— «•- £ O ~~
i— c w in a* *^ *D OJ **—
*>x: 3 e x: re > — o
u +j — o -CUE re
— •— re E • u u T3
at -^ - aj ID — ^ *•••» aj
u m o>a> a> X) x: -^- o_ in c
Ou'— O O)^ *J 5 * —
ro »— s- -u z: x: c

*>uu'— o 3 t- a> in

TD *-» \ C OX: 3 CTO O -^
a* re ^"- *^- t- m O" o •»— -
S ^ u "O re ^*« u *re m — "TO
*- *^ a* in c *j oj * •
xt o 3 o o •— at o o
t_s **- x> Q. o u •£ x: a> o (o CT
— -o
'ai -> ai
•O U in
g-i- ^-3
•D O
a> o c -
c o i- c a> a>
O —> Cv*»- 0) in
t. "o *o m at u
10 O> C •— i- V)
o m ro a> at *D 4->
3 > x: x: i-
•— in O> O. 3
re O) */> ^"~ u> *•— o
x» c at o a> u
O •*- in w E "O
»- 01 w O •*•* O C
u xi re o re E *^-

c **- ex:
01 W Q. C O O --'
^J-gM*-°**'|*S5J.5
re D> C tn*o re c in" S * ji lo ^ in

f— -o . "5 E a> S"u S 2! "S £ c -2

§re o *re *J -o a. t. o.— *> >> >
"O U C i- O>in3OU(OX)L.
^ ^ h~ 0) c *t~ in E *D O tn a> ^ tn
oreo u aix: a» *J a> MMX:CX>

re i— •—
re re

* •s t
"D iH CO CO
o r*^ re co re co
^ r-t UJ i-l UJ f-l
I I — l-l 1 C
•1i *i*lg-ls §
in s- at -*j — r re «^-
a> rei-^-c ^-»-cu*J

ID at -^- *^- m a> re ^-> >> 3 re
^- E -J ••- c -D x: o. c re u a>
X) U T- 4.) U *•- O
re . >>•--> u in . at in u
in a. QJ i- re at E >,•*- c CL
re o 3 c CT QJ at re — ••- o at
o

V)
at
•o
re

\.S
»— s.
fO 3

O C
U U

_
i re
•— O U in
re at •*- *—
XI Ol E 0)
o o a> "o
•— •*- x: o
U X) U E
0 U

o w E
C re O
•o >j
-- ^- X)
re re E
> at
in
^ -o c
a> e re
•o re
o c
E — —
-D

U •—
^ E S!
re o 3
c w-o .
o a» a)
O)^ O TO
at at u •.->
ce jt Q. w
at
OJ • (
*A —•* 3 1
O ro D .
a> *-* > u w
(/> rH -^- 4->
re .. >, t- u
U ^ »— D) 3
T*- u re -o
*^ •-< ro o
L. i— S» I-
3 0) 3 O O.
O L) (J 1-
U C— t—
(u ^-* a> ro
t— •«- s- — i-
O U (0 X) 3
O l/> Q_ (0 *•>
1
*U t in E S-
0) 3 1 3 U 1 O
roc c o c o o ID in .

E^- at u TO — x: re re o

a> u o *» ~D o a>
f— •— ••- o)atoo"-ec»—
tu ro o o *^ to CJ>*— xt
r— *j L. a> DIX: -tJ o
UJ i-> £ U i« O-*-> — X>i«*^'o)

re o> *—

0) O •*->
0)
C T3
O C ^"
Ift 10 •—
i_ o 4-> ^- a* CM
i- r* ** f*» z r*.
I rH Z i-* O »-l
a>
i
in

OJ ro
U A
C 0
ro •—
2 i- "i?
o •—
in **- OJ
in "O
ro in o

ro o in
in c* ro
<*^*t- OJ OJ
in c >,
r- 10 O
ro ja v,»-
c <~ o
o t- QJ
— OJ E m
Ol > O C
a, — — QJ
OC S- £> *J

r^ f^.
§1 1C
O u in
O OJ •*- *—
— 01 E QJ
Ol O OJ "O
QJ — .e o
ce j3 o E
01 C C
I- 1- — O

C O QJ in --> OJ
"- — C in ^ u » i

3 *^ OJ E QJ C CJ

ro S . it. > c >,
U -f- O D. U C
C ro +J t~ E • 1. OJ

•— •— O QJ O U
•— TO ••- *J TJ in —
rO U U • OJ *-

C 3 ••- D. in
OJ X QJ in U QJ ro u 3 c -D
— t— in u >i E OJ
4J *J •— O 1- in 1.
rorcl-QJ O^COJ
I- "ol Oj O O f .1- — -i-
U in
ce i ->
Z 1 3 C
c — 01 ro
O>'»- D) OJ *—
c c i- a.
— -O OJ
*J C -O fc-
*J ro • c OJ
ai 01 ro s
in in c O
•O "- - Q.
cs-D.cn
— ro o c 01
•o •— i- c
•O C QJ i- —
OJ ro > QJ *J
in *J OJ QJ ro
3 in -a c —
OJ
— i -D
*j xi e en
C S- l~ C Q)
"" fc -g S. I o.
Ol U ••- OJ E
ro ro in c u o
i. C 0 3 U
15 •*•» O m
oj cj £ ro c
*J OJ U Ol OJ
ro Q. u E OJ
•a o • ro a
«- u fO OJ I-
*— o -4J in OJ m
ro in 1C OJ *J ro
> "- T3 1- OJ i
I-
» in i — a>
O "- ro *>
§OJ U m
O '5l • T3 OJ
UI E O C C 1C %
t— — ro ro f-
s- o 6 en -o
» OJ ••- £ t-( C
uj j= •? e 01 ro
u^ 4^ re 4-> 3 -
§o u o x QJ
OJ •— 1C TO ID
•o s- ui p- *: -t- r~
*— i c o o f-u rerji
ro i
C OJ 1
— I. - U
•— OJ in OJ 3
**- C t. U
OJ •*- «*- O O *»
*j o •— •— E m
re -o *>
*J in re QJ QJ
m c >, 3 E *J
i o t- o~ o re
>,— Q QJ m i.
•o <-> c o •
re 3 •*- *— .c A QJ
QJ •— "o re -4J re t-
4J O 1- — f- — 3
i/) in o *J 2 OJ **

O 1
w • s-
in •-•
N E re
J5 -
i in in

O in QJ
— 0 .C
Ol u m ^
OJ OJ &. in
t. N— OJ TJ
.0 *J OJ
3 re re .c:
v« Z > in
"re -0
u c
Ol £-
O O! O
•— C Q. in
O— m —
— — C 01
•o u re -o
re >, s- o
cr o — ' E

QJ
Is
x: en
v-H

-------
27

•o
Ol
3
C

C
O
U
rH
UJ
—I

tn —
3 <•-

re o
c -o
01 01
o o>
.Lg

IA
C
o

re
u
"o.

c
o

re
13

i/i
01

X
UJ

"re
U Ol
^> s
11
f «-

*io
u
o
Is

X. TO
TO in

TO
Q.
LO

^_
OJ OI
*O Q.

I I
O QJ
a.— 4->
ro 3 re
X in tn
0) =•

o u
w . u
*/> re
4^ £ GJ
15 4J Z
L> in i.
•*• tft O)
**- C
•o ro 4->
O> TO

^ O E
TO 4J £
u o •
Q) QJ «*- in
C 4»* O)
5m c —
.— •»- O)

•O
C C 14-
.*. o> O
*- .J in
3 C
**- U O
QJ 0 —
in <<- 4->
3 U
.Q.Z .,- (J
*» -O TO
X v> C.
TO QJ t. E
Z -*-> 0 —

O>
c

tn tn
O> 01 >>
4J in &
-£*
re <-> oi
U O 4-> .

Q. O) 0)
E »- = .C
UJ O ift •»•>

re
ic "»J
r^ o> CO
en ID
t-4 IA ert

f— 01 in
— 01 C
oi cm —
ai en 01
O O rH —1
t- C
Q 01
Ol I- •
C Ol W
— •*- C
— 1- O
c "° ID
>,£>§•
•— re oi
F— c
3 -5 're
in s- —
r> o *>

^
z
re i i—
U E 3 01
— JU 1 -0
O) in 3 E
I. >> U
0 in >•,
01 O T3 <->
JC U C —
>- oi re c
1 "-
4J U *J
re 01 01
c c: re js 4-> 01
(C — S- E fC £_ •
> O — .C "D O IA
01 IA w- c c E *J
£ l"g^.£ll 1
TO in QJ 4-> QJ ro •*->
>,^ O i- U i-
^— O)^ 4J C Q.
TO - • C QJ TO E
••- . *— . TO U. "O O
O • ^- C QJ OJ
M 0> Oi O> >>Z > U
^^ £i o> —
tn >, o r- 4J>
•^ in c in O
U 4-> I- — OJ O» E —
•D QJ X) ro LJ C *J ro
GJ >— E C QJ «t) in
£ *£ 3 * S"^ 5»* c
i
tn o
OJ -J

u in
>, O. in 4J
TD in c in
3 O QJ -
*J i*- Q. i- in
in o w O E
Oj t*_ QJ
O tn L. 4->
4J 4-> C in
U •— O >»
"^ QJ QJ in
QJ **- > XO
in **- QJ o U
13 O> ^- (/) O>
TO

TO £ O 1
"O O! QJ QJ W

*•> ^ OJ TO t C O — •
C t-C O*t-*~TOTO
*f-4->TOOW U O **4-> C
O!C *J QJ t- C O *D TJ
TO 4J t- re >» o«-
in QJ ^ QJ -^- 4-> 4-» in
"O QJ £ s- o -M 4J ro cm
QJ £ 4^ 3 o re u J-> QJ ro
in t- -•»- U Oi C O
H- O O D.CNJ U in > Q..Q
rH in in
Q) Q)
^- in ^
c a* c c
W fO i— 1 4-) TO TO

S- 4J 4^ 4J
c a> w c t. t.
O) JS «O -^ TO TO
4J (X> U O) •*: D)V£> OlO
4jr*w 1.3 *J sr** 3co

Q. i— 1 ^(/> 00 CO rH tr> rH

c.
3 •
c
t-> O
— at •—
TO in 4^
U -r- TO
•-- b. 4-> in »—
*» TO C C 3
in QJ oj O E
10 c c. f- •£
£ -r- QJ 4J in
U •— l^- TO
0 C ••- 3 S-
4-> O -r- O" O
l/» C ^ 0> **-

in
£.2
TO I-
0> 3
*^1r
jo a>
c
o
in
4^ in in
in QJ r-
QJ O QJ
S- U "O
O 3 O
u. in E
O1 m
5 i c
c c E (^
>> O I O — • X *• —
in i_ u O in — ai i-
C Q; — in a. jC 4-*
O O 4J >i«-> OJ O.4-J 3

*J .— OJ S- — S- *O
~O ro > ^ 3 O *- JJ

>.^=JZS-4->SlnJ.-S
£ *" E o "- -r- ^ O
•OT5O .«*-E*r.3 >>
J£C -"SS*"^ u STO
r- rein TOO —En
TO *J QJ »O 4J QJ X O -r-
CtOJintn ^ O •*" Oi
U ft) o • •— "D U
C QJ U ••- X QJ *— ^ C O
•r-fsjO4->O>^ 3 4^ ro
t.'»- fc- U — '~ O*r-^ I/)
C^ in O. TO CL— X » ** TO

1 4->
OJ L. in i c i
IA JZ TO CJ C Qj TO

O — tr> CL U n
n- TJ I- **- CJ E — —
3 OJ O U O **- r-

QJ in •.- QJ (- 3 U JD •
in in U O. E w V
3*DCC OJ t-COJ
O> O fO*-*^*^- O TO^-
Olr- U*' «J Ift-r-W (J^
^ — l_ r- >, E

c 4^ ro Q. u *J 4-> o m
TO CJ£ E-r-S- C TO"- m
CJ"O4->'r-4-» O 01^4-* TO
1 1
— s. oi
t. O T3
TO W C
> •— '
•o -o
QJ QJ OJ i-n
4^ 4-) in . *J
in TO 3 C in
QJ 4^ O O)
§-J C
«-> 0

•— C in jQ O* S
§«— fl) •*> TD o
re — — c S
in oi.fi ID 01 c
=> re re u o. 3

CVJ
en
r-i
V)
"c
c

£- >)
re i —
c u re •— a> c
•*~ o 3 re oi o
•o er t t- •-
u. ^- o) ft* re *->
o re c *— re
S- *^ U CJ Hi 3
re c c ID e
Ol 0> 01 — *
c t- L. • Q. in in
••- 01 01 W 3 r—
r— s- >*- C 01 Ol
c w s- o «•> -o -o
o — — — oi o o
X "O -a •*•> m o E
.

3 I. u re
•^ - >.
•> "O *i— >^
C OJ L> »- O
O 4J QJ *r- 4J
•r- TO O- f
O) 1- tn T3 >.
QJ A i X. *-
W *^ QJ TO lAff
3 TO ••- TO QJ
l/> U i/i "O ?

O)
•2 m
CO »—
QJ
ft. "O
CO O
TO 1
E 13 TO
QJ C 4-> .
•— 5 C in
^ v- OJ U
O w —
I- TO CJ E

O Q. =
4-> QJ >,
OJ £ *" "°
w inr- c
3 f- TO O
C (J *r-
r- m— -J
TO £ t ro

e i to.
01 01 O
o c >,
•o •— o
-> -o ro
re re -4-1 c
£c o
in oi —
U *^ E 4-*

— 1/1

re 3 Ci &^-> 01
_> C O m f
Ol f— — •— Ol U
j-j - .* o> a. m
m u re >
*> •»- E oi -o ->
in •_ "O 0) C
oi c -J ai
c re o o re E
— — *J u Ol

•D re — -o oi re
01 U U 01 ^ C
IA O Ol IA C ID
= ^ -o r> — E

en
en
CO rH
en tsi

U 4->
re 3
X CJ)

1 1A
m § o?
U CC Q£
•*- <
tn in •
re c in
u .* ai c
o c -o o
*j a> o —
i« -S E w
^

c
3
a>

"o.
i
V)

Q.^
O QJ
i. 0)
O J
o>
TO- tn
4-) QJ
in a. in

*•> cx ai
"s i- o
£ U E
-------
28

•o
QJ
3
C

•£
O
u
1— (
LU
OD

in
•o c
c o
TO •»-

m re
O> LJ
in *f
3 V-

re o
4J
c -o

O OJ

in
C
O

o.
•S-

c
o

re
•o

ro
>

in
OJ

i
TO
X
LU

TO
U 0>
4-> 3
re *-»
E U
OJ 3
.C U

TO in
E

're
o
Is

\ TO
»— U
TO in
re
D.
LO

'flJ OJ
•D D.
O >»

>> re
c IE re ""

O) -T *—
7i QJ U 1

TO i- C
:>> — TO
"- in n.^
— — £ u
TO fti ft) ft)
— -O E •
U O C TO
O S O «- 0)
in O t.
4J TO
V) . C C
*•> *-

ftj v- CfO O
t, M— ft) "O i-
a. at -o < CL
j -o
>> oJ 'o>

U) -D -
GJ cn-o
s- o c c
o 4J ••- re
*- u
>> ? S.4J
^O j-> m in
in ft>
•D 3 ft) >
QJ -D C U
=>•- E £

4-> re

c re

ro
cn aj
TO 4->
re
•o s-
•4-> ja in
m re •*•»

t— 'ftJ «

rH C\J

r— r-H
o>
^ >»
U OJ

— L.
E <

U 1
OJ 3
•*-«/)£
**- C "-
— O »n
*O •<••
O 4-> U
•^ j- ro o
*J TO 3 **-
in ft) CT
TO C ftt D. C
JT -^ 3 O
U — 0» ••-
O C U -4-> *>
-j o c 0) re
iy> c oi in -—

^
TO C
fti o
^1*^
»H ro

TO O
X- ^
+>
>> 1
4^ QJ in
tn O>'~
O> TO OJ
I- C T3
o ro o
lZ £ £

"c
re

,_
QJ

U
0

u •*->

"O OJ
ft) *»-
Q. ft)

w- .. i/T QJ

o» •»- t-

U 0) >>X1 —
OJ -J -C O OJ
*— re c 4J TD
«*- t. - ro o
in c ft)
i_ ._ ^ ft)
QJ D, U - Q.—
>*- O TO U 3 TO
^H u m ft) u m
t t
TO •—
> u

4-» O' D.
ID in X
c JD a1 -
"- o ro
re "04-1
O) a> c ro
re 4-> re "O
re
13 t, — r-
ft) o re TO
*j .a c ->
in re o c

H- *OJ •*-» £

O 00
x r*-

O i-H
C
QJ

QJ in ro
u- c o
w- O —

^ 4J 4->
TO U
>> 3 ft) .
fc- O"— in
re ft) ft) O)
= O
•i- •— O »—
"O ro in CQ
fc- — •— C
O *•* cu ro

\ t.
4J ro
c o>

,_
CL o
QJ
*— in
CJ5 i-
C 3
•»- O
LO ^

1
QJ in
*Dl C QJ
C TO T3

Wl Q. E
in
"i
•— OJ
O:
in t.
— re

•D

o 5
QJ X in
t- 4J C
3 i- O
10 ••- 4->
c ja re

u u t.

to CL. i-
LO « CL

c
o

JE ^

U *J
«C 01

in b.
C I- O -
•*• W **- C
TO r- O

ro E aj 4J
— in re
-O v> 3 t.
0> J3
4J TO 4J -F-
l/> *J TO •—
O) re ^ re
H- "D 4J 0
CO
en

OJ >t
.- i?
ro £
E U

Q> ^

-*J T3
OJ C
— 1 (D

re
u
a< .
en u

n E
re
(- C
ns >>
g?

-J C
o

TO
£ O

in
•0 in

j= fti
in QJ •>—
t- 5 re
0) "*v 3
*•> 0-J C
re E c
^ ^ re
i in
+j — •
0 •§
1 O
— ' E
3

« CL
-------
29
(1) Regional bi'ogeochemical models. These models were developed
for use In legal proceedings related to the regulation of DDT. They
are regional in scale, with all biotic components aggregated into
trophic levels. All components of the ecosystem, from abiotic
compartments through top carnivores, are included. The models are
formulated as differential equations describing mass balance; the
number of state variables can vary between 3 and about 15. All are
coded in FORTRAN. No user's manuals exist. Examples of these models
can be found in Harrison et al. (1970) and Hett and O'Neill (1974).

Regional biogeochemical models have been used to predict
accumulation of DDT in biotic compartments, especially in top
carnivores. They can also predict toxic effects of DDT. Although it
would be difficult to modify these models to predict effects of
chemical substances other than DDT, the same principles and ideas
could be useful in formulating new models for those substances.
Regional DDT cycling models are relatively low in generality, require
relatively large amounts of data for calibration, and are difficult to
validate. However, they are comparatively easy to use, and their
temporal scales match the temporal impact scale (years to decades) for
chemical substances.

(2) Radiological cycling and transport models. These models
were developed as hazard assessment tools for isotope releases. They
are airshed models for food chains and are used to predict the food
chain transport of isotopes from airborne dispersion from a point
source. As a result, the probable dose to humans can be predicted.
Chemical effects could not be predicted without substantial model
modifications.

Large amounts of data are required. If sufficient data are
available, they can be readily validated. Socially relevant
predictions are made, such as the dose to man resulting from the
release of a chemical substance from a point source. However, the
time scales of these models do not match basic impact scales for toxic
substances. The models are multiplicative chain models. State
variables are concentration ratios, rates of release, and decay rates.
FORTRAN is the programming language used. User's manuals and
interactive codes are available and descriptions of the models have
been published in the open literature (e.g., Hoffman et al. 1977;
Killough and McKay 1976; Schaeffer et al. 1978).

(3) Global biogeochemical models. These models were designed to
predict the C02 concentration in the atmosphere resulting from
combustion of fossil fuels. Some have been adapted for modeling
global cycling of DDT (Woodwell et al. 1971).

The models used coupled differential equations that are usually
linear, but have selected nonlinear terms. FORTRAN is usually the
programming language used; however, a few models are in simulation
languages such as CSMP (e.g., Gowdy et al. 1975). In most cases,
-------
30
documentation is sketchy, although all are described in journal
articles (e.g., Killough 1980; Emanuel et al. 1980a and b; Gowdy et
al. 1975; Bacastow and Kealing 1975; Bjakstrom 1979). These models
are modified frequently.

Extensive modification would be required to handle cycling of
toxicants. The cycle must be analagous to that of major chemical
elements material inputs. They cannot predict the effects of chemical
substances nor do they predict socially relevant impacts. Moderate
amounts of data are required. Validation is relatively easy.
(4) Forest
succession
succession models.
over long time scales.
vegetation and predict the effects
deletion on forest succession. These
research aid to understanding ecosystems;
of stochastic nonlinear difference equations and are coded in ANSI
Standard FORTRAN. Most of the succession models cited in Table 4.2
are documented in the open literature.
All these models simulate forest
They model soil compartments and
of S02, climate change, and species
models were developed as a
they are formulated in terms
TABLE 4.2 FOREST SUCCESSION MODELS
Name
Forest type
Citation
JABOWA Northern Hardwood Forest

FORET Southern Hardwood Forest

BRIND Australian Eucalypt Forest

FORMIS Floodplain Forest

SELVA Puerto Rican Rain Forest

FORAR Mixed Oak-Pine Forest

KIAMBRAM Complex Notophyll Vine
Forest
Botkin et al. 1972

Shugart and West 1976

Shugart and Noble 1980

Tharp 1978

Dolye et al. 1981

Mielke et al. 1978

Shugart et al. 1980
No model modifications are required to predict the effects of
chemical substances. The temporal scales modeled are applicable for
either long-lived chemicals or for continual release of short-lived
chemicals. The predictions, such as changes in timber production,
made by these models are obviously socially relevant.
-------
31
(5) Theoretical ecosystem model. DeAngelis et al. (1975)
developed a generalized model for use in theoretical studies of
ecosystem structure and function. The model can be applied to any
type of ecosystem. Nonlinear algebraic equations are used; these can
also be used as terms in differential equations (O'Neill 1976). The
functions are completely documented in the journal article cited
above. No programming is required for steady-state analysis of simple
models. Analysis by numerical simulation would require writing a
program.

This model has never been used to predict effects of toxic
substances on a real ecosystem and would require extensive
modification to predict such effects. This model cannot be easily
validated, does not make socially relevant predictions, and does not
suggest a monitoring protocol. Like the similar models described in
Section 4.2.3, this model appears to be useful primarily in
formulating research strategies and in initial screening studies.

4.2.2 Aquatic Simulation Models

The working group on aquatic simulation models identified three
types of such models: ecosystem models, fate models, and
fate-and-effects models. Ecosystem models focus on biological
processes, such as primary production, grazing, predation, and
decomposition. Physical and chemical interactions and transport
phenomena are either ignored or treated superficially. Ecosystem
models can predict the biological effects of chemical substances, but
cannot predict the movement and fate of chemicals. Conversely, fate
models emphasize physical and chemical interactions and transport
phenomena at the expense of biological realism. Fate models can
predict the movement, chemical transformations, and fate (including
bioaccumulation) of chemical substances, but cannot predict biological
effects. Recently, efforts have been made to develop hybrid models,
here called fate-and-effects models, that can predict both the fate
and biological effects of chemical substances in aquatic ecosystems.

(1) Ecosystem models. Aquatic ecosystem models exist for entire
lakes, streams, estuaries, or open seas (see Table 4.3). The original
purposes for developing these models are varied. For example, some
were constructed for research purposes, some for predicting the
effects of eutrophication (e.g., Chen and Orlob 1975), and one (Tetra
Tech, Inc. 1979) for predicting the effects of power plants. Although
none of these models have been used to predict the effects of toxic
substances on ecosystems, all are detailed enough for the effects of
toxic substances on organismal physiology to be extrapolated to
population and ecosystem effects.

The level of aggregation for these models varies; some are
aggregated into trophic levels, and some into two or more functional
groups within trophic levels. In general, the lower trophic levels
are modeled in the greatest detail, although most models include
levels through piscivorous fish.
-------
32

_l
UJ
o
i
i

«
g

in
l—
if

1§
re-
in re
CJ U
in -^-
3 w
<— *5
re o

C T3
^g •§
O 0)
c

re
u

TO
•o
re

x
LD

re
U CU
•«- i-
•*•* 3
re +J
fy
3
JZ U
*> 4->
i"

p_
re
o
Is
*-> *—

•— o
re v>

re
D.

ft) CU
•o o.

i. o
l/> U ~ r— TO **-
j-> .P- .hj a) cu a;
*J .C • U > Wl i/> >>
c CL in ™ a? o di •—
*— o Q.-CJ — a* az .c «—
v- u 3 a> .0 --> TO
cu *j o t. u •<-
S- Q."- — > • C U
d_) C O"' ^ *^"- O •
uo o a. 3 »— CL in
•— •— -»oscuci/> —
•3 m re J- -c o cu cu
j-uocj ^£-j *ai
o. c "~ *^ t. 4 re ^3 ^~
re *•* ••*• o) cu o> Q/
o *J u 3 5 J-* re *c u
5 5 *"•—.£:»•- & S IE
-D z; 4- i- tn uca.

3 U O -» U — *J
•~ QJ in C "O O
.O O f— U ^-t-^CU
c > 01 *^- re 3 "o CD
s- u
O 1 GJ

c •r-
3 i_ -_> 4-> W (/)
TO u re — >
--> a> *D ^ u c

o i) tr> w ^ t—
E J- 3 0- CL CL

•f- 'o c D ey
*n **-•»— Tp in
3 •»-» c
4-> If) -F- (O

o to *-* c to
•+-* re >i
re *J JD *J *J
i. re me
.D "O "D C 0)
•^ 0) » re c
re c o o» a>
w o •— re a.
*> *

t. i-H
^ U.
re
- I re
tu re

re cu cu

re r^ a> r*» t.

re i
II

c
o •—
c re

,>>•£

re t- •
^ ai in
ai »*- c
C*- 0

in
i.
f— re
gre c 01
-*•> 01 >>
in j£ re o o
>> re o *>
in t— u -
O o in
0 • • C >>
"8?-5
ai o; re .— ~-.
— t. 3 01 •—
o f *> f re
§m IA a. ai
^ 01 3 in

o>
*>

IA
o
u
UJ

E
O>

i/)

O
cu

o
1
s-
5

Ol •
J3 m
in -o
ii
CU
4J-D
re c
JJ •*- TO
U
•^ T3 «/l
•D C CU V)
Ci TO T3 ^~
t- *r- TO
D.*J U *->
I_— CU
O O — » E
4-> o. m
w cu >,
TJ C CL >
cu TO re
tn u >*- cu
=>*J 0 J=
C £
CU i in
O> "O ui O
* «- 0> i- U
-_> O *j -O E O
CU U O U
J3 C "- -O (- U
o 13 cy •*- •—
C — CU > E
O 4-» U i- C
w> 3 CL cu O s-
•»- J3 u> -^- O
t- -1- U J2 4->
re t- cu o 3 -o •
t-j u ,n«— re
m re "o •«- cu *j
-»- i- c s- — re

en -D
r^- c
•o cr> re
r— C i-H
re re Gc f
c or^
— > c cu i- en
O> O *n +J f-n
IA C (A

E en o o re in
irt f-H ac *^ u- ^

£-, ,
i re *^- TO
C C— 3

t. o m »—
o ^-*- re

u w u *> .
TO re re c in
eu cu CL cu c
c c i- o

CU (A
+J U
(A TO
>> CU

o
u o
cu •»-»
cu */>
— t-
11

Ol
re
01 ai -a

—i O S- Ol
u s ai
•t- I/) W- C O
•3 «J in
Oi u .C in ^~
*- 01 u ••- re

s- e 01 ai
(-01 C !-
o 4-> •'- re
i— -a 3 *>
c J3 in in
^ re 01 01
3 -4-> l«
i*- - in re
01 Ol 01 *D J3
in .*> u c
s re c re re
•4- re -t>
>, +* +* re
r— - in C -O
"m HI'S E-a
•F- o in Q. 01 •
4J Q. <3 > "O
C in u *— O 01
01 C ••- Ol t ^
*J re x > p. 01
O >- O 01 E U
a. ^> «< -o» c

c
01
01
J3

re -o
f ai

Ol •—
C D.
o a.
•X. re

' in
in i/> *J *J .
re »— w o ~o
E 01 01 C Ol
in 01 ^ •*- *-»
•o *> o n- 01 to
O in E 01 > -a
j= >, re —
*» m ai our F—
01 o <-> c n
E u TJ-— in >
ai u- •>->•—
ai in ai c
E I- TJ — T3 Ol
03 0 C X O 01
i/> n- re uJ & ja
X
ai

I
TO
O
y U3

U- rH

L*
01 C
1- O

•o re

t- cr
re ai

•f- re
o e

i
*J IA
in u
in ai
0 X
u
01 O
01
^ in
It
1

1 U
cu cu
*4 «*.
re *-
u. cu
-------
33
Most of these models use nonlinear differential or difference
equations describing mass balance. FORTRAN is the program language
used most frequently. Machine dependency varies. In general, model
documentation is very good (e.g., Scavia et al. 1976; Steel and Frost
1977; Anderson and Ursin 1977). User's manuals are available for most
models listed.

As would be expected, model validation was considered easier to
achieve for lower trophic levels and abiotic compartments than for
fish.

(2) Fate models. Fate models are available for lakes, streams,
and estuaries (Table 4.3). These models were constructed to predict
the transport and fate of pesticides or other chemical substances.
Bioaccumulation is also included. The level of aggregation in fate
models usually includes abiotic campartments plus trophic levels,
although a few, such as PEST (Park et al., unpublished draft), model
functional groups within trophic levels. The abiotic compartments are
usually modeled in detail. Biotic compartments include primary
producers such as phytoplankton and/or macrophytes and primary
consumers such as zooplankton and/or benthos. Fish may or may not be
included.

These models use nonlinear or linear differential equations.
Sometimes partial differential equations are used to simulate
two-dimensional hydrodynamics. As with ecosystem models, the program
language most frequently used is FORTRAN. In general, documentation
of these models is not as good as that for ecosystem models, although
the EXAMS (Lassiter et al. 1978) and PEST (Park et al., unpublished
draft) models are reasonably well documented. Documentation for the
European models (Fagerstrom and Asell 1973; Mogenson and Jorgensen
1979) is not known.

These models cannot predict the effects of chemical substances on
ecosystems, but must be coupled to ecosystem models. PEST, for
example, is compatible with CLEANER (Park et al. 1974).

(3) Fate-and-effects models. Efforts are now being made to
develop models that predict the tranport, fate, and effects of toxic
substances in aquatic ecosystems. For example, Falco and Mulkey (1976)
have described a pesticide fate-and-effects model. Falco and Mulkey
coupled a one-dimensional fate model (used to predict the movement and
transformation of Malathion® in a river) to a simple dose-response
model that predicts reductions in standing crop of fish caused by
exposure to Malathion®. The model consists of linear and nonlinear
ordinary and partial differential equations, and is coded in FORTRAN.
No user's manual is available. Although fate-and-effects models are
specifically designed to predict the effects of chemical substances on
aquatic biota, none have been used in practical applications to date.
-------
34
4.2.3 Generalized Multipopulation Models

Ecosystem simulation models are intended to be realistic
representations of particular ecosystem types. Modifying them to
model a different ecosystem can be time-consuming and expensive.
Alternately, it is possible to construct simple, highly generalized
multipopulation models that can be rapidly and inexpensively tailored
to fit any system of interacting populations, aquatic or terrestrial.
Using this modeling strategy, no attempt is made to model every
component of an ecosystem; only those processes believed to be
critically important are modeled. Transport phenomena are not
incorporated in these models. Thus, the models can be used to predict
the effects of chemical substances on systems, but not the fate of
those substances. These models are not thought to be appropriate for
detailed chemical and site-specific assessments; however, they can be
used in the early stages of a hazard assessment to rapidly explore the
possible effects of chemical substances. Results of such studies can
aid in determining whether a more detailed (i.e., expensive and
time-consuming) modeling effort is warranted.

Four categories of generalized multipopulation models were
identified. In order of increasing complexity, these are:

Functionally simple, not environmentally coupled.
Functionally simple, environmentally coupled.
Functionally complex, not environmentally coupled.
Functionally complex, environmentally coupled.

Within each category, models can be either spatially homogeneous
or spatially complex and either age-dependent or not. Table 4.4 lists
some examples of most of the categories. Although many of these
examples were developed with particular systems of populations in
mind, the principles employed can be applied to other systems as well.

4.2.4 Alternative Methodologies

For the purposes of this workshop, "alternative methodologies"
were defined as any modeling technique that does not fit into one of
the other three categories. Four such techniques were identified.
Two of these, loop analysis and time-series averaging, are methods of
analyzing the qualitative behavior of systems of coupled differential
equations. They could be applied to many of the generalized
multipopulation models discussed in Section 4.2.3. Input-output
analysis is a method of econometric analysis that has been modified
for use in analyzing material budgets in ecosystems. Natural
selection models are applications of population genetics theory to the
problem of predicting the evolutionary response of populations to
toxic substances.

(1) Loop analysis. This modeling technique is designed to
analyze partially specified systems, i.e., systems in which the
-------
35
TABLE 4.4 EXAMPLES OF GENERALIZED MULTIPOPULATION MODELS
Age . Spatial fa
Type structure structure Reference
DeAngelis et al. 1975;
Canale 1970;
Rescigno and Richardson 1957
1
1 +
2
3
3 +
4
4 +
4 +
+ Levins 1974
Hassell and Comins 1976;
Pennycuick et al. 1968
Emanuel and Mul noil and 1975
Hsu et al. 1977
Travis et al. 1980
Craig et al. 1979
Eggers 1975
+ Andersen and Ursin 1977

1 = Functionally simple, not environmentally coupled.
2 = Functionally simple, environmentally coupled.
3 = Functionally complex, not environmentally coupled.
4 = Functionally complex, environmentally coupled.

- = Age structure (or spatial structure) absent.
+ = Age structure (or spatial structure) present.
-------
36
patterns of interaction among the component variables are known, but
parameter values and functional forms are not (Levins 1974; Lane and
Levins 1977). The level of aggregation and trophic levels modeled are
arbitrary. This type of analysis can predict effects such as local
stability, direction of change of variables in response to altered
parameter values (e.g., input of chemical substance) and correlations
among variables responding to different inputs (Levins 1974; Lane and
Levins 1977).

Coupled differential or difference equations are used. The
analysis can be validated by comparing the predicted response of a
system to a parameter change against the actual response of a
perturbed system. This type of analysis has not been used to predict
the effect of chemicals. Loop analysis can be used to (a) predict the
response of a multipopulation system to an applied stress, (b)
identify critical parameters that should be measured, and (c) identify
system properties that enhance or reduce impacts. It may not be
applicable to systems that are far from equilibrium.

(2) Time averaging. Time averaging can be used to model any
system of interacting populations or aggregates of populations (Levins
1979). This methodology, which is complementary to loop analysis, was
developed to extend ecological theory to nonequilibrium systems. The
kinds of effects predicted are changes in variances and covariances
among variables as affected by parameter change.

Coupled differential equations are used. Analysis focuses on
statistical moments of variables, especially second-order statistics.
Validation can be accomplished by comparing predicted changes in
variances and covariances against actual responses of a perturbed
system by using pre- and post-perturbation time series. This method
has not been used to predict the effects of chemical substances.

Time averaging is potentially useful for analyzing time-varying
systems that are far from equilibrium. Time averaging can be used to
(a) characterize microcosms before addition of chemical substances,
(b) distinguish populations that are directly affected by a chemical
substance from those that are indirectly affected, and (c) provide
warning about possible structural change caused by chemical
substances.

(3) Input-output analysis. Input-output analysis has been used
to compare material cycling patterns in different ecosystems (Finn
1976; Hannon 1973; Lettenmaier and Richey 1978). It has been
hypothesized that structure and cycling indices derived from
input-output analysis might also be useful as indicators of
environmental stress. These indices can be computed from a matrix of
material flow coefficients using a computer program written in FORTRAN
IV. This program can be coupled to "process" models that compute the
flow coefficients.
-------
37
The analysis can be applied either to whole ecosystems or to
subsystems within ecosystems. The indices are now useful primarily
for descriptive purposes and as indicators of system dysfunction
caused by stress. In theory, input-output analysis can be used to
predict changes in material flow patterns in response to stress to the
biota, but further development and testing are required before it is
known whether this is feasible in practice.

(4) Natural selection models. Population biologists have used a
variety of models to study the evolution of populations and systems of
interacting populations in response to changes in their environments
(Levins 1968; Kimura and Ohta 1971). All of these models relate rates
of changes in gene or phenotype frequencies to selective pressure,
heritability, and genetic variance within populations. They can be
used to predict adaptive responses of species to chemical substances
and to predict the effects of those responses on population size,
location, behavior, and interactions with other species.

Natural selection models can be validated by comparing predicted
changes in gene frequencies to actual changes in a population
experimentally exposed to stress. Although they have not been used to
predict the effects of chemical substances on populations, they are
potentially valuable for this purpose because populations in nature
frequently evolve in response to exposure to toxic substances.
Pesticide tolerance in insects and antibiotic resistance in pathogens
are notorious examples. Practical applications would require
experimental work to measure the genetic variances in tolerance within
and between populations for species of interest and to estimate
selection intensities in the field.

4.3 Conclusions

We recognize that few, if any, existing models of any kind have
been demonstrated to be useful for predicting the effects of chemical
substances on ecosystems. Ecosystem simulation models are the only
type to have had significant applications to date, but their
complexity makes their use comparatively difficult and expensive.
Generally, available mathematical models appear to be better suited
for use as relatively inexpensive and rapid qualitative tools for
preliminary screening to explore the possible effects of chemicals
than for use as detailed chemical and site-specific hazard
assessments.

4.3.1 Criteria for Evaluating and Selecting Models

No single model or model type can fulfill all needs associated
with environmental hazard assessment. For this reason, one of the
workshop tasks was to develop criteria that could be used to evaluate
the usefulness of existing models, modified versions of existing
models, and new models. These criteria include not only the
properties of the models themselves but also the match between the
-------
38
capabilities and deficiencies of the models and hazard assessment
objectives.

Nine criteria for evaluating the usefulness of models were
developed:

I. The degree of modification required for handling toxic
material inputs. Can chemical inputs be modeled directly?
Are the physical and chemical processes that govern the
transport and fate of chemical substances included in the
model? Are the biological processes directly affected by
toxic materials included in the model?

2. Data requirements. Is the amount of data required for
parameterizing the model consistent with the available
resources (i.e., time and money)?

3. Generality. Can the model be used for only one geographic
region or ecosystem type or can it be easily applied to
others?

4. Ease of validation. Has the model been validated against
baseline data? Are the output variables (i.e., those that
must be measured to test the model's predictions) easily
measurable? Do modifications required for handling chemical
substances invalidate the model? Can the model be tested
with microcosm systems as well as with field data?

5. Social relevance. Is the model output relevant to
regulatory needs?

6. Relevance to monitoring. Does the model suggest an environ-
mental monitoring protocol? For example, does it suggest
indicator variables that are easily measureable and that
could be used as early warnings of environmental effects?

7. Spatial/temporal scales. Do the spatial and temporal scales
of the model match the basic impact scale?

8. Ease of use. Is the model documentation comprehensible,
consistent, and complete? Is the computer code readily
available? How much modification is required to implement
the code on a different computer system?

9. Acceptance by the scientific community, especially the
ecological community. Are these models based on biological
ideas and mathematical procedures accepted by most of the
ecological community?

Figure 4.1 presents a scheme that could be used to identify
specific models for use in hazard assessments; it was prepared using
-------
39
(IDENTIFICATION)
IDENTIFY SPECIFIC
REQUIREMENTS AND
OBJECTIVES, !.».,
Law and/or Social
Ralavancy
"CONSIDERATIONS"

1. SCREENING: SIMPLER
MODELS ADEQUATE
2. DISPUTED RULINGS:
MORE COMPLEX
3. POLITICAL AND LEGAL
CONSTRAINTS
"ALTERNATIVES"
FATE
MODELS
I
ECOSYSTEM
MODELS
I
CRITERIA FOR IDENTIFICATION
AND DEVELOPMENT

1. INCORPORATE BIOLOGIC PROCESS
LEVELS
2. - ALL TROPHIC LEVELS
3. PORTABLE
4. DOCUMENTED (ESPECIALLY PEER
LITERATURE)
5. CAPABLE OF "TRADITIONAL" +
FATE = EFFECTS
_ E. G. CLEANER
HYDROCOMP
TETRATECH

EFFECTS
EXAMS
PEST
MODEL SELECTION

1. BASED ON BENCHMARK DATA SET
FROM EITHER "FIELD" OR "MANUFACTURED
TEST' DATA

2. BY AN INDEPENDENT PANEL (EPA SELECTION)
IDENTIFICATION OF SPECIFIC MODEL
FOR SPECIFIC TASK
Figure 4.1. Hypothetical scheme for selecting appropriate models
for use in hazard assessments.
-------
40
aquatic simulation models as examples, but it could apply equally to
any type of model. The scheme highlights nonmodeling decisions that
must be made before appropriate models can be selected for an
assessment problem. These include formulating the specific legal or
social questions that the model will be expected to answer and
specifying whether the purpose of the assessment is the screening of
many chemical substances for potential effects or the detailed
evaluation of particular substances in connection with regulatory
actions.

4.3.2 Research Priorities

The workshop participants identified several kinds of research
and development activities that are needed to increase the usefulness
of mathematical models for predicting the effects of toxic substances
on ecosystems.

Further development and testing of ecosystem simulation models is
necessary. Improvements are needed in the models themselves and in
the data bases used to parameterize and test them. Standard models,
specially tailored for the prediction of toxic effects, and standard
data sets are needed for representative terrestrial and aquatic
environments. As an aid to future model development, we recommend
that an ecosystem parameter handbook be compiled. This handbook would
include definitions and standard notations for parameters that are
used in ecosystem models. The handbook would also include a
codification of properties of ecosystems relevant to modeling (e.g.,
numbers of trophic levels and functional groups in different ecosystem
types, relationships between primary and secondary production, and
average numbers of prey species fed on by various predators). We also
recommend that selected aquatic ecosystem and fate models be coupled
to form effects models. The coupled models should then be verified
using benchmark data sets. In addition, new methodologies are needed
to solve two problems related to fate modeling. First, regional mass
balance models are needed to quantify the movement of chemical
substances between aquatic ecosystems. Second, specific methodologies
are needed to project loadings of important substances.

Theoretical studies using generalized multipopulation models and
alternative methodologies should be performed to define the possible
responses of systems to chemical substances. Examples of the kinds of
results that could be obtained are the identification of (a) system
properties that confer resilience or vulnerability to chemical sub-
stances, and (b) conditions under which sublethal exposures to
chemical substances can cause destabilization of competitive or
predator-prey systems. Results of such studies, which can be
conducted relatively rapidly and inexpensively, would suggest
processes that should be incorporated in more complex models and
hypotheses that should be tested using ecosystem simulation models,
microcosm studies, and field studies.
-------
41
Regardless of how many and what kinds of models are available, an
overall strategy for selecting and applying models will be required to
use models productively as part of the hazard assessment process.
This strategy should include the development of flow chart
decision-trees for selecting the best model(s) for any given
assessment problem. It should also include modifications of current
laboratory protocols to provide appropriate input data and the
systematic testing of models using microcosm and field data.
-------
42
4.4 References

Andersen, K. P., and E. Ursin. 1977. A multispecies extension to the
Beverton and Holt theory of fishing, with accounts of phosphorus
circulation and primary production. Medd. fra. Danmarks Fisk.
Havunders. 7:313-435.

Arney, J. D. 1972. Computer Simulation of Douglas-fir tree and
standard growth. Ph.D. dissertation, Oregon State University
Corvallis. 88 pp.

Bacastow, R. B., and C. D. Kealing. 1977. Models to predict future
atmospheric C02 concentrations. In: Workshop on the Global
Effects of Carbon Dioxide from Fossil Fuels. Miami Beach,
Florida.

Bjakstrom, A. 1979. A model of C02 interactions between atmosphere,
oceans, and land biota, pp. 403-451. In: The Global Carbon
Cycle—Scope 13, B. Bolin, E. T. Degens, S. Kempe, and P. Ketner,
eds. John Wiley and Sons, New York.

Botkin, D. B., J. F. Janak, and J. R. Wallace. 1972. Some ecological
consequences of a computer model of forest growth. J. Ecol.
60:849-872.

Canale, R. P. 1970. An analysis of models describing predator-prey
interactions. Biotechnol. Bioeng. 12:353.

Chen, C. W., and G. T. Orlob. 1975. Ecologic simulation for aquatic
environments, pp. 475-588. In: Systems Analysis and Simulation
in Ecology, B. C. Patten, ed. Academic Press, New York.

Craig, R. B. , D. L. DeAngelis, and K. R. Dixon. 1979. Long- and
short-term dynamic optimization models with application to the
feeding strategy of the loggerhead strike. Am. Nat. 113:31-51.

DeAngelis, D. L., R. A. Goldstein, and R. V. O'Neill. 1975. A model
for trophic interaction. Ecology 56:881-892.

Dixon, K. R. , R. J. Luxmore, and C. L. Begovick. 1978. Ceres--A
model of forest stand biomass dynamics for predicting trace
contaminant, nutrient, and water effects. I. Model description.
II. Model Application. Ecol. Model. 5:17-38, 93-114.

Doyle, T. W. , H. H. Shugart, and D. C. West. 1981. A forest
succession of the lower montanes rain forest in Puerto Rico.
ORNL/TM-7709. Oak Ridge National Laboratory, Oak Ridge,
Tennessee.

Eggers, D. M. 1975. A Synthesis of the Feeding Behavior and Growth
of Juvenile Sockeye Salmon in the Limnetic Environment. Ph.D.
Dissertation. University of Washington.
-------
43
Emanuel, W. R., and J. R. Mulholland. 1975. Energy based dynamic
model for Lago Pond, Georgia. IEEE Trans. Auto. Control
AC-20:98-101.
Emanuel, W. R. , and J. S. Olson, and G.
expanded use of fossil fuels by the
dioxide problem. J. Environ. Manage.
G. Killough.
U.S. and the
10:37-49.
1980a. The
global carbon
Emanuel, W. R. ,
Calibration
pp. 642-649.
Cybernetics
Electronics,
G. G. KiHough, and H. H. Shugart, Jr. 1980b.
and testing of models of the global carbon cycle.
In: Proceedings of the International Conference on
and Society. The Institute of Electrical and
Inc., New York.
Fagerstrom, T. , and B. Asell. 1973.
an aquatic food chain: A model
planning. Ambio 2:164-171.
Methyl mercury accumulation in
and implications for research
Falco, J. W. , and L. A. Mulkey. 1976. Modeling the effect of
pesticide loading on riverine ecosystems. pp. 156-160.
Proceedings Conference on Environmental Modeling and Simulation.
ORD, DPM, USEPA, Washington, D.C. EPA-600/9-76-016.

Finn, J. T. 1976. Measures of ecosystem structure and function
derived from analysis of flows. J. Theor. Biol. 56:363-380.

Gowdy, C. M., R. J. Mulholland, and W. R. Emanuel. 1975. Modeling
the global carbon cycle. Int. J. Sys. Sci. 6:965-976.

Gutierrez, L. T. , and W. R. Fey. 1975. Simulation of secondary
autogenic succession in the shortgrass prairie ecosystem.
Simulation Councils, Inc.

Hacker, C. S. 1978. Autoregressive and transfer function models of
mosquito populations. In: Time Series and Ecological Processes,
H. H. Shugart, ed. SIAM Publications, Philadelphia.
Hannon, B. 1973,
41:535-546.
The structure of ecosystems. J. Theor. Biol.
Harrison, H. L., 0. L. Loucks, J. W. Mitchell, D. T. Parkhurst, C. R
Tracy, D. G. Watts, and V. J. Yannacone, Jr. 1970. Systems
studies of DDT transport. Science 170:503-508.

Hassell, M. P., and H. N. Comins. 1976. Discrete time models for two
species competition. Theor. Pop. Biol. 9:202-221.

Hett, J. M., and R. V. O'Neill. 1974. Systems analysis of the Aleut
ecosystems. Arct. Anthropol. 11(1):31-40.
-------
Hoffman, F. 0., C. W. Miller, D. L. Shaeffer, and C. T. Garten, Jr.
1977. Computer codes for the assessment of radionuclides
released to the environment. Nucl. Saf. 18(3):343-354.

Hsu, S. B. , S. Hubbell, and P. Waltman. 1977. A mathematical theory
of single nutrient competition in continuous cultures of
microorganisms. SIAM J. of Appl. Math. 32:366-383.

Innis, G. S. 1972. Simulation Models of Grasslands and Grazing
Lands. Preprint No. 41, Grassland Biome, Natural Resource
Ecology Laboratory, Colorado State University, Fort Collins,
Colorado.

Kercher, J. R. , and H. H. Shugart, Jr. 1975. Trophic structure,
effective trophic position, and connectivity in food webs. Am.
Nat. 109(966):191-206.

Killough, G. G., and L. R. McKay. 1976. A methodology for
calculating radiation doses from radioactivity released to the
environment. ORNL 4992.

Killough, G. G. 1980. A dynamic model for estimating radiation dose
to the world population from release of 14C to the atmosphere.
Health Phys. 38:296-300.

Kimura, M., and T. Ohta. 1971. Theoretical aspects of population
genetics. Monographs in Population Biology 4, Princeton
University Press, Princeton, New Jersey.

Lane, P. A. , and R. Levins. 1977. The dynamics of aquatic
ecosystems, 2. The effects of nutrient enrichment on model
plankton communities. Limnol. Oceanogr. 22(3):454-471.

Lassiter, R. R. , G. L. Baughman, and L. A. Burns. 1978. Fate of
toxic organic substances in the aquatic environment. pp.
219-246. In: State of the Art in Ecological Modelling, Vol. 7.
Proceedings of the Conference on Ecological Modeling, S. E.
Jorgensen, ed. International Society for Ecological Modelling,
Copenhagen, Denmark.

Lettenmaier, D. P., and J. E. Richey. 1978. Ecosystem Modeling: A
Structural Approach. J. Environ. Eng. Div., Proc. A.S.C.E.
104:1015-1021.

Levins, S. A. 1974. Dispersion and population interaction. Am. Nat.
108:207-228.

Levins, R. 1968. Evolution in Changing Environments. Monographs in
Populations Biology 2. Princeton University Press, Princeton,
N.J. 120 pp.
-------
45
Levins, R. 1979. Coexistence in a variable environment. Am. Nat.
114:765-783.

Levins, R. 1974. The qualitative analysis of partially specified
systems. Ann. N.Y. Acad. Sci. 231:123-138.

Mielke, D. L., H. H. Shugart, Jr., and D. C. West. 1978. A stand
model for upland forests of Southern Arkansas. Oak Ridge
National Laboratory. ORNL/TM-6225.

Mitchell, K. J. 1975. Dynamics and Simulated Yield of Douglas-fir.
Forest Sci. Monograph 17:1-39.

Mogenson, B., and S. E. Jorgensen. 1979. Modelling the distribution
of chromium in the Danish Firth. In: First International
Conference on the State of the Art in Ecological Modelling, S.
E. Jorgensen, ed. Copenhagen, Denmark.

O'Neill, R. V. 1976. Ecosystem persistence and heterotrophic
regulation. Ecology 57:1244-1253.

O'Neill, R. V., R. A. Goldstein, H. H. Shugart, Jr., and J. B. Mankin.
1972. Terrestrial ecosystem energy model. EDFB-MR-72-19. 38pp.

Park, R. A., L. S. Clesceri, C. I. Connolly, H. F. Herbrandson, J. R.
Loehe, S. Ross, D. D. Sharma, and W. W. Shuster. Documentation
of PEST: A model for evaluating the fate of toxic substances in
aquatic environments. Unpublished draft. Available from R. A.
Park, Center for Ecological Modeling, Rensselaer Polytechnic
Institute, Troy, New York.

Park, R. A., R. V. O'Neill, J. A. Bloomfield, H. H. Shugart, R. S.
Booth, R. A. Goldstein, J. B. Mankin, J. F. Koonce, D. Scavia,
M. S. Adams, L. S. Clesceri, E. M. Coloni, E. H. Dettman, J.
Hoopes, D. D. Huff, S. Katz, J. F. Kitchell, R. C. Kohberger, E.
J. La Row, D. C. McNaught, J. Peterson, J. Titus, P. R. Weiler,
J. W. Wilkinson, and C. S. Zahorcak. 1974. A generalized model
for simulating lake ecosystems. Simulation 22:33-50.

Pennycuick, C. J. , R. M. Compton, and L. Beckingham. 1968. A
computer model for simulating the growth of a population or two
interacting populations. J. Theor. Biol. 18:316-329.

Rescigno, A., and I. W. Richardson. 1957. The struggle for life. I.
Two species. Bull. Math. Biophys. 29:377-388.

Scavia, D., B. J. Eadie, and A. Robertson. 1976. An ecological model
for the Great Lakes, pp. 629-633. In: Proc. Conf. Environ.
Modeling and Simulations, U.S. EPA. EPA 600/9-76-016.
-------
46
Shaeffer, D. L., C. W. Miller, and C. T. Garter, Jr. 1978.
Proceedings of a Workshop on the Evaluation of Models Used for
the Environmental Assessment of Radionuclide Releases,
CONF-770901. Oak Ridge National Laboratory, Oak Ridge,
Tennessee. 131 pp.

Shugart, H. H. , and I. R. Noble. 1980 (in press). A computer model
of succession and fire response of the high altitude eucalyptus
forests of the Brindabella Range, Australian Capital Territory,
Australia. J. Ecol.

Shugart, H. H., and D. C. West. 1976. Development of an Appalachian
deciduous forest succession model and its application to
assessment of the impact of the chestnut blight. J. Environ.
Manage. 5:161-179.

Shugart, H. H., and D. C. West. 1980. Forest succession models.
Bioscience 30:308-313.

Shugart, H. H., M. S. Hopkins, I. P. Burgess, and A. T. Mortlock.
1980 (in press). The development of a succession model for
subtropical rain forest and its application to assess the effects
of timber harvest at Wiangaree State Forest, New South Wales.
J. Environ. Manage.

Smith, J. H. , W. R. Mabey, N. Bohonos, B. R. Holt, S. S. Lee, T. W.
Chou, D. C. Bomberger, and T. Mill. 1977. Environmental
pathways of selected chemicals in freshwater systems, Part I.
EPA 600/7-77-113.

Steele, J. H. , and B. W. Frost. 1977. The structure of plankton
communities. Philos. Trans. R. Soc. London, Ser. B.
280:485-534.

Tetra Tech, Inc. 1979. Methodology for evaluation of multiple power
plant cooling system effects. VOL. I-IV. EPRI Report No.
EA-IIII. Electric Power Research Institute, Palo Alto,
California.

Tharp, M. L. 1978. Modeling major perturbations on a forest. M.S.
Thesis. Univeristy of Tennessee, Knoxville, Tennessee. 52pp.

Travis, C. C., W. M. Post, and D. L. DeAngelis. 1980 (in press).
Analysis of compensatory Leslie matrix models for competing
species. Theor. Pop. Biol.

Waide, J. B., and J. R. Webster. 1976. Engineering systems analysis;
applicability to ecosystems, pp. 329-371. In: Systems Analysis
and Simulation in Ecology, Vol. 4, B.C. Patten, ed. Academic
Press, New York.
-------
47
Walters, C. J. 1971. Systems ecology: The systems approach and
mathematical models in ecology, pp. 276-292. In: Fundamentals
of Ecology, E. P. Odum, ed. W.B. Saunders, Philadephia.

Woodwell, G. M., P. P. Craig, and H. A. Johnson. 1971. DDT in the
biosphere: Where does it go? Science 174:1101-1107.
-------
METHODS FOR MEASURING
EFFECTS OF CHEMICALS ON TERRESTRIAL
ECOSYSTEM PROPERTIES
January 15 and 16, 1980
Glenn W. Suter, II
Chairman

Environmental Sciences Division
Oak Ridge National Laboratory
-------
50

PARTICIPANTS

G. W. Suter, II, Chairman
WORKING GROUP A - EFFECTS OF CHEMICALS ON
MICROBIAL PROCESSES IN THE TERRESTRIAL SYSTEM
R. T. Belly (Group Leader) B. P. Spalding
Eastman Kodak Company Oak Ridge National Laboratory

D. A. Klein R. F. Strayer
Colorado State University Oak Ridge National Laboratory
WORKING GROUP B - MODEL ECOSYSTEMS
Peter Van Voris (Group Leader) J. D. Gile
Battelle Columbus Laboratories U.S. Environmental Protection Agency

R. V. Anderson D. R. Jackson
Western Illinois University Battelle Columbus Laboratories

N. T. Edwards D. W. Johnson
Oak Ridge National Laboratory Oak Ridge National Laboratory
OBSERVERS
J. V. Nabholz A. S. Hammons
U.S. Environmental Protection Oak Ridge National Laboratory
Agency
-------
51
SECTION 5

METHODS FOR MEASURING EFFECTS OF
CHEMICALS ON TERRESTRIAL ECOSYSTEM PROPERTIES
5.1 Introduction

This workshop considered the availability and utility of
multispecies laboratory systems to display the responses of ecosystem
level properties to toxic chemicals. The workshop was divided into
two working groups. Working group A discussed soil-microbe laboratory
systems and microbial processes. Working group B discussed model
ecosystems (i.e., systems that contain multiple trophic levels and
some diversity in physical structure). Three tasks were assigned to
each working group:

1. Identify laboratory microbial systems or model ecosys-
tems that display community and ecosystem level
properties and organize them into categories. These
categories should be sufficiently distinct to permit
generalization about their potential utility in
toxicity testing.

2. Evaluate the test systems in terms of the following
criteria:

Replicability

Standardization (potential for interlaboratory
transfer)

Sensitivity

Generality (to what range of ecosystems can the
results be applied?)

Equivocality of results

Statistical bases for interpreting results

Standards for rejecting results

Frequency of failure

Level of training and expertise required

Time required
-------
52

Cost per chemical

Appropriate positive controls

Other

3. Consider how these systems could be developed and
applied to toxicity testing. For example,

What test systems are sufficiently developed to
serve as test protocols?

What test systems require further development?

What types of development are needed: Development
of techniques, interlaboratory round robins
(validation), field verification, etc.?

What types of modeling or analysis would be
necessary to relate test results to the real
world, to generalize the results to a variety of
environments or to relate the parameters measured
to socially significant parameters?

What simpler tests could be confirmed by the
results of this multispecies test? What more
complex test might confirm the results of this
multispecies test?

The following sections present the results and discussions of the
working qroups. Although these sections are based primarily on
The following sections present the results and discussions of th
working groups. Although these sections are based primarily on
tten material produced during the workshop by participants, they
» been exnanded on the hasis nf notes and rernrdinas from the
two
written
have been expanded on the basis of notes and recordings from the
discussions.
5.2 Results and Discussion

5.2.1 Microbial Processes

Working group A reviewed laboratory testing and evaluation
procedures for assessing effects of chemicals on terrestrial microbial
processes. Basic microbial processes such as carbon, nitrogen, and
sulfur transformations are common to the nutrient cycles of all
ecosystems. Disruption or promotion of these processes may have
ecologically and socially significant impacts.

These discussions emphasized development of a balance between
proposing test methods that are practical to perform and ecologically
meaningful. The degree of environmental testing required to assess
the hazard of any chemical depends on its production, distribution,
-------
53

use, disposal, and fate in the environment. Because of the large
number of chemicals produced and the requirement of the Toxic
Substances Control Act (TSCA) for cost effective testing, a tiered
system involving different levels of testing complexity may be
required. Likewise, it is difficult to develop one test method to
accommodate the wide range of physical and chemical properties of
chemicals that need to be tested in the terrestrial environment. For
example, insoluble compounds may require a different set of test
conditions than their water soluble counterparts. In addition, the
binding behavior of chemicals in soil may play an important role in
determining observed toxic effects. Effects may not be observed until
the binding capacity of a soil is exceeded or the bound chemical is
released into solution by exchange processes. An understanding of
various physical and chemical properties of a chemical, as well as its
binding properties, is crucial to determining possible toxic effects
in the environment.

(1) Identification and evaluation of test systems. The general
criteria discussed in Section 5.1 were used to evaluate the methods
currently proposed by the EPA for measuring effects on cellulose
decomposition and nitrogen and sulfur transformations under TSCA (U.S.
EPA 1979). Other existing test systems were also evaluated and an
alternate test was proposed by workshop participants. These tests are
discussed in the following sections.

(a) Cellulose decomposition. The cellulose decomposition test
proposed by the EPA employs the measurement of C02 formation from an
axenic culture of Trichoderma longibrachiatum growing on cellulose as
a test substrate to determine effects on microbial carbon
mineralization. Because this test uses a pure culture and a well
defined substrate under controlled laboratory conditions, there are
apparent advantages in the potential reproducibility of this method
from laboratory to laboratory. In addition, such techniques have been
well studied.

Some disadvantages of the proposed cellulose decomposition test
are:

1. With an insoluble powdered substrate, degradation of
the substrate may be limited by surface area and
mixing.

2. The appropriateness of cellulose as a model
carbonaceous substrate is quite dubious. Although
cellulose is a major component of both plant litter and
soil organic matter, it normally occurs in a "masked"
form as an intimate physical and chemical association
with lignin.

This lignocellulose complex is much less susceptible to
microbial metabolism than purified cellulose itself.
-------
54
3. Trichoderma may or may not be the major organism
performing cellulose degradation in soil.

4. Although a single organism test may be very sensitive
to specific chemicals, its sensitivity or responses
would probably not be representative of the general
microbial community.

(b) Nitrogen transformation. The EPA proposed method for
testing the effects of new chemicals on nitrogen transformation in the
soil measures the production of ammonia from the microbial
decomposition of urea. The advantages of this test are that natural,
multispecies soil communities are used, instead of pure cultures, as
recommended in the cellulose decomposition and sulfate reduction
tests. However, we believe that urea is not an ideal choice of
organic N substrate. Extracellular soil urease activity can be quite
high and can persist in soils in which microbial activity has been
eliminated. Thus, ammonification of urea can continue after ureolytic
microorganisms have been inhibited by a toxic substance. In other
words, the production of ammonia from urea is not a growth-related
process and, thus, may be unaffected by a toxic chemical. A second
criticism is that an enrichment for ureolytic microorganisms may occur
with time and these microorganisms may differ in their tolerance to
the test substances when compared to the indigenous microflora in soil
not amended with urea.

Furthermore, ammonia may be oxidized to nitrite or nitrate,
making interpretation of results, based solely on the measurement of
ammonia, impossible.

(c) Sulfur transformation. The method proposed by the EPA for
determining possible effects on microorganisms in the sulfur cycle
involves a semi-quantitative measure of H2S production by a pure
culture of Desulfovibrio desulfuricans in the presence and absence of
a test substance. Sulfate reduction is a major process only in
anaerobic zones of flooded soils and sediments. In most soil systems,
sulfate reduction has only minimal effects on plant productivity
(Brock 1966; Russel and Russel 1950).

The proposed method for measuring sulfate reduction is
non-quantitative and is based on a single enzymatic
oxidation-reduction activity in a single microorganism. This method
largely ignores the complex interactions and diversity that can occur
in natural microbial populations. It is doubtful that the results of
this technique can possibly be extrapolated to the environment.
Furthermore, some microorganisms can convert organic sulfur to sulfate
directly without the intermediate production of H2S (Alexander 1961).

Although more expensive and time consuming, a more accurate
approach would be to use 35S to determine possible effects on the
major steps of the S cycle. Such radioisotope methods have been
-------
55
applied to both the soil and sediment system (for review see Brock
1966).

(d) Other existing test systems. Many experimental techniques
are used to measure specific or general microbial activities in soils
and sediments. A summary of the advantages and disadvantages of many
of these methods is provided in Table 5.1. A numerical forced-choice
rating system (0 to 4) was employed to estimate various attributes of
each test. At present there appear to be no protocols available for
testing effects on microbial processes. In estimating the utility of
each potential test, generality, nearness to validation, and
simplicity were emphasized. Less emphasis was placed on sensitivity
because this property is generally unknown for any broad range of
chemicals. Similarly, suitable positive controls are generally
lacking for all tests; their development will be the best approach to
estimating the sensitivity of a test.

Test systems that directly or indirectly measure carbon and
nitrogen mineralization were carefully evaluated because of their
importance in most ecosystems. Direct carbon cycle tests include the
evolution of C02 from soil with and without amendment of substrates.
Direct nitrogen mineralization tests include the production of
inorganic nitrogen from indigenous organic matter or amended
nitrogeneous substrate. Most of the direct tests have been shown to
be valid measures of ecosystem microbial processes. Rates of C02
evolution and nitrogen mineralization in laboratory tests are well
correlated with field processes (e.g., Bunnel et al. 1977; Stanford
and Smith 1972). We believe that responses to toxic chemicals could
also be validated.

Oxygen uptake methods using Warburg techniques are commonly used
to measure microbial activity in soil (for description of methodology,
see Parkinson et al. 1971). As pointed out by Stotzky (1965), carbon
dioxide evolution is more appropriate than oxygen uptake as a measure
of microbial respiration in soil. Among the limitations of oxygen
uptake methods are: (1) gases other than C02 may be evolved as a
result of microbial activity and may interfere with the manometric
measurement; (2) for oxygen uptake to be an accurate reflection of
soil respiratory activity, the soil environment must be completely
aerobic; (3) under anaerobic conditions, carbon dioxide is evolved
without oxgyen uptake, and non-biological factors may interfere; and
(4) chemicals acting as uncouplers of oxidative phosphorylation may
produce erroneous results.

Problems also occur when C02 evolution measurements are used as
indicators of microbial activity. These include the non-biological
production of C02 through chemical decarboxylation, cell-free enzymes,
or from free carbonates in soil. Phototrophic C02 fixation can be
minimized by incubating samples in the dark, although in some soils
chemoautotrophic C02 fixation may pose a problem.
-------
56
TABLE 5.1. EVALUATION OF MAJOR TEST SYSTEMS FOR EVALUATING EFFECTS OF MICROBIAL PROCESSES3
Assay test
I. Aerobic & General
1. 02 consumption
2. Dehydrogenase assay
3. Glucose l*C mineralization
4. Enzymatic activities
a. Protease
b. Cellulose
c. Amylase
d. Phosphatase
e. Arylsulfatase
5. Soil /substrate
Additive - C02
a. Glucose
b. Cellulose
c. Starch
d. Protein
e. Plant material
f. Soil humic material
6. Heat evolution
7. Microbial biomass
a. Counts
b. Chlorophyll
c. ATP
8. Nitrification
9. Anmonification
a. Urea
b. Protein
c. Plant material
10. Nitrogen fixation
II. Anaerobic
1. Sulfate reduction with
Desulfovibrio
2. Gas-generation (CH4)
Biogeochemical cycle
CHO N P S
X
X
X

X X
X
X
X
X

X
X
X
X
XX X
XX X
XX X
X
X
X
X
X
X
X
X
X
X

X
X
Comments
Too complex and difficult to measure
Possible toxicity, interpretation problems
Cost and apparent operator qualifications

Possible lack of sensitivity
Possible difficulty in establishing
Unknown validity
Unknown validity
Unknown validity

Unknown validity/enrichment
Unknown validity/enrichment
Unknown validity/enrichment
Unknown validity/enrichment
Probable validity
Poor sources commercially
Too complex/costly

No meaning
Surface soils only (no generality)
Cost and apparent operator qualifications
Too sensitive/secondary microbial process

Poor generality/validity
More validity
Most validity
Good validity

Minor biogeochemical importance
Cumbersome and space consuming
"Higher numbers indicate generally desirable attributes and lower numbers indicate disadvantageous attributes.
High values for time, operator skill, and cost were taken to be advantageous (i.e., short-time, low skill or low cost).
-------
57
TABLE 5.1 (continued)
Assay test Repllcability
Standard-
ization
Sensitivity
Generality
Equivo-
cality
Statistical
I. Aerobic & General
1.
2.
3.
4.

6.
7.

8.
9.

10.
02 consumption
Oehydrogenase assay
Glucose 14C mineralization
Enzymatic activities
a. Protease
b. Cellulose
c. Amulase
d. Phosphatase
3. Arylsulfatase
Soil /substrate
Additive - C02
a. Glucose
b. Cellulose
c. Starch
d. Protein
e. Plant material
f. Soil humic material
Heat evolution
Mlcrobial biomass
a. Counts
b. Chlorophyll
c. ATP
Nitrification
Ammonifi cation
a. Urea
b. Protein
c. Plant material
Nitrogen fixation
2
2
4

1
1
1
1
1

2
2
2
2
2
2
1

1
2
2
3

3
3
2
4
1
3
3

2
2
2
2
2

3
3
3
3
2
2
1

1
2
1
2

3
3
2
3
3
0
3

0
0
0
0
0

2
2
2
2
1
1
0

1
0
0
4

2
2
2
3
3
2
3

1
1
1
1
1

3
3
3
3
3
2
3

1
0
2
1

1
2
3
2
2
1
3

2
2
2
2
2

2
2
2
2
2
2
2

1
1
1
2

2
2
2
2
0
0
0

0
0
0
0
0

0
0
0
0
0
0
0

0
0
0
0

0
0
0
0
II. Anaerobic
1. Sulfate reduction with
Desulfovibrio
2. Gas-generation (CH4)
-------
58
TABLE 5.1 (continued)
Assay test
I. Aerobic & General
1. O2 consumption
2. Dehydrogenase assay
3. Glucose 14C mineralization
4. Enzymatic activities
a. Protease
b. Cellulose
c. Amylase
d. Phosphatase
e. Arylsulfatase
5. Soil /substrate
additive - C02
a. Glucose
b. Cellulose
c. Starch
d. Protein
e. Plant material
f. Soil humic material
6. Heat evolution
7. Microbial biomass
a. Counts
b. Chlorophyll
c. ATP
8. Nitrification
9. Ammonifi cation
a. Urea
b. Protein
c. Plant material
10. Nitrogen fixation
II. Anaerobic
1. Sulfate reduction with
2. Gas-generation (CH.)
Rejection
standards
0
0
0

0
0
0
0
0

0
0
0
0
0
0
0

0
0
0
0

0
0

0
0
Failure
frequency
0
0
0

0
0
0
0
0

0
0
0
0
0
0
0

0
0
0
0

0
0
Training
expertise
requirement
2
2
1

2
2
2
2
2

3
3
3
3
3
3
1

2
2
1
3

3
3
3
2

2
2
Time
requirement
3
2
3

2
2
2
3
3

3
3
3
3
2
2
2

1
3
2
1

2
2
2
2

1
1
Cost per
test
2
3
1

2
2
2
2
2

3
3
3
3
3
3
1

1
3
1
2

2
2
2
2

2
2
Positive
controls
0
0
0

0
0
0
0
0

0
0
0
0
0
0
0

0
0
0
0

0
0
-------
59
Indirect tests include various enzymatic assays, ATP assays, heat
production, and estimations of microbial populations. Most of the
indirect tests have not yet been shown to be generally valid measures
of microbial processes or generalizable to different ecosystems. As
pointed out by Parkinson et al. (1971), the use of the dehydrogenase
assay has not been considered a useful quantitative method for
assessing metabolic activity of microorganisms in soil because it is
reputed to have less than 5% of the efficiency of oxygen uptake
measurements. However, some workers have provided data that indicate
this method can provide rough comparative estimates of microbial
activity. Other tests based on enzymatic activity (e.g., protease,
cellulase, amylase phosphatase, and arylsulfatase - see Table 5.1)
have limitations because many extracellular enzymes can persist in
soil in which microbial activity is inhibited; these tests are not
based on growth related processes. Because of these limitations, we
concluded that such tests would have dubious value as indicators of
effects of chemicals in the natural environment.

Tests that include radioisotopic methods (e.g., glucose 14C
mineralization) are considered to have limited use because of cost and
training requirements associated with use of a radioactive substrate.
Likewise, the usefulness of tests based on heat evolution is limited
because they are time-consuming and difficult to perform.
Measurements based on microbial biomass (counts, chlorophyll, ATP) are
difficult to perform and interpret.

To summarize discussions on currently available methods, it
appears that the measurement of C02 evolution is the most practical
and meaningful test for measuring the effects of chemicals on
microorganisms involved in the carbon cycle.

(e) Proposed alternate tests. Group A proposed a test system to
measure the effects of chemicals on carbon and nitrogen
mineralizations simultaneously using environmentally relevant, high
nitrogen substrates and mixed microbe populations that can be
manipulated easily by technicians with minimal training. The test
uses a soil system and can be modified for aerobic or anaerobic
testing.

The basic test design involves the use of either a standardized
protein (Pharmamedia or a similar product) or ground alfalfa meal as a
substrate amendment and measurement of the following parameters at
appropriate time intervals under aerobic or anaerobic conditions:

Assay Aerobic Anaerobic

C02 evolution * *
NH4 * *
N02 and N03 *
-------
60
For this set of assays the following matrix of treatments would
be used:

Substrate
Test chemical
a b

c d
The characteristics of the proposed substrates are summarized in
Table 5.2. Flasks or small specimen bottles, up to 500-800 mL
capacity, were suggested for use in the assay. Approximately 30 to 40
g of wet soil could be used in each test flask. The soil should be at
approximately 60 to 70% of water holding capacity in a layer not more
than 10 mm in depth to assure that oxygen diffusion will not be
limited in the nominally aerobic system. The jar should be set up to
contain a small vial of alkali that can be exchanged at desired assay
times if the same jar will be used for sequential assays. Sequential
sampling of soil from the same flask should be avoided because it
interferes with C02 measurement.

For anaerobic systems, the 30 to 40 g of soil per test sample
could be placed in appropriate screw-cap top 30 x 70-mm test tubes
with sufficient water to water-log the samples and assure that the
systems will become anaerobic through normal oxygen depletion
processes. In addition, the soil-water mixture could be purged with
nitrogen to remove most of the dissolved oxygen.
-------
61
TABLE 5.2. CHARACTERISTICS OF PROPOSED SUBSTRATES
Standardized Protein
Advantages
Disadvantages
Can be purchased to standard
specifications - easily
reproducible.

Easily soluble, and no
subsequent direct mixing
problems.
Within limits, N and S levels
can be specified.
Responses are due to a limited
part of the microbial population.
Microbial responses may be too
rapid to allow usefulness in
detecting differences due to the
test chemicals.
Alfalfa
Advantages

Plant materials are represented
in an environmentally relevant
form.

C, N, and S cycling can be
monitored.
Disadvantages
Insolubility may lead to inhomo-
geneous dispersion.
Grinding and mixing may influence
rates of activity and make inter-
pretation difficult.

Reproducibility of materials from
different sources may be difficult
to assure.
Any soil falling in the range of properties listed in
Section A-3.5 (d)(i) (U.S. EPA, 1979 p. 162F) would be appropriate for
the proposed microbial functions test. The soil should be either
moistened or dried to 60 to 70% of the water holding capacity.
Increments of the test substance should be added to the soil as water
solutions and mixed thoroughly with the soil to yield the desired soil
concentrations. If the test substance does not have adequate water
solubility, it can be dissolved in an appropriate organic solvent that
can then be added to the soil. Residual solvent would be evaporated.
When this approach is used, all controls should be treated with
equivalent volumes of the solvent alone. Soil water content should be
checked after solvent evaporation and readjusted if necessary.
Another strategy that can be used is to add the test chemical plus
-------
62
solvent to perhaps 10 to 25% of the test soil and, after solvent
evaporation, to mix this treated soil with a larger volume of regular
soil, using appropriate solvent controls.

As a screen, a single C02 assay can be performed at perhaps day 3
or 4 for the substrate-amended system, and days 7 to 11 for the plain
soil systems. This will require a total of 24 flasks per test
chemical (Table 5.3). For kinetic studies a schedule similar to that
shown in Table 5.4 that allows the C02 assays to be completed with
incubation in 12 days (2 work weeks) could be used. With this type of
schedule, a series of sequential tests could be conducted biweekly.
Per assay day, 12 or 24 assays would have to be completed. Inorganic
N or other mineral nutrient determinations would be made on two soil
samples at day 0 and on all test soil after the last C02 assay.
Additional replicates would be required if a positive control is used.
TABLE 5.3. NUMBER OF C02 MEASUREMENTS PER TEST
Assay
Sequential
Unamended Amended
Single
Assay times
Replications
Substrate amendment
Test substrate concentrations

3
2
1
6
36
6
2
1
6
72
1
2
2
6
24

TABLE 5.4. SCHEDULE FOR SEQUENTIAL SAMPLING: AEROBIC SYSTEM

Day
Substrate
amended

M
0
Setup

T
1
X

W
2
X

T
3
X

F
4
X

Time
S S M
567
X

T W T F
8 9 10 11
X

Substrate Setup
not amended
Alternate tests proposed by Working Group A employ natural soil
microflora, composed of many species, and substrate amendments that
-------
63
range from those that are easily decomposed (protein) to those that
are more persistent (plant materials and soil organic matter). Major
ecosystem processes that are monitored include carbon, nitrogen, and
possibly sulfur mineralization and nitrification. Tests that are
simpler than our test system include pure and mixed culture systems
with similar or probably less complex substrates. Our proposed tests
should confirm most negative results from simpler tests.

If a test chemical has no effect on C02 production from a variety
of substrates by a pure culture, the same result will nearly always be
obtained in a complex soil system. However, simpler systems may give
positive results that are not confirmed by the proposed test system.
In the simpler, pure culture systems, absorption and/or other
physico-chemical fates of the test substances are minimal. However,
in a more complex soil system, absorption of the test substance on
organic matter or inorganic soil components may effectively lower the
concentration of potential toxicants to levels that are no longer
inhibitory. Futhermore, in multispecies systems, succession or
adaptation to the potential toxicant may occur, resulting in little
noticeable changes in the levels of C or N mineralization. However,
it is possible that a microbial culture would partially degrade a test
substance into a more toxic form. Because this is more likely to
occur in a mixed culture, this system could be more sensitive than
single species cultures in a few cases.

Tests that involve use of a ground plant material and soil with
its natural microbial community, sacrifice some potential
reproducibility for clearer validity in extrapolating test results to
actual microbial processes. The use of positive controls in such
tests would facilitate comparisons between different laboratories that
use different soils or plant material.

(2) Protocol development. Development can be considered in
terms of standardization and validation. Considerable work has been
done on the laboratory measurement of microbial processes for
agronomic and ecological investigations and to a lesser extent for
studies of pesticides and other toxic substances. Despite this
considerable experience, none of these systems have been standardized
and shown to give comparable results among laboratories when used to
test the effects of chemical substances. Such standardization must
occur before any protocol is adopted. Validation studies are needed
for all of the proposed tests. Various definitions of validation
should be considered in development of these research strategies.

As noted in previous discussions and in Table 5.1, the use of
protein or ground alfalfa decomposition in a soil environment appears
to be well suited for use in evaluating the possible effects of test
materials on terrestrial microbial processes. This is based on the
ability to measure both carbon and nitrogen mineralization in the same
system and to evaluate the release of carbon and nitrogen from natural
substrates (protein and alfalfa) and from the native soil organic
-------
64

matter, processes that involve the functioning of varied groups of
soil microorganisms.

In addition to this test, as noted in Table 5.1, several other
assays are considered to be potentially useful. These include 14C
glucose mineralization, nitrification, and C02 evolution from a series
of general substrates, including glucose and cellulose. The 14C
glucose mineralization assay, in spite of its known sensitivity, will
probably not be useful because of its high cost, the needs for
specialized equipment and radioactivity control requirements, and the
needs for specialized training and clearances for technicians. Carbon
dioxide evolution from glucose could be considered; this involves a
soluble substrate, but a major concern with this substrate is that it
would only allow evaluation of carbon processing performed by a narrow
range of soil microorganisms. Cellulose, as noted earlier, is a
difficult substrate because of its insolubility and difficulties in
mixing it into test systems. In addition, pure cellulose is not a
common material in soil systems.

Assuming that the test systems proposed by Working Group A might
be considered for use in hazard evaluations, the following factors
will require further development:

I. Optimization of soil-substrate and soil-test substances
ratios.

2. Establishment of minimum amounts of C and N
mineralization needed to complete assay.

3. Optimization of incubation time for control and
substrate-amended systems.

4. Establishment of experimental variability and number of
replicates required.

5. Use of known toxic agents to evaluate sensitivity of the
assays.

6. Development and testing of a sampling schedule that will
be useful under a wide range of conditions.

7. Selection of a test incubation vessel of minimum cost
and simplifying the procedures of the test as much as
possible.

8. Evaluation of the utility of sequential versus single
assay analysis for the evaluation of C02 evolution,
ammonification, and nitrification.

9. Selection of the best measurement procedure for evolved
C02, ammonium, and combined nitrite and nitrate.
-------
65
5.2.2 Identification of Model Ecosystems

(1) Test descriptions. Model ecosystems were divided into two
categories, those that are synthesized in the laboratory and those
that are excised from natural systems. A reference and a brief
description of each system considered are provided below. The order
in which they are listed does not represent a ranking.

(a) Synthetic systems.

Coleman and Anderson. A 50-mL Erlenmeyer flask with 20 g of seived
sterilized soil and a defined (gnotobiotic) community including
species of bacteria, protozoa, and nematodes (Anderson et al. 1978).

H. T. Odum. A 16-cm diameter desiccator with soil, litter, one
transplanted bromeliad, fern, lichen, moss, and algal clump
(Odum 1970).

Lichtenstein. A 86-mm diameter, 1-L plastic cylinder containing
layers of treated and untreated soil and corn seedlings. Leachate is
collected (Lichtenstein et al. 1977).

Lighthart and Bond. A plastic lined, No. 300 can or 600-mL beaker
with 150-g homogenized soil and 15-g sifted litter (Lighthart and
Bond 1976).

MeteaIf and Cole. A 19-L, wide-mouth jar containing vermiculite, corn
seedlings, earthworms, isopods, slugs, saltmarsh caterpillars, and a
vole (Cole et al. 1976).

Nash and Beall. 150-cm-long, 115-cnrhigh, and 50-cm-wide, closed
glass box with a 15-cm layer of seived soil and crop plants. Air and
leachate are monitored (Nash et al. 1977).

Rosswall. A plastic pot with sterilized forest soil and litter, a
defined soil biota, and a pine seedling (Rosswall et al. 1978).

TMC. The terrestrial microcosm chamber consists of a glass box (1 m x
0.75 m x 0.61 m) with 20 cm of synthetic soil, alfalfa, ryegrass,
nematodes, earthworms, enchytraeid worms, isopods, mealworms,
crickets, snails, and a pregnant vole. Air and leachate are monitored
(Gile and Gillett 1979).

(b) Excised systems.

Grassland Core. A 15- or 30-cm diameter intact soil core with
vegetation retained, encased in heat shrunk plastic (Jackson et al.
1979; Van Voris et al. 1978).
-------
66

Outcrop. 90- x 90-cm sections excised from communities that develop
in depressions in granite or sandstone outcrops arranged in 1- x 6.5-m
concrete simulated outcrops (McCormick and Platt 1962).

Soil Core. A 5-cnrdiameter by 5 to 10-cm-deep, intact soil core with
vegetation clipped at ground level, encased in heat shrunk plastic
(Jackson et al. 1977).

Treecosm. An approximately 2-m-tall sapling in 45 cm x 45 cm x 25 cm
excised block of forest soil, sealed at the sides and contained in a
plywood box (Jackson et al. 1978).

(2) Measurable parameters. Table 5.5 shows the community and
ecosystem responses that have been or can be measured in each of these
systems. Net primary productivity is the only one of these parameters
that has direct social relevance. This parameter may be slow to
respond, however, if it is mediated by changes in reproduction, soil
fertility, or other intermediate factors.

A second category of parameters was considered relevant because
of their relation to primary production and because they are likely to
be worth measuring in model ecosystems. The first of these is loss of
mineral nutrients (total N, N03-N, NH3-N, Ca, P, K, S) and dissolved
organic carbon. The utility of this parameter as an early indicator
of ecosystem stress was hypothesized by O'Neill et al. (1977). It has
been partially confirmed by a series of microcosm experiments
conducted at Oak Ridge National Laboratory (ORNL) (Harris 1980) and
Corvallis Environmental Research Laboratory (CERL) (Gile et al. 1979).
A number of field studies indicate that nutrient dynamics are
sensitive to both perturbations and internal successional processes
(Jackson and Watson 1977; Likens et al. 1970; Richardson and
Lund 1975; Best and Monk 1975; and Vitousek et al. 1979).

Respiration is another potentially important and readily
measurable attribute of model ecosystems. The more common methods of
measuring respiration in terms of C02 output have technical problems.
Static absorption of C02 in KOH solution is inefficient unless
respiration rates are low; static absorption on soda lime is more
efficient but can cause drying of the soil. Use of a flow-through
system can increase the efficiency of KOH absorption and eliminate the
drying problem with soda lime, but both absorptive methods can have
low precision because of errors in titration or weighing. Infrared
gas analysis is precise and efficient, but has a high initial cost.
The problems associated with measuring ecosystem respiration are
discussed in more detail in the following references: Eckardt 1968;
Monteith 1968; Woodwell and Botkin 1970; Odum 1970; Edwards and
Soil ins 1973; Minderman and Vulto 1973; and Van Voris et al. 1978.

The ratio of carbon to major mineral nutrients was also
considered to be a potentially important indicator of system
disruption. Rates of nitrification and the mineralization of nitrogen
-------
67
TABLE 5.5. PARAMETERS MEASURABLE IN TERRESTRIAL MODEL ECOSYSTEMS

Test Systems

0)
r—
O C
O •!-
0)
+ +>
U)
>*- c
i— 01
TO -t->
Parameters and £ "5
Attributes* z 5
Diversity
Succession
Trophic level
interaction
Competition
Net primary
productivity Y Y
Nutrient
retention P P
Respiration Y
C/Nutrient
ratio P P
c
0
i.
0>
•a
c
^c

^
^~
C r—
TO TO
£ X
J=. O r— U)
TO Z O 0
•z *— u ce.
P P

P Y Y P

Y Y Y Y

P P Y Y
P P P

P P Y P

+•> -P
a> 3
o

u at

£
-------
68

and other mineral nutrients are tied to carbon dynamics (Cairns 1963;
Johnson and Edwards 1979; and Johnson and Edwards 1980). Thus,
simultaneous measurement of C02 production and mineral nutrients may
supply a more sensitive and explanatory indicator of toxic response
than either parameter alone. Mineral nutrients may be measured in
leachate or in soil subsamples.

A third category of parameters consisted of those that could
possibly be measured in model ecosystems but were not recommended
because they present technical problems and are not necessary for the
understanding of ecosystem response. ATP and enzyme assays are
technically difficult, and the results are not easily interpreted.
Community characteristics such as diversity and succession could only
be measured with microbes or micro-invertebrates because of space and
time limitations. These groups present severe taxonomic problems and
the analysis would be difficult and time consuming. Interactions
between specific species, such as competition and predation, are
limited to the same groups of organisms and present the same problems
unless simple gnotobiotic systems such as Coleman and Anderson's are
used.

5.2.3 Evaluation of Model Ecosystems

(1) Synthetic systems.

Coleman and Anderson. This system is rated good for replicability,
standardization, and cost of maintenance. It is rated low for level
of training and frequency of failure because of the difficulty of
maintaining gnotobiotic conditions. Its ability to represent real
ecosystems is questionable. It may have potential for assessing
effects on specific microorganismic couplings.

Lichtenstein. This system is rated very good for testing effects on
primary productivity; it has good potential for testing effects of
nutrient loss from soil. Its rating is good for replicability,
standardization, cost, and level of training.

Lighthart. This system is rated good for replicability,
standardization, level of training, and cost. It is rated low for
applicability to testing ecosystem effects.

Metcalf and Cole. The utility of this system is limited in terms of
substrate realism and validity of productivity measurements. It is
not well designed for determination of effects on ecosystem processes.
It is rated good for replicability, standardization, cost, and level
of training required.

Nash. This system is good for measuring productivity. It has good
replicability and standardization. It is rated low in terms of cost
and size. It is not applicable in its current design for testing
ecological effects.
-------
69

H. T. Odum. This system is rated moderate for replicability. The
level of training required is low, and it is low in cost. Its
applicability to toxicants is good. It has a relatively high failure
rate because of transplantation.

Rosswall. This system is rated good for replicability and
standardization. It is limited in terms of small size and
productivity. It is rated good in level of training required and
cost.

TMC. This system is rated good for testing effects on competition and
productivity, and potential for revealing effects on soil processes.
It is limited by cost and size, and to some extent by level of
training required. It is rated good for replicability and
standardization.

In general, artificial soil substrates may increase replicability
and standardization at the expense of application to testing effects
on processes in real ecosystems.

Homogenized soil systems are most applicable to agricultural
ecosystems and less applicable to uncultivated ecosystems. They are
also more replicable and more easily standardized than intact
microcosms.

(2) Excised systems. Grassland microcosms are considered more
desirable than forest systems for testing purposes because of less
restriction on size. However, size may be most limited by the need
for expensive environmental control systems and labor costs in
excision. Minimum sizes suggested were 10 to 20 cm in diameter and no
less than 15 cm in depth.

In excised systems, replicability and standardization are
difficult problems that need additional study before definitive
recommendations can be made. In general, replicability should improve
as size increases.

The measurements of ecosystem processes in these systems can be
standardized, but the test systems themselves are defined by the
ecosystems from which they are excised.

(3) Uncertainties concerning all model ecosystems.

Size. Despite some work on the effects of size on the responses of
soil cores (Harris 1980), the effects of system size on most system
types and parameters is unknown.

Soil Types. The physical and chemical properties of soils profoundly
affect the fate and toxicity of chemicals. It is not clear that a
sufficiently limited number of regionally representative soil types
can be defined.
-------
70
Boundary Effects. It is not clear whether the boundaries of a model
ecosystem, which contrast with the continuity of natural ecosystems,
affect the responses observed.

Construction. The container for the model ecosystem can affect system
response by adsorption, channelization, and release of constituent
chemicals into the soil.

Variance. The trade-offs involved in reducing variance by using
larger systems or artificial substrates have not been defined.

Field Verification. While the utility of model ecosystems for studies
of the fate of chemicals has been field validated, almost no work has
been done to field-verify toxic responses observed in model ecosystems
(Jackson et al. 1979).

Equivocality. The relationship of the parameters measured in model
ecosystems such as C02 and nutrient output to socially relevant
parameters such as primary production is not well defined.

5.2.4 Protocol Development for Model Ecosystems

None of the test systems are sufficiently developed to serve as
test protocols. Nevertheless, the following test systems are worthy
of further development:

Coleman and Anderson. These defined (gnotobiotic) systems are more
likely to have explainable results than the undefined systems. Level
of training and rate of failure resulting from gnotobiotic technology
are impediments to their use in routine testing. These systems may
have the potential for overcoming the disadvantages of the homogenized
and excised systems as they are currently conceived.

Intermediate-si zed grassland microcosms. Homogenized and excised
microcosms of this type come close to meeting many of the criteria
(size, cost, level of training, replicability, etc.) addressed at this
workshop. However, further developments are needed in terms of
round-robin analysis, field validation, and replicability. In
addition, C02 analysis should be refined further. Large excised
systems such as the excised tree microcosms and rare systems such as
McCormick's outcrop microcosms may serve as excellent research tools
for ecosystem processes. However, the size and expense of these
systems may limit their usefulness for testing chemicals.

The small microcosms (5x5 cm) offer advantages in screening
chemicals on the basis of the number of units that can be examined per
unit cost. However, their inability to sustain primary production and
the small volume of soil involved seriously limits the interpretation
of results.
-------
71

5.3 Conclusions

5.3.1 Microbial Processes

Any test for predicting terrestrial ecosystem effects on
microorganisms should be as close to the natural system as reasonably
possible. Certain pure culture systems could be developed that are
more sensitive to toxic effects than the whole soil population.
However, these pure culture systems ignore the complex interactions
occuring in soil and the genetic variability and diverse nature of
soil microbial populations. Therefore, it is extremely difficult, if
not impossible to extrapolate the results of such pure culture systems
to the natural situation.

Tests using natural soils were considered more meaningful than
tests using soil suspensions that disturb the soil structure and sub-
structure. Furthermore, soil slurry systems are very sensitive to
oxygen transfer effects. Therefore, tests based on soil slurry
systems may be difficult to standardize.

The natural system is not static but rather a dynamic system
undergoing constant change. With the exception of nitrification, all
microbial processes in the carbon, nitrogen, and sulfur cycles are
performed by a diverse group of microorganisms. Because of this
genetic diversity and the ability of microbial populations to adapt to
the presence of a chemical, effects may be only temporary. Therefore,
to mimic the natural system, kinetic (sequential) testing, rather than
static testing, should be employed. This is not to say that a single
measurement may not be useful as an initial screen for detecting
possible inhibitory effects to microorganisms. However, to
extrapolate the meaning of this single measurement to the environment,
kinetic studies are required. Preferably, these kinetic studies would
initially be performed in the laboratory and followed by field
studies.

In addition to using natural soil microorganisms and the soil
matrix as a test system, we concluded that measurement of effects of
general metabolic processes occurring in the whole population or
community (e.g., C02 formation, 02 consumption) would be more
meaningful than results from more selective tests based on a single
enzymatic criterion (e.g., sulfatase, phosphotase, amylase). Because
many of these extracellular enzymes can function in soil even after
cell death, the results of these tests are difficult to interpret.

Considerable effort has been directed at studying the ability of
microorganisms to transform chemicals in the terrestrial environment.
In contrast, there are only a few studies defining the effects of
chemicals on the microorganisms involved in the carbon, nitrogen, and
sulfur cycles. Because of the genetic diversity of microorganisms,
their ability to detoxify or polymerize certain pollutants, and the
binding of pollutants to soil, it remains to be determined whether
-------
72
microorganisms in the terrestrial environment are sensitive indicators
of the effects of chemicals. The development of suitable techniques
for evaluating the effects of chemicals on microorganisms and their
use to test compounds such as pesticides and polychlorinated biphenyls
that have well established toxicity to other forms of life will lead
to an answer to this question by providing a basis for comparison.

5.3.2 Model Ecosystems

No model ecosystem, synthetic or excised, is considered ready to
serve as a test protocol. Nevertheless, defined systems such as
Coleman and Anderson's and intermediate-sized grassland microcosms are
recommended for further development. In addition, the following
research activities are recommended:

1. Evaluation of encasement materials in terms of
leachabiltiy, adsorptive capacity, optical properties,
and durability.

2. Round-robin evaluation in all cases.

3. Field validation in all cases.

4. Determination of the relevance of measured parameters
(nutrient loss) to major ecosystem processes.

5.3.3 General Discussion

During the general discussion that concluded the workshop, a
number of common concerns were raised. One was the degree of realism
required in a test system. Although we agreed that a minimum degree
of realism is necessary, increasing realism involves greater cost and
technical difficulty. The larger number of physical and biological
elements in more realistic systems tends to reduce replicability.
They also reduce sensitivity by increasing absorption and degradation
of the test substance and by increasing the functional redundancy in
the system. Finally, the presence of multiple biological elements can
interfere with the interpretation of results in terms of the processes
involved. For example, the presence of plants creates a more
realistic environment for soil microbes, but plant respiration and
nutrient uptake interfere with the measurement of mineralization and
microbial respiration.

Therefore, it is necessary to determine whether a particular
simplification of the test system will cause a qualitative difference
in response. Brian Spalding suggested that the many years of
experience in agronomic microbiology indicate that sieved soil
adequately represents the dynamics of major nutrients in the field.
However, the toxicant responses of whole natural ecosystems have not
been well studied. Peter Van Voris suggested that the ecological
importance of interactions between microbes, animals, plants, soil,
and litter makes them worth considering in a testing scheme.
-------
73

A second major issue is the effect of variation in the chemical,
physical, and microbiological properties of soil on responses to
chemicals. We agreed that, while this problem is difficult and highly
significant, the direct approach of providing a standard test soil is
impractical. The best alternative is to set limits on the major
physical and chemical properties of test soils and then to use
positive control substances to determine the relative sensitivity of
each test soil. One must also recognize that variation in soil
properties can affect the choice of response parameters. For example,
nitrate loss was found to be a good indicator of toxicant response for
soil cores from Oak Ridge but not for cores from Corvallis (Gile et
al. 1979).

Finally, the problem of dose delivery and scheduling was briefly
considered. Because TSCA controlled substances, unlike pesticides,
are usually not deliberately released into the environment, the proper
mode of delivery is not obvious. Test substances could be delivered
in sprays, irrigation water, organic solvents, particulates, or
vapors.

Substances could be delivered in a single acute dose or in
continuous or episodic chronic doses. The mode of dose delivery can
significantly affect the outcome of a test and, therefore, must be
carefully considered in the development of a hazard assessment scheme.
-------
74

5.4 References

Alexander, M. 1961. Introduction to Soil Microbiology. John Wiley
and Sons, New York.

Anderson, R. V., E. T. Elliot, J. F. McClellan, D. C. Coleman, C. V.
Cole, and H. W. Hunt. 1978. Trophic interactions in soils as
they affect energy and nutrient dynamics. Ill Biotic
interactions of bacteria, amoebae, and nematodes. Microb. Ecol.
4:361-371.

Best, G. R., and C. C. Monk. 1973. Cation flux in hardwood and white
pine watersheds, pp. 847-861. In: Mineral Cycling in Temperate
Forest Ecosystems, F. G. Howel, J. B. Gentry, and M. H. Smith,
eds. CONF-740513. National Technical Information Service,
Springfield, Virginia.

Brock, T. D. 1966. Principles of Microbial Ecology. Prentice Hall,
Englewood Cliffs, New Jersey.

Cairns, R. R. 1963. Nitrogen mineralization in solonetzic soil
samples and some influencing factors. Can. J. Soil Sci.
43:387-392.

Cole, L. K., R. L. Metcalf, and J. R. Sanborn. 1976. Environmental
fate of insecticides in terrestrial model ecosystems. Intern. J.
Environ. Stud. 10:7-14.

Eckardt, F. E. 1968. Techniques de mesure de la phtosynthese sur la
terrain basees sur 1'emploi d1enceintes climatisees. pp.
289-319. In: Functioning of Terrestrial Ecosystems at the
Primary Production Level, F. E. Eckardt, ed. Proceedings UNESCO
Copenhagen Symposium. UNESCO, Copenhagen.

Edwards, N. T. , and P. Sollins. 1973. Continuous measurement of
carbon dioxide evolution from partitioned forest floor
components. Ecology 54:406-412.

Gile, J. D., J. C. Collins, and J. W. Gillett. 1979. The soil core
microcosm - a potential screening tool. EPA-600/3-79-089.
Corvallis Environmental Research Laboratory, Corvallis, Oregon.

Gile, J. D., and J. W. Gillett. 1979. Terrestrial microcosm chamber
evaluations of substitute chemicals. pp. 75-85. In:
Terrestrial Microcosms and Environmental Chemistry, J. M. Witt
and J. W. Gillett, eds. National Science Foundation, Washington,
D.C.

Harris, W. F. 1980. Microcosms as potential screening tools for
evaluating transport and effects of toxic substances: Final
report. ORNL/EPA-4. Environmental Sciences Division, Oak Ridge
National Laboratory, Oak Ridge, Tennessee.
-------
75
Jackson, D. R., B. S. Ausmus, and M. Levine. 1979. Effects of
arsenic on nutrient dynamics of grassland microcosms and field
plots. Water Soil Air Pollut. 11:13-21.

Jackson, D. R., J. J. Selvidge, and B. S. Ausmus. 1978. Behavior of
heavy metals in forest microcosms: I. Transport and
distribution among components. Water Soil Air Pollut. 10:3-11.

Jackson, D. R., C. D. Washburne, and B. S. Ausmus. 1977. Loss of Ca
and N03-N from terrestrial microcosms as an indicator of soil
pollution. Water Soil Air Pollut. 8:279-284.

Jackson, D. R., and A. P. Watson. 1977. Disruption of nutrient pools
and transport of heavy metals in a forested watershed near a lead
smelter. J. Environ. Qual. 6:331-338.

Johnson, D. W. , and N. T. Edwards. 1979. The effects of stem
girdling on biogeochemical cycles within a mixed deciduous forest
in eastern Tennessee II. Soil nitorgen mineralization and
nitrification rates. Oecologia 40:259-271.

Johnson, D. W. , and N. T. Edwards. 1980. Nitrogen mineralization,
immobilization and nitrification following urea fertilization of
a forest soil under field and laboratory conditions. J. Soil
Sci. Soc. Am. 44:610-616.

Lichenstein, E. P., K. R. Schultz, and T. T. Liang. 1977. Fate of
fresh and aged soil residues of the insecticide (14C)-N-2596 in a
soil-corn-water ecosystem. J. Econ. Entom. 70:169-175.

Lighthart, B., and H. Bond. 1976. Design and preliminary results
from soil/litter microcosms. Int. J. Environ. Stud. 10:51-58.

Likens, G. E., F. H. Bormann, N. M. Johnson, D. W. Fisher, and R. S.
Pierce. 1970. Effects of forest cutting and herbicide treatment
on nutrient budgets in the Hubbard Brook watershed-ecosystem.
Ecol. Mono. 40:23-47.

McCormick, J. F., and R. B. Platt. 1962. Effects of ionized
radiation on a natural plant community. Radiat. Bot. 2:161-188.

Minderman, G., and J. C. Vulto. 1973. Comparison of techniques for
the measurement of carbon dioxide evolution from soil.
Pedobiologia 13:73-80.

Monteith, J. L. 1968. Analysis of the photosynthesis and respiration
of field crops from vertical fluxes of carbon dioxide.
pp. 349-358. In: Functioning of Terrestrial Ecosystems at the
Primary Production Level. F. E. Eckardt, ed. Proceedings of the
UNESCO Copenhagen Symposium, UNESCO, Copenhagen.
-------
76
Nash, R. G., M. L. Beall, Jr., and W. G. Harris. 1977. Toxaphene and
1, 1, 1 trichloro-2,2-bis (p-chlorophenyl) ethane (DDT) losses
from cotton in an agroecosystem chamber. J. Agric. Food Chem.
25:336-341.

O'Neill, R. V., B. S. Ausmus, D. R. Jackson, R. I. Van Hook, P. Van
Voris, C. Washburne, and A. P. Watson. 1977. Monitoring
terrestrial ecosystems by analysis of nutrient export. Water Air
Soil Pollut. 8:271-277.

Odum, H. T. , and A. Lugo. 1970. Metabolism of forest-floor
microcosms, pp. 1-35 to 1-56. In: A Tropical Rain Forest. H.
T. Odum, ed. U.S. Atomic Energy Commission, Washington, D.C.

Parkinson, D., T. R. G. Gray, and S. T. Williams. 1971. Methods for
studying the ecology of soil microorganisms, IBP handbook no. 19.
Blackwell Scientific Publications, Oxford and Edinburgh.

Richardson, C. J, and J. A. Lund. 1975. Effects of clear-cutting on
nutrient losses in aspen forests on three soil types in Michigan.
pp. 673-686. In: Mineral Cycling in Temperate Forest
Ecosystems. F. G. Howe 11, J. B. Gentry, and M. H. Smith, eds.
CONF-74513. Technical Information Center, U.S. ERDA.

Rosswall, T., E. U. Lohm, and B. Sohbnius. 1978. Development d'un
Microcosme pour L1etude de la Mineralisation et de L'absorption
Radiculaire de L'asote dans L1humus d'un Foret de Coniferes
(Pinus sylvestris L.) Lejeunia Rev. Botanique, Novelle Serie No.
84. 26 pp.

Russell, E. J. , and E. W. Russell. 1950. Soil Conditions and Plant
Growth, 8th ed. Longmans, Green and Co., London.

Stotzky, G. 1965. Microbial respiration. pp. 1550-1572. In:
Methods of Soil Analysis, Part 2, Chemical and Microbial
Properties. C. A. Black, ed. American Society of Agronomy,
Inc., Madison, Wisconsin.

U.S. Environmental Protection Agency. 1979. Toxic Substances Control
Act premanufacture testing of new chemical substances
(OTS-050003; FRL-1069-1) Fed. Regist. 44(53):16240-16292.

Van Voris, P., R. V. O'Neill, H. H. Shugart, and W. R. Emanuel. 1978.
Functional complexity and ecosystem stability: An experimental
approach. ORNL/TM-6199. Oak Ridge National Laboratory, Oak
Ridge, Tennessee.

Vitousek, P. M., J. R. Gosz, C. C. Brier, J. M. Melillo, W. A.
Reiners, and R. L. Todd. 1979. Nitrate losses from disturbed
ecosystems. Science 204:469-474.
-------
77
Woodwell, G. M., and D. B. Botkin. 1970. Metabolism of terrestrial
ecosystems by gas exchange techniques: The Brookhaven approach.
pp. 73-85. In: Analysis of Temperate Forest Ecosystems. D. E.
Reichle, ed. Springer-Verlag, Berlin-Heidelberg-New York.
-------
METHODS FOR MEASURING EFFECTS OF CHEMICALS

ON AQUATIC ECOSYSTEM PROPERTIES
February 5 and 6, 1980
Jeffrey M. Giddings

Environmental Sciences Division

Oak Ridge National Laboratory
-------
80
PARTICIPANTS

J. M. Giddings, Chairman
Oak Ridge National Laboratory
GROUP A (LENTIC WORKING GROUP)
S. M. Adams Allen Medine
Oak Ridge National Laboratory University of Connecticut

B. G. Blaylock P. J. Mulholland
Oak Ridge National Laboratory Oak Ridge National Laboratory

J. W. Leffler Hap Pritchard
Ferrum College U.S. Environmental Protection Agency

Donald Levy J. B. Waide
University of California Oak Ridge National Laboratory
GROUP B (LOTIC WORKING GROUP)
John Bowling J. D. Newbold
Savannah River Ecology Lab. Oak Ridge National Laboratory

J. W. Elwood John Rogers
Oak Ridge National Laboratory North Texas State University

Alan Maki F. S. Sanders
Proctor and Gamble Oak Ridge National Laboratory
OBSERVERS
A. S. Mammons J. V. Nabholz
Oak Ridge National Laboratory U.S. Environmental Protection Agency
-------
81
SECTION 6

METHODS FOR MEASURING EFFECTS OF CHEMICALS
ON AQUATIC ECOSYSTEM PROPERTIES
6.1 Introduction

This workshop explored the potential uses of laboratory
experimental ecosystems for assessing the effects of chemicals on
aquatic ecosystem structure and function. The workshop objectives
were:

1. To produce as complete a list as possible of laboratory
experimental systems that have been (or could be) used to
measure effects of chemicals on aquatic ecosystem
properties.

2. To evaluate the capabilities of each model ecosystem for
measuring system-level properties.

3. To evaluate each model ecosystem as a tool for routine
screening of chemicals.

4. To outline logical test protocols for three different
types of model ecosystems.

The workshop participants were divided into two working groups
designated the Lentic Working Group (Group A) and the Lotic Working
Group (Group B). The Lentic Working Group considered nonflowing
freshwater model ecosystems and various marine systems; the Lotic
Working Group discussed model stream research. Each working group
addressed the following three general topics in a series of three
working sessions:

Working Session I: Identification and Description of Test Systems

1. Identify model aquatic ecosystems (Group A: lentic;
Group B: lotic) that display community and
ecosystem-level properties. If possible, organize
these model ecosystems into categories that are
sufficiently well defined to permit generalization
about their potential utility in hazard assessment
requirements of the Toxic Substances Control Act
(TSCA).

2. List community and ecosystem-level properties
measurable in these model ecosystems.
-------
82
3. Construct a matrix with test systems on the horizontal
axis and measurable properties on the vertical axis.
For each intersection in the matrix, assess the
feasibility of measuring that property on that model
ecosystem.

Working Session II: Evaluation of Test Systems

1. Construct a matrix with test systems on the horizontal
axis and experimental criteria on the vertical axis.
For each intersection in the matrix, evaluate that test
system in terms of that criterion. The criteria are:

Replicability
Potential for interlaboratory transfer
Sensitivity to chemical stress
Generalizability to other aquatic ecosystems
Unequivocality of results
Statistical basis for interpreting results
Existence of standards for rejecting results
(test failure)
Frequency of test failure
Level of training and expertise required
Time required
Cost per chemical
Others

Working Session III: Protocol Development

1. Write a first-approximation protocol for testing
chemicals for effects on ecosystem properties. Of
course, many questions would remain to be resolved
before these protocols could be adopted by the
Environmental Protection Agency (EPA)—these questions
should be identified as they arise in the protocols.
An outline for a protocol could be as follows:

I. Hypotheses to be tested by this protocol

II. System design

Size
Abiotic and biotic components
Light, temperature controls
Strategies for maximizing replicability

III. Test Procedure

Controls
Controls
Replicates per treatment; treatments per chemical
-------
83
Introduction of test chemical into system
Measurement of effects
IV. Analysis of Results
Statistical basis for identifying effects
Criteria for rejection of test results
Interpretation of results
Generalization to other aquatic ecosystems
V. Development Needs

How to validate the test
Major questions that need to be resolved
before finalizing protocol

6.2 Results and Discussion

6.2.1 Test Descriptions and Measurable Ecosystem-Level Properties

(1) Lentic Working Group. Seventeen model ecosystems or model
ecosystem types within the scope of this workshop were listed (Table
6.1). Various approaches to organizing the list into general classes
were suggested. Classification schemes based on distinctions between
open and closed systems, between static and flow-through systems,
between synthesized (gnotobiotic) and naturally-derived systems,
between systems with and without sediments, and between large and
small systems were all rejected because the variety of model
ecosystems makes such categorizations unworkable. The list of 17 was
later condensed to 9 representative systems [Section 6.2.2(1)].

A detailed list of measurable ecosystem properties was then
compiled (Table 6.2). Because some of these properties were
operationally defined, the list did not clearly distinguish between
system-level parameters per se and methods that are commonly used for
measurements. For the purposes of constructing an evaluation matrix,
each property or method on the list was considered individually.

The model ecosystems in Table 6.1 were then evaluated in terms of
their suitability for measuring the properties in Table 6.2. A simple
3-level rating system was used, with a rating of 1 indicating that a
parameter was easy to measure in a given system; a rating of 3,
difficult; and a rating of 2, intermediate. The full matrix is
presented in Table 6.3.

In general, Group A thought that nearly all the ecosystem-level
parameters could be measured in any of the model ecosystems, but that
some measurements were more difficult than others. Destructive
sampling over time is difficult in smaller systems; therefore, the
number of measurements that can be made on a single experimental unit
is limited. Sampling of larger, more complex systems is complicated
-------
84
TABLE 6.1. MODEL ECOSYSTEMS
(Lentic Working Group)
1. Gnotobiotic mixed flask culture (Nixon 1969; Taub 1969)
2 Derived mixed flask culture, closed (Corden et al. 1969)
3. Derived mixed flask culture, open (Leffler 1977; Neil! 1975;
Thomas 1978)
4. Carboy microcosm (McConnell 1962)
5. Pelagic community (Harte et al. 1978)
6. Pond community (Brockway et al. 1979; Giddings and Eddlemon
1979)
7. Artificial food chain, aquatic (Isensee et al. 1973)
8. Artificial food chain, terrestrial/aquatic (Metcalf et al.
1971)
9. Salt marsh box core (Kitchens 1979)
10. Marine pelagic, 150 L (Perez et al. 1977)
11. Marine littoral, benthic (Henderson et al. 1976)
12. Estuary, compartmentalized (Cooper and Copeland 1973)
13. Marine sediment core (Pritchard et al. 1979)
14. Freshwater sediment core (Medine et al. 1980)
15. Sediment-water systems, general (Neame and Goldman 1980)
16. Narragansett Bay, 13 m3 (Pilson et al. 1980)
17. Marine plankton, deep tank (Strickland et al. 1969)
-------
85
TABLE 6.2. MEASURABLE ECOSYSTEM PROPERTIES
(Lentic Working Group)
A. Chemical matrix (Eh, pH, etc.) L.
B. Electrical potential M.
C. Nutrient levels (mass balance)
N.
D. Nutrient flux
E. Dissolved organic carbon 0.
F. Biomass P.
G. Chlorophyll Q.
H. Fluorescence R.
I. Turbidity S.
J. ATP T.
K. Size spectrum of particles
Taxonomic description
Primary production,
respiration, P/R
Oxygen dynamics (input-
output)
Heterotrophic activity
Dehydrogenase activity
Enzyme systems
Grazing rate
Colonization rate
Spatial/temporal patterns
-------
86

Q
iy^
LU
P
oc
LU
a.
of
S
h™
W>
S
LU

5
QC
3
^g
z
oe
o
LU
X
LU
r-

O
U
LU
_j
LU
o
S
Z
U_
0
o
*—
i
3
5
LU

a
I

^^
rH
3
.
LA
r-t
.
S
.
I-H

.
CM
rH
.
rH
rH

.
If
in .
s?0"
a
u
UJ .
00
r—
01
•o
o .
*"*'

t^O
.
tf)

.
^f

1-4 f-Hi-H .-ir-iro.-icsjevjr-ioom^-en

i-lrH (*> iH^-ti-li-1r-*f-ICMrHCMe>J^'rOfO^-^

i-^f— (r— 4COr— (r-*rHrHi-HrHrHC\li-11t>
u>
^-
5
c
•r-
01
J3
to
£
w

3 O
in c
a
i- e O

u» >> o o>
(O *J *» r-
O r- U
*>'|^ Or^
u?'5'«E §•
..« -t-
U) HI HI T3 4->
O1 i * o
C >. TO >> C
*> at ~o at >>
a > ai > *?
QE -r- S f- S.

s- to a) nj a.
O r— *> •— O
Ol C 0) 1.
c ae i— i ae o.
o
+J u II u u
C rH 04 « *
"a.
LU

^
*fl»
>
*>
S
&
in
in
01
i.
at
2
a.
"S
a
in

^
U>
>>
(A
01
0)
•a
§
M-
O
C
O
-t—

ia
u
t-

c
di
•S

t-
o
t-
CM
VO
^J
c
ia
rH
IA

in
0)

a
i—

at
ai
CO
10

-------
87
by the difficulty of spatial heterogeneity within each system, which
necessitates a stratified sampling regime.

Chemical and physical parameters are relatively easy to measure
in the model ecosystems considered. Autotrophic activity is also
easily determined (except, of course, in systems lacking autotrophic
components). Heterotrophic activity is more difficult to measure
because of the methods now available. Nutrient levels are easy to
measure, but large numbers of samples may be required to overcome
spatial and temporal variability. (Measurement of nutrient flux, as
opposed to instantaneous concentrations, is more difficult;
radioisotopes are generally needed). Finally, taxonomic descriptions
are difficult and time consuming in all but the simplest model
ecosystems.

(2) Lotic Working Group. Three major classes of lotic model
ecosystems were recognized: (a) closed (completely recirculating)
systems, (b) partially recirculating systems, and (c) open (once-
through flow) systems. Each class was further broken down as shown in
Table 6.4. These system types generally fall along a gradient from
small, completely recirculating laboratory devices to large-scale,
outdoor streams. Examples of each type are given in Table 6.4.

Measurable system-level parameters were identified and classified
as structural or functional (Table 6.5). "Transport" in this list
refers to all downstream (or upstream) movement of organisms and of
dissolved and particulate matter, rather than toxicant transport in
the usual sense; "fate" refers to the chemical fate of the toxicant.
Except for transport, fate, and behavior, the properties listed by the
Lotic Working Group correspond closely with those listed by the Lentic
Working Group.

The Lotic Working Group used a three-level rating system, similar
to that of the Lentic Working Group to assess the feasibility of
measuring particular properties in each model ecosystem category. By
and large, all properties are measurable in all systems. The major
exception to this generalization is the closed recirculating tube
system, which is too small and ecologically incomplete for some
measurements. The matrix of model ecosystems vs. ecosystem properties
is presented in Table 6.6.

6.2.2 Evaluation of Test Systems

(1) Lentic Working Group. Of the original list of 17 model
ecosystems, nine were evaluated as TSCA testing protocols. This
condensation was achieved by grouping similar systems into one
category, and by including only those experimental systems that were
considered to have the most potential. The condensed list forms the
vertical axis of Table 6.7.
-------
88

TABLE 6.4. MODEL ECOSYSTEMS
(Lotic Working Group)
Closed systems
Closed recirculating tubs (e.g., Gushing and Rose 1970)
In-Situ (e.g., Bott et al. 1978)
Recirculating trough (e.g., Kevern and Ball 1965)
Laboratory channel (e.g., Kehde and Wilhm 1972)

Partially recirculating systems
Boxes, tubes, etc. (e.g., Mclntire et al. 1964)
Laboratory channel (e.g., Brocksen et al. 1968)

Open systems
Wood or concrete channels (e.g., Maki and Johnson 1976)
Stream-side flume (e.g., Armitage 1980)
In-stream flume (e.g., Manuel and Marshall 1980)
Large-scale outdoor stream (e.g., Kania et al. 1976)
-------
89
TABLE 6.5. MEASURABLE ECOSYSTEM PROPERTIES
(Lotic Working Group)
Functional
Primary production Biochemical measurements
Community production Functional group diversity
Community respiration Transport
Secondary production Behavior
Nutrient dynamics Fate

Structural
Standing stock
Diversity - equitability
Colonization (recolonization)
Physical properties (pH, DO, DOC, etc.)
-------
90
TABLE 6.6. EVALUATION OF MODEL ECOSYSTEMS
FOR MEASURING ECOSYSTEM PROPERTIES
(Lotic Working Group)
Ecosystem
properties

£
u
01
Closed
systems
c
o
•^
•w
IO
Partially
recirculating
(A
01
.Q
3

concrete
Open
systems
3

01
Functional:

Conun. Respiration
2° Production

Biochemical
Functional Group Diversity
Transport*
Behavioral
Fate
u
01
s.
•o
4) M
IA 0)
5-8
u *»

N
N
Y
3
10
e

Y
N
Y
•w
IO
"3 <—
O "*TJ v
(- DI e
•^3 C
«2 -85
ce +J — J u

Y Y

Y Y
N M
Y Y
^
3
"»
10 C
01 • C
X U J3 10
o *> m .c
oa o> — i o

Y Y

Y Y
M M
Y Y
u -o
ic'aJ 6 u
c a o> *> 01
•o c 01 s MB
O JZ *> r— C i—
3 u «/> *• 1-1 1-

Y Y Y

Y Y Y
Y Y Y
Y Y Y
SL
§o
S
10 3
—1 O

Y
Y
Y
Structural:
Standing stock
Di versi ty-equi tabi 1 1 ty
Colom'z. , Recol.
Physical prop. (pH, 00,
DOC, etc.)
"Dissension expressed
Y
Y
N
Y

Y
Y
N
Y

Y
Y
M
Y

Y
Y
Y
Y

Explanation of Ratings:
Y = YES; N = NO; M = MAYBE
-------
91

TABLE 6.7. EVALUATION OF TEST SYSTEMS

(Lentic Working Group)

a
Model ecosystem type
(1)
(2), (3)
(5)
(6)
(7)
(9)
(10), (11)
(13)
(14)

JK
+>
•r"
r—
•i—
.a

r—
a.
a>
G
G
G
G
F
F
G
G
G
.a
•r*
•^
-Q

N
•i-.
^f
t~
tj
TJ
C

4J
G
G
G
G
G
G
G
G
G
•o
•a
0)
•r-
3
cr
a)
£_

O)
C
•1—
c
•r—
<0
i.
1—
PC6
1
1-7
1-7
PC
1-14
1-14
1-2
1-21
(A
0)
0)
U (A

C
o
(A "O
•P -P
U
0)

2 +
2
2
(A
O
U

c
o
•r-
•P
TO
t-
Ol
Q.
O
^
-
+
+ 1 >3
1
+ 4
4
h
2
1 +
2 +
2
1
1 +
-
+
+
-
—

^.
) *
•r-
>
urm
4->

(A
C
i/)
7
7
7
7
7
7
7
7
7

aNumbers refer to Table

bG = Good
6.1

d +
-

= relatively
= relatively

expensive

inexpensive

F = Fair
+ = equilibration necessary
- = equilibration not necessary
PC = experience with pure cultures
needed
-------
92

Because time for discussion was limited, seven criteria were
selected for evaluating these systems. These criteria are listed on
the horizontal axis of Table 6.7. The following conclusions were
reached:

1. Replicability. Most of the systems were judged to be fairly
replicable, although replicability is essentially a function
of the parameter being measured. Gross parameters, such as
total primary productivity, are more similar between
replicate microcosms than population-level measurements. It
was also observed that variations in some properties over
time may be out of phase between replicates; replication can
still be considered good if replicate systems behave the
same even though fluctuations are out of phase.

2. Interlaboratory transfer. All of the systems considered
could be built and operated at different laboratories
without difficulty.

3. Training required. All of the systems are fairly easy to
set up and to operate. Estimates of the time needed to
train an inexperienced technician to use these systems
ranged from 1 day to 3 weeks. The synthetic systems require
expertise in maintaining pure stock cultures. The
measurement of ecosystem properties would be the most
difficult aspect of these tests. The amount of training
required would therefore depend on which parameters were
being measured.

4. Time required. The time required to conduct tests with
model ecosystems is a function of experimental objectives
and measured responses. Some of the systems require an
equilibration time after set up to allow conditions to
stabilize before beginning an experiment. Acute effects
might be measurable in as little as 6 h; chronic effects
could be observed for 3 months or longer.

5. Cost of testing. None of the systems are particularly
expensive to set up or operate. Taub's gnotobiotic systems
are probably more expensive than most of the
naturally-derived systems considered, because the
gnotobiotic systems require maintenance of stock cultures.
The salt marsh models of Kitchens and the 3-phase sediment
cores of Medine and Porcella include some simple plumbing,
which would increase their costs. For all of the tests, the
major expenses are chemical analyses and other costs of
measuring ecosystem responses. As with any chemical testing
system, some expenses may be incurred in protecting workers
from exposure to potentially harmful materials.
-------
93
6. Sensitivity. The sensitivity of these systems was
considered to be dependent on the response parameters
measured rather than on the test systems themselves. It is
not yet known which responses are most sensitive to chemical
stress; this is an area where more basic research is needed.

7. Generalizability. Extrapolating from laboratory tests to
natural systems was recognized as the major problem in model
ecosystem research. To some extent, the generalizability of
model ecosystem results depends on the responses being
measured. Two basic strategies for making ecosystem-level
tests "representative" emerged from the discussion. One
approach is to construct a generic ecosystem (i.e., a system
exhibiting important ecosystem properties but not mimicking
any particular natural ecosystem). Such a system could be
used to rank chemicals in order of toxicity, in the early
stages of hazard assessment. A second approach is to model
a selected natural ecosystem as closely as possible, so that
microcosm results could be taken as indicative of probable
effects in the natural ecosystems; the problem then becomes
one of generalizing from one natural ecosystem to other
natural ecosystems. A predictive, or mimicking, model
ecosystem would probably be most useful in later stages of
hazard assessment.

(2) Lotic Working Group. The 10 general categories of model
ecosystems discussed in Working Session I were evaluated according to
13 criteria important to a hazard assessment process. The results of
this evaluation are summarized in Table 6.8. The criteria were
defined as follows:

1. Replicability. The Lotic Working Group interpreted
"replicability11 to mean ease of setting up replicate
systems. Smaller systems are more replicable, by this
definition, than larger systems. The question of similarity
between replicates was not discussed.

2. Interlaboratory transfer. The methodology of smaller model
systems is more easily transferred between laboratories than
that for larger, more complex systems.

3. Sensitivity. Small systems are probably more sensitive to
chemical stress than larger systems. However, sensitivity
is primarily a function of the responses measured.

4. Generalizabi1ity. The ability to relate model ecosystem
results to natural ecosystems was judged to be the most
important criterion for evaluating model ecosystems as
chemical testing tools. Small, closed systems were rated
poor in this regard; only larger, open systems are enough
like natural streams to permit reliable predictions. Even
-------
94
TABLE 6.8. EVALUATION OF TEST SYSTEMS
(Lotic Working Group)
Criteria

Danl 4 faH-i 1 < *•»/

potential ror inter iao.
transfer
Sensitivity to chemical
stress
uenera i i zaoi i i ty to
other aquatic ecosystems
Linearity
Unequivocal ity of results
statistical oasis tor
interpreting results
Existence of standards
for rejecting results
(test failure)
Frequency of test
failure
Level of training and
expertise required
Time required
Cost per chemical
Cause-effect
Interpretation
H = High G = Good
N = Medium F = Fair
L = Low P = Poor
Closed Partially
systems reclrculating Open systems
Closed reclrculating
tubes

U
G

L
L
H

a
M
fc

U
G

L
L
H

((•circulating trough
Lab channel
Lab channel
G/* /»

b u
G G G

U U U
G G G

U U U

L M M

L M M
L M M
H M M

b
x s i i
• « S i ^
- u *- 3 o
we i—
tl O HI H- 01
* ° 5 . -5
*» -o M 3 u
§1 a 2 !*
M £ 3 *> 01
01 -a c o> i/> a
x o C «0
S u «/> N* — '

F F F F F

U U U U U
G G G G G

U U U U U

M H H H H

M H H H H
M H H H H
N L L L L

U - Unknown
-------
95
with larger model streams, doubts about ecological realism
remain.

5. Linearity. This criterion refers to the effects of scale on
the properties of model streams. To what extent can model
ecosystem results be "scaled up" to full-sized natural
streams? This question cannot be answered for any model
ecosystem at present.

6. Unequivocalityof Results. The group interpreted this
criterion as "faith in the results". The conclusion was
that any test system could be designed to provide
unambiguous responses. However, the ability to interpret
experimental results will depend on an improvement in our
current knowledge of the complex interactions occurring in
ecosystems.

7. Statistical Basis for Interpretation. Statistical analysis
of results is easiest with small systems. The inherent
variability of larger systems means that more samples are
needed to achieve a given level of confidence in the
measurements, and that temporal trends are more difficult to
detect. The difficulty of replicating larger systems also
detracts from the ability to detect statistically
significant effects.

8. Standards for rejection. Simple standards for rejecting the
results of any particular test might be possible for smaller
model systems but less likely for larger ones.

9. Frequency of test failure. The frequency of test failure is
unknown for any model stream, mainly because failures are
rarely reported in the literature.

10. Training required, time required, and cost. Small, recir-
culating systems require less training, less time, and less
money to operate than larger systems. The large, open
systems received the worst ratings by these criteria.

11. Cause-effect interpretation. The complexity of a large,
open system makes interpretation of chemical effects
difficult. Direct toxic effects are not easily untangled
from the web of secondary effects and interactions occurring
in a complex system. Cause and effect are most easily
distinguished in small, simple systems.

An interesting generality that emerged from this evaluation
exercise was that systems meeting the operational criteria for a
screening test (replicability, potential for interlaboratory transfer,
statistical interpretation, low cost, short time, and low level of
expertise required) are the same systems that are least generalizable
-------
96
to natural ecosystems. This echoes the conclusion of the Lentic
Working Group that two types of ecosystem test systems are
desirable—one for screening, the other for prediction. Rapid, simple
screening tests cannot be designed without sacrificing ecological
realism. Model ecosystems designed with ecological realism as their
objective are most appropriate at later stages in the assessment
process.

6.2.3 Protocol Development

The protocol development exercise served two functions. First,
outlines of three possible ecosystem-level effects tests based on
three widely-used model ecosystem types were produced. Second, the
protocols constituted focal points for discussions of common problems
encountered in system-level testing, and strategies for dealing with
them.

There were interesting differences in the approaches taken by the
working groups. The Lentic Working Group chose to develop two
protocols, one involving sediment communities (freshwater or marine),
and one including primarily pelagic organisms. The sediment protocol,
adapted from methods of Medine (freshwater) and Pritchard (marine),
provides for small-scale, simplified simulation of specific aquatic
environments. The pelagic protocol is essentially the mixed flask
culture method of Beyers, Leffler, and others; the experimental unit
is a highly-simplified, naturally-derived community that simulates no
specific natural ecosystem, but exhibits system-level properties
common to all ecosystems. Both of these protocols were developed by
selecting familiar model ecosystem designs that seemed most amenable
to routine toxicity screening. The Lotic Working Group approached the
task differently. They began by selecting a small number of
system-level responses that were considered most ecologically
meaningful, and then designed a model ecosystem that would reflect
these responses as realistically as possible. The protocol that
resulted from these deliberations is a nonrecirculating laboratory
channel that, according to the conclusions reached in Working Session
II, does not meet the requirements for a screening test but could be
used in the advanced stages of a chemical hazard assessment. The
Lotic Working Group thought that model streams simple enough to be
used for screening would not represent natural streams realistically.

Each protocol obviously contains many unresolved problems, and
none of the protocols could be used in hazard assessment without
extensive refinement and validation. They are sketches of feasible
ecosystem-level tests as envisioned by a group of experienced
scientists. The protocols are presented here as prototypes for
further research and development.

(1) Sediment core effects test procedure. This procedure tests
the hypothesis that a chemical will alter the system-level properties
(chemical matrix, primary production, heterotrophic activity, nutrient
-------
97
cycling) of a sediment-water system. The test consists of cylinders
containing homogenized sediment or intact cores, water, and natural
biota; continuous or semi continuous flow. Hypolimnetic, littoral, or
marine benthic environments are simulated.

Cores are contained in 1- to 10-L cylinders, with depth 5 to 6
times the diameter. Cylinders may be lucite or glass; lucite is
recommended for testing metals, but is unusable with some organic
materials; glass is more expensive, but deteriorates less rapidly than
plastic. A sediment core or homogenized sediment (depth equal to
cylinder diameter) is placed in each cylinder. Cylinders are then
filled with natural water (prefiltering not recommended) or defined
medium (preferable for mass balance calculations). Biota are included
with the sediment and/or water. Medium exchange is continuous or
semicontinuous, with a residence time of 2 to 10 d. Cylinders are
illuminated from above by horizontal Duro-Test lights (approximating
sunlight spectrum) on a 12-h photoperiod (hypolimnion can be simulated
by running test in darkness). Temperature is maintained at 15 to
25°C, depending on environment; temperature is controlled within 0.5°C
by water jackets or by an environmental chamber. The water is stirred
with sufficient mixing energy to prevent stratification without
resuspending sediment. Aeration is not required.

A test should include two replicates per treatment, plus two
controls (plus two carrier controls if appropriate). Each chemical is
tested at three concentrations at least. The chemicals are introduced
with the medium, or added separately by syringe pump; introduction of
chemicals without a carrier is preferred, if possible. Homogenized
cores should be equilibrated for 15 to 20 d before introducing the
chemical; no equilibration is necessary with intact cores. Measure-
ments to be made on treated and control systems include:

1. Chemical matrix (Eh, pH, TOC, DOC, contaminant levels).

2. Primary production and respiration (by light/dark bottle
technique, measurement of diurnal dissolved oxygen
fluctuations, or oxygen mass balance).

3. Nutrient levels, especially N03-N, NH4-N, organic N.

4. Heterotrophic activity (14C02 production from labelled
compounds, algal lysate, or detritus; or carbon balance).

Measurements should be made at least weekly. Two weeks of
exposure are required to determine no-effect levels; three sampling
intervals are necessary to confirm the magnitude of effects observed.

Treated microcosms are compared with controls to identify effects
of chemical treatment. Test results are rejected if controls or
treatment are lost; controls should exhibit ecosystem measurements
consistent with local environmental conditions. Interpretation of
-------
98
results is arbitrary until comparisons with other microcosm
experiments or field studies can be made (high level of funding
recommended for this purpose).

(2) Pelagic ecosystem effects test procedure. This procedure
tests the hypothesis that a chemical will alter the system-level
properties (chemical matrix, primary production, heterotrophic
activity, nutrient cycling) of a generalized aquatic ecosystem. The
test consists of bacteria, algae, and microinvertebrates maintained in
artificial growth medium. Typical ecosystem properties are exhibited
but no particular natural ecosystem is simulated.

Mixed aquatic communities are maintained in loosely-capped 4-L
beakers. Samples of biota from local ecosystems are used to establish
laboratory stock cultures; the taxonomic composition of these cultures
is not important as long as major functional groups (autotrophs,
grazers, detritivores, and decomposers) are present. When stock
cultures reach a fairly stable taxonomic composition, they are used to
inoculate the test communities. A standard growth medium (e.g.,
Taub's #63 or Gerloff's) is used. Cultures are stirred continuously
or only during sampling. Light is supplied for 12 h each day. A
constant temperature of 20 to 25°C is maintained. During the
pretreatment period, replicates are cross-seeded periodically to
increase uniformity of composition. Evaporative losses are replaced
with equal volumes of stock culture.

Each test includes five controls plus five replicates per
treatment. Chemicals are tested at three treatment levels
corresponding to 1/10 X, IX, and 10 X the predicted environmental
concentrations. After a 6-week equilibrium period, microcosms are
treated once with the test chemicals; treatments are not repeated.
Measurements on treated and control systems include:

1. Chemical matrix (Eh, pH)

2. Primary production and respiration (3-point oxygen
method)

3. Nutrient levels (N03-N, NH4-N, ortho-P, SRP, total C,
DOC)

4. Heterotrophic activity (14C02 release from labelled
substrate, epifluorescent bacterial counts, or oxygen
changes)

5. Autotrophic biomass (chlorophyll, phaeophytin,
fluorescence)

6. Turbidity
-------
99
Measurements that can be automated should be made twice daily
(just before the lights turn on and just before they turn off); other
parameters should be measured weekly. The systems should show a
response within 2 weeks after the chemical is added. System response
should be monitored for a period sufficient to allow recovery, or a
maximum of 10 weeks.

Deviations of measured parameters from the 95% confidence limits
of the normal operating range of controls are integrated over time to
provide an index of effect. This index is "total relative stability"
(Leffler 1977); it provides a single measure of an ecosystem's total
stress response. This index can be compared with those produced by
standard chemicals tested in the same systems. Any deviation from
controls signifies a positive response, because 95% confidence limits
are built into the analysis. Test results are rejected if the test
system fails to rank four standard chemicals properly, or if the
behavior of the controls is aberrant. The ranking of four standard
chemicals serves as the criterion for assessing the reproducibility of
the test system. Thus, even though the species composition and
proportions may vary within each batch of microcosms, the batches are
considered as the same test system as long as they rank the four
chemicals in standard order.

All aspects of the protocol need much more testing and
development. Methods of introducing chemicals must be refined. The
optimal time frame of the test has not been established.

(3) Model stream effects test procedure. This procedure tests
the hypothesis that a chemical will alter the system-level properties
(primary production, respiration, community production) of a stream
ecosystem. The test consists of laboratory-scale, nonrecirculating
model streams, containing natural substrate and naturally derived
biota. Regional characteristics of small stream ecosystems are
simulated.

Model streams are assembled in indoor troughs, 2- to 10-m long
and 30-cm wide. A substrate (limestone, gravel, rock) is placed in
the troughs, and well water is pumped through the systems with a
current speed of 0.01 to 0.04 m/sec. A naturally derived autotrophic
community (system-dependent) and a nondrifting invertebrate grazer
(cosmopolitan) are the major biota. Overhead lighting is provided at
about half the normal sunlight intensity for the region. Temperature,
pH, and DO are controlled within the normal regional range.

A single test includes two replicates at each of three treatment
levels, plus two replicate controls. The effects of the chemical are
measured after 2 to 4 weeks of continuous exposure. Primary
production and respiration are measured on stream substrate
communities placed in respiration chambers. Net community production
is measured by destructive sampling and total biomass determination
after 30 d exposure. Nonparametric procedures are used to compare
treated streams with controls.
-------
100
The protocol must be checked in inter!aboratory round-robin
tests. The protocol may be validated by comparing test results with
conventional toxicity tests and any available field data.

6.3 Conclusions

Most ecosystem properties (primary productivity, secondary
productivity, ecosystem respiration, nutrient cycling, total biomass,
functional or structural diversity, etc.) can be measured in most
model ecosystems. The ease with which a particular property may be
measured, and the degree to which that measurement is representative
of natural ecosystems, depends on the laboratory system and the
property being measured. Research is needed to identify which
properties are most easily measured, most sensitive to chemical
stress, most critical to the functioning of the ecosystem, and most
generalizable to other systems. This research must precede the
adoption of standard testing protocols.

The model ecosystems that are the most replicable, inexpensive,
simple, rapid, and easily standardizable are also the least realistic
and most difficult to extrapolate to natural ecosystems. There seem
to be two potential roles for model ecosystems in hazard assessment.
The first role is a general, nonrepresentational model (i.e., mixed
flask culture). This type of system exhibits universal ecosystem
properties without mimicking any particular natural ecosystem. Such a
system is easily replicated, simple, and cost efficient, and could be
used early in the testing sequence to screen chemicals for their
ability to disrupt ecosystem processes. The second role involves more
detailed representation of specific natural ecosystems such as ponds,
lakes, streams, or coastal environments. These systems can provide
information on the magnitude and direction of ecosystem effects, as
well as details about sensitive organisms, sensitive processes,
indirect effects, and ecosystem recovery. Because of the greater
expense and expertise required to use such model ecosystems and
because their results are not necessarily generalizable to other
ecosystem types, these systems are best used in the later stages of
hazard assessment.
-------
101

6.4 References

Armitage, B. J. 1980. Effects of temperature on periphyton biomass
and community composition in the Browns Ferry experimental
channels. In: Microcosms in Ecological Research, J. P. Giesy,
ed. (in press).

Brocksen, R. W., G. E. Davis, and C. E. Warren. 1968. Competition,
food consumption, and produciton of sculpins and trout in
laboratory stream communities. J. Wildl. Manag. 32:51-75.

Brockway, D. L., J. Hill, IV, J. R. Maudsley, and R. R. Lassiter.
1979. Development, replicability, and modeling of naturally
derived microcosms. Int. J. Environ. Stud. 13:149-158.

Bott, T. L., J. T. Brock, C. E. Gushing, S. V. Gregory, P. King, and
R. C. Petersen. 1978. A comparison of methods for measuring
primary productivity and community respiration in streams.
Hydrobiologia 60:3-12.

Cooper, D. C., and B. J. Copeland. 1973. Responses of
continuous-series estuarine microecosystems to point-source input
variations. Ecol. Monogr. 43:213-236.

Cushing, C. E., and F. L. Rose. 1970. Cycling of zinc 65 by Columbia
River periphyton in a closed lotic microcosm. Limnol. Oceanogr.
15:762-767.

Giddings, J. M. , and G. K. Eddlemon. 1979. Some ecological and
experimental properties of complex aquatic microcosms. Int. J.
Environ. Stud. 13:119-123.

Gorden, R. W., R. J. Beyers, E. P. Odum, and R. G. Eagon. 1969.
Studies of a simple laboratory microecosystem: Bacterial
activities in a heterotrophic succession. Ecology 50:86-100.

Harte, J., D. Levy, E. Lapan, A. Jassby, M. Dudzik, and J. Rees.
1978. Aquatic microcosms for assessment of effluent effects.
Electrical Power Research Institute EA-936.

Henderson, R. S. , S. V. Smith, and E. C. Evans. 1976. Flow-through
microcosms for simulation of marine ecosystems: Development and
intercomparison of open coast and bay facilities. Naval Undersea
Center, San Diego, California, NUC-TP-519.

Isensee, A. R., P. C. Kearney, E. A. Woolson, G. E. Jones, and V. P.
Williams. 1973. Distribution of alkyl arsenicals in model
ecosystems. Environ. Sci. Technol. 7:841-845.

Kania, H. J., R. L. Knight, and R. J. Beyers. 1976. Fate and
biological effects of mercury introduced into artificial streams.
Environmental Protection Agency, EPA-600/3-76-060.
-------
102
Kehde, P. M., and J. L. Wilhm. 1972. The effects of grazing by
snails on community structure of periphyton in laboratory
streams. Am. Midi. Nat. 87:8-24.

Kevern, N. R., and R. C. Ball. 1965. Primary productivity and energy
relationships in artificial streams. Limnol. Oceanogr.
10:74-87.

Kitchens, W. M. 1979. Development of a salt marsh microecosystem.
Int. J. Environ. Stud. 13:109-118.

Leffler, J. W. 1977. A Microcosm Approach to an Evaluation of the
Diversity-Stability Hypothesis. Ph. D. dissertation, U. Georgia.

Maki, A. W., and H. E. Johnson. 1976. Evaluation of a toxicant on the
metabolism of model stream communities. J. Fish. Res. Board Can.
33:2740-2746.

Manuel, C. Y. , and G. W. Marshall. 1980. Limitations on the use of
microcosms for predicting algal response to nutrient enrichment
in lotic systems. In: Microcosms in Ecological Research, J. P.
Giesy, ed. (in press).

McConnell, W. J. 1962. Productivity relations in carboy microcosms.
Limnol. Oceanogr. 7:335-343.

Mclntire, C. D., R. L. Garrison, H. K. Phinney, and C. E. Warren.
1964. Primary production in laboratory streams. Limnol.
Oceanogr. 9:92-102.

Medine, A. J., D. B. Porcella, and V. D. Adams. 1980. Heavy metal
and nutrient effects on sediment oxygen demand in three-phase
aquatic microcosms. In: Microcosms in Ecological Research, J.
P. Giesy, ed. (in press).
alf, R. L. , G. H. Sangha, and I. P. Kapoor. 1971. Model
ecosystem for the evaluation of pesticide biodegradability <
ecological magnification. Environ. Sci. Technol. 5:709-71;
Neame, P. A., and C. R. Goldman. 1980. Oxygen uptake and production
in sediment-water microcosms. In: Microcosms in Ecological
Research, J. P. Giesy, ed. (in press).

Neill, W. E. 1975. Experimental studies of microcrustacean
competition, community composition, and efficiency of resource
utilization. Ecology 56:809-826.

Nixon, S. W. 1969. A synthetic microcosm. Limnol. Oceanogr.
14:142-145.
-------
103
Perez, K. T., G. M. Morrison, N. F. Lackie, C. A. Oviatt, and S. W.
Nixon. 1977. The importance of physical and biotic scaling to
the experimental simulation of a coastal marine ecosystem.
Helgol. Wiss. Meeresunters. 30:144-162.

Pilson, M. E. Q., C. A. Oviatt, and S. W. Nixon. 1980. Annual
nutrient cycles in a marine microcosm. In: Microcosms in
Ecological Research, J. P. Giesy, ed. (in press).

Pritchard, P. H. , A. W. Bourquin, H. L. Frederickson, and T. Maziarz.
1979. System design factors affecting environmental fate studies
in microcosms, pp. 251-272. In: Microbial Degradation of
Pollutants in Marine Environments, A. W. Bourquin and P. H.
Pritchard, eds. EPA-600/9-79-012. pp. 251-272.

Strickland, J. D. H., 0. Holm-Hansen, R. W. Eppley, and R. J. Linn.
1969. The use of a deep tank in plankton ecology. I. Studies of
the growth and composition of phytoplankton crops at low nutrient
levels. Limnol. Oceanogr. 14:23-34.

Taub, F. B. 1969. A biological model of a freshwater community: a
gnotobiotic ecosystem. Limnol. Oceanogr. 14:136-142.

Thomas, C. L. 1978. A Microcosm Study of the Interactions of
Nutrients and Cadmium in Aquatic Systems. M. S. Thesis, U.
Georgia.
-------
METHODS FOR MEASURING EFFECTS OF CHEMICALS
ON TERRESTRIAL POPULATION INTERACTIONS
February 26 and 27, 1980
Glenn W. Suter, II
Chairman

Environmental Sciences Division
Oak Ridge National Laboratory
-------
106
PARTICIPANTS

G. W. Suter, II, Chairman
Oak Ridge National Laboratory
WORKING GROUP A: EFFECTS OF CHEMICALS
ON MICROBIAL INTERACTIONS IN TERRESTRIAL SYSTEMS
R. M. Atlas (Group Leader) R. F. Strayer
University of Louisville Oak Ridge National Laboratory

Ralph Baker
Colorado State University
WORKING GROUP B: EFFECTS OF CHEMICALS ON
INTERACTIONS OF TERRESTRIAL PLANTS AND OF PLANTS AND MICROBES
J. L. Ruehle (Group Leader) A. S. Heagle
Forestry Sciences Laboratory North Carolina State University

Udo Blum D. S. Shriner
North Carolina State University Oak Ridge National Laboratory
WORKING GROUP C: EFFECTS OF CHEMICALS ON
TERRESTRIAL ARTHROPOD INTERACTIONS
John All (Group Leader) D. C. Force
University of Georgia California State Polytechnic University

W. J. Webb (Recorder) M. S. McClure
Oak Ridge National Connecticut Agricultural Experiment
Laboratory Station

David Craig D. M. Nafus
University of Illinois U.S. Department of Agriculture
OBSERVERS
J. V. Nabholz A. S. Hammons
U.S. Environmental Protection Oak Ridge National Laboratory
Agency
-------
107
SECTION 7

METHODS FOR MEASURING EFFECTS OF CHEMICALS
ON TERRESTRIAL POPULATION INTERACTIONS
7.1 Introduction

This workshop considered the availability and utility of
multispecies laboratory systems to display the responses of
terrestrial population interactions to toxic chemicals. The workshop
was divided into three working groups. Working Group A discussed
interactions between microbe populations and within microbial
communities. Working Group B discussed interactions between plant
populations and between plants and microbes. Working Group C
discussed interactions between arthropod populations and between
arthropods and plants. Each working group participated in three
sessions. The objectives of each session were outlined as follows:

Working Session 1: Identification and Description of Test Systems

Identify laboratory systems that display terrestrial population
interactions or community properties. If possible, organize these
systems into categories that are sufficiently well defined to permit
generalization about their potential utility in the Environmental
Protection Agency's (EPA) Toxic Substances Control Act (TSCA) hazard
assessment processes.

List population and community properties measurable in each
system.

Working Session II: Evaluation of Test Systems

Construct a matrix with test systems on the horizontal axis and
experimental criteria on the vertical axis. For each intersection in
the matrix, evaluate that test system in terms of that criterion. The
criteria are:

Replicability

Potential for interlaboratory transfer

Sensitivity to chemical stress

Generalizability to other terrestrial ecosystems

Statistical basis for interpreting results

Existence of standards for rejecting results (test failure)
-------
108

Frequency of test failure

Level of training and expertise required

Time required

Cost per chemical

Others

Working Session III: Protocol Development

Based on our experience to date, it may be possible to write a
first-approximation protocol for testing chemicals for effects on
certain population interactions. Of course, many questions would
remain to be resolved before these protocols could be adopted by the
EPA—these questions should be identified as they arise in the
protocols. An outline for a protocol could be as follows:

I. Hypotheses to be tested by this protocol

II. System design

Size
Abiotic and biotic components
Light, temperature controls
Strategies for maximizing replicability

III. Test Procedure

Controls
Replicas per treatment; treatments per chemical
Introduction of test chemical into the system
Measurement of effects

IV. Analysis of Results

Statistical basis for identifying effects
Criteria for rejection of test results
Interpretation of results
Generalization of other terrestrial ecosystems

V. Developmental Needs

How to validate tests
Major questions that need to be resolved before
finalizing protocol

Many other systems may seem appropriate for development.
Identify the systems that seem to be most appropriate and describe the
types of development that are necessary for each (e.g., testing of
species, development of media).
-------
109
What types of modeling or analysis would be necessary to relate
test results to the real world, to generalize the results to a variety
of environments, or to relate the parameters measured to socially
significant parameters?

What simpler tests could be confirmed by the results of this
multispecies test? What more complex test might confirm the results
of this multispecies test?

7.2 Results and Discussion - Microbial Populations

The microbial process working group in the terrestrial ecosystem
properties workshop (Section 5) considered the soil microbiota in
terms of their contribution to ecosystem processes, that is, as a
black box that changes the chemical characteristics of its substrate.
In this workshop, Working Group A considered the internal structure of
the black box at two levels. The first level consists of the basic
interactions between pairs of populations: predation, parasitism,
competition, antagonism, and mutualism. The second level is the
community characteristics that are immediate products of population
interactions: taxonomic composition, diversity, succession, and
resistance to invasion. Each of these areas is discussed in the
following sections.

7.2.1 Population Interactions

The feedback mechanisms associated with population interactions
tend to moderate the effects of changes in the soil environment
resulting in a weak homeostasis. Disruption of these interactions can
be experimentally examined. The potential utility of these
experimental systems for testing the effects of chemicals is evaluated
in Table 7.1.

(1) Predation and parasitism. Predation and parasitism of
bacteria can only be studied with reasonable ease in liquid cultures.
Addition of a clay suspension to the cultures may provide an
approximation of conditions in the soil. An example of such a culture
system is the one devised by Roper and Marshall (1978). This system
could be used with Escherichia coli as the host and Bdellovibrio as
the parasite or with a protozoan as the predator. The system should
be run in the dark at 25°C for 12 d. The response parameter is the
concentration of host and predator or parasite cells measured over
time with and without the test chemical. The purity and viability of
the cultures must be monitored.

While the procedures for this test system are well established,
the applicability of the test to real world conditions is
questionable. The importance of these processes in the soil and the
generality of responses measured in this system need to be determined.
Field validation of this system would be difficult.
-------
no

to
z
o
t—
u
LU
y

Z.
0
p—

O
Q.
— 1
1— 1
CO
o
ce
£

ce
o
u.

00
LU
r—
to
to
1—

o
z
o
1—

1
^£

^
i
i

1-4
LU
9

in
*> S-
in re
O P—
O r—
O
•o
01 in
E >,
•r- re
t— -a
OTO f~*
C Ol Ol
•p- «- 01
e V- t.
•--30)
re o" o>
S- OJ T3
p— S- >^

Ol -p-
3 re
01
fc. V
u. o

in
u

>^
i
re
£_
Ol
C
01
u

>>
>
in
01
to

j_ ^
o o
re t-
i— t Ol
re o »-
•p- .Q in
•u re c
c r— re
01 s- s-
•U UJ *>
o -u
0- C

>,
^
JD
re
u
a.
at
ee
re
.21
'^
0
O.

in
01
H-

o o o o
0 000
«r x m CM CM

o r^ ^- CM
ro <-H «•

to to t/) to
CQ CO CQ CO

£ — i X — I

in
<:
> > > u
1 11?
(— 1

13 z I =>

ac z x i

z or z ar

^ IT) rH C\J i— 1

TJ
C J3
10 C in
E O C E
C in •!- 01 in
o -p- +J o *p-
•t- *> -r- in O C
*> -p- *J C C O
nJ in o> o> re O)
•o re CL f J= re
01 S- E U *> +>
s. re o -I- 01 c
Q. Q. O — 1 f. «

o o o
0 00
CM CM O X
r-l

oo r^ o o
CM PO tO
A

i/> to to to
CO 00 CO CO

—I —III

c
0 0
•r- 1/1
in a > i
01 o o i
i-T3 Z. 'Z. 1
0) C «t =)

3 => =1 =5

X X Z X

Z =5 Z S

CM ro in in

in
"a!

in
•r- C C
in o o
i— re -p- -p-
re 4J 4J in
> in re in
•t- -p- i — 0)
> 0) 3 u
s- c a. u
3 3 O 3
i/> u. a. to

o
**
*"^
0)
]^
*-*
1-1

E
o
u

"re
u
re
c.
o

^J
p
3
1
re
ai
t.
at

u
&
X
a>
01
+J C t-
re » o
t. 0
jc ai c o)
O)*O ^ ^ C
•p- O O C »P-
XZ_J=3 -U
re
s.
xii li=s re*

•o
ai

i(-
u
Ol
o.
w>

in
in
o
z.
J3
-------
Ill
(2) Competition. While microbial competition undoubtedly occurs
in the soil, it is very difficult to measure. We recommend that
competition be tested in flow-through systems with defined limiting
nutrients that model aquatic systems more appropriately than
terrestrial systems.

(3) Mutualism. Some microbial species enter into stable
mutualistic relationships that perform ecological functions that
cannot be performed by the individual symbionts. Toxicants can
disrupt these mutualistic relationships.

(a) Lichens. The algal-fungal symbiosis that constitutes
lichens has been shown to be sensitive to air pollutants and to
physical factors that cause imbalance in growth rates of constituent
species. Lichens are important components of tundra ecosystems and
appear to contribute significantly to nitrogen dynamics in coniferous
forests of the Pacific Northwest.

The primary response parameters are respiration, measured as
either 02 consumption or C02 output, photosynthesis, which may be
measured as 14C02 assimilation, and the ratio of photosynthesis to
respiration (P/R). In addition, N2 fixation can be measured by
acetylene reduction in lichens that contain N2-fixing blue-green
algae. Test chemicals may be applied as gases, mists, or soaking
solutions. Care must be taken to ensure that the experimental
conditions are appropriate, that is, that the control thalli remain
healthy.

Lichens are particularly useful for measuring the effects of
chemicals transported in the atmosphere. The P/R ratio may be a good
indicator of responses that lead to overgrowth of one symbiont by the
other. The mode of response of lichens to toxicants is poorly
understood and their sensitivity has only been demonstrated for sulfur
oxides and gaseous oxidants.

(b) Methanogenesis. Methane production can occur in chronically
wet soils, rice paddies, swamps, and marshes. The "methanogenic
consortium" performs interspecies H2 transfer between heterotrophic,
anaerobic bacteria and H2-consuming methanogenic bacterial symbionts.
Methanobacteriurn strain M.O.H. and the so-called S organism are used
in the test system described by Bryant et al. (1967). This mixed
culture converts ethanol to methane according to the following
reaction.

S Organism ^— C05

CH3CH2OH CH3COOH + H2-
^\ /
Methanogen
-------
112
The S organism is inhibited by its own product, H2. But the H2
is needed by the methanogen, which converts it to CH4 and thus keeps
H2 levels low enough for the S organism to continue converting ethanol
to H2 and acetate. A chemical may disrupt this system by affecting
either the S organism or the methanogen.

The test could use the procedure of Miller and Wolin (1974) with
the two membered culture growing on ethanol, under strictly anaerobic
conditions. The test chemicals should be dissolved in a prereduced
mineral salts solution and introduced into the culture. Methane
production should be measured using the syringe displacement method
(Healy and Young 1979) or gas chromatography. Exposure to any 02 will
completely stop methanogenesis. Thus, rates of zero are suspect and
should be repeated.

The test only applies to the few terrestrial ecosystems where CH4
is naturally produced, such as landfills, swamps, bogs, rice paddies,
and flooded soils.

(4) Antagonism. Some plant diseases are precluded by
antagonistic relationships between microorganisms normally associated
with the plant and pathogenic populations. For example, Rhizoctonia
sol am' is a common plant pathogen having a wide host range. The
presence of Trichoderma spp. in soil associated with the host induces
suppressiveness to R. solani, effectively controlling the disease.
Similar suppressiveness may be induced in certain soils by the
presence of fluorescent Pseudomonas which are antagonistic to the
Fusarium (vascular) wilt pathogens. Pseudomonas spp. are also
important in plant growth responses. We hypothesize that the addition
of a test chemical can alter the host-pathogen-microbial interactions
characteristic of these systems.

Soils have become suppress!ve to plant pathogens when crops are
grown in monoculture. A model system that demonstrates this
phenomenon consists of repeatedly planting radishes at weekly
intervals in soil infested with R. solani (Hem's et al. 1978; Rouse
and Baker 1978; Hem's et al. 1979; Wijetunga and Baker 1979). In a
typical experiment containing an initially low inoculum density, 100%
disease incidence occurs after three replants. By the fifth replant,
however, disease incidence has decreased to almost zero. Associated
with this is the increase in propagule density of Trichoderma spp.
from an initial density of 102 propagules/g to 106 propagules/g at the
end of 5 weeks. The effects of a toxic substance on these
interactions can be examined.

To maximize replicability, certain factors of environment and
inoculum are important. The most important environmental factor
influencing the system is soil pH. Trichoderma is most active in acid
soils. Thus, the soil should have a natural or adjusted pH of around
6. Effects of the added chemical on soil pH would have to be taken
-------
113

into consideration. The other important factor is the size of the R.
solani inoculum. When propagules (sclerotial) of size > 250 urn are
used as inoculum in alkaline soils, suppressiveness is not induced,
and there is no influence on propagule density of Trichoderma.
However, suppressiveness is readily induced in acid soils when
propagules of size > 250 urn are used as inoculum. If propagules of
size < 250 urn are used, suppressiveness is induced and Trichoderma
spp. increase in either alkaline or acid soils.

(a) Test procedure. Ten containers, each with 100 g of soil at
a matric potential of 0.7 bar, are each seeded with 32 radish seeds.
These are arranged in a radiating pattern with the aid of a vacuum
seed planter. In five of the containers, inoculum of R. solani is
introduced into the center. Five others are left as noninoculated
controls. After a 1-week incubation period, a conducive index (CI)
can be determined (described later). The inoculum is grown in a
chopped potato-soil mix and is composed of both large and small
propagules. If acid soil is available, this mixture can be used. If
alkaline soil is used, the large propagules (> 250 urn) should be
screened out. Containers are covered with clear plastic during
incubation to maintain a relatively constant matric potential.

The chemical to be tested can be mixed into the soil at different
concentrations at the time that the initial matric potential is
established. One week after the first seeding and after the CI is
determined, the plants are uprooted. All replicates of each treatment
are combined and the soil redistributed (as before) in the containers.
Thirty-two radish seeds are planted again in each container. This
process is repeated at weekly intervals. Usually the soil develops
suppressiveness in about 5 weeks (replants). At this time, fresh
inoculum can be introduced into the center of the pot and the CI can
be determined again.

During the period of the test, ambient laboratory temperatures
are satisfactory for incubation. Light is supplied but need not be of
high intensity because radishes are not grown for more than a week.

Development of suppressiveness in the soil during the course of
the test is associated with increase in propagule density of
Trichoderma spp. Some soils in nature contain low numbers of this
microorganism (102 propagules/g in Fort Collins clay loam). One
naturally suppressive soil from Bogata, Columbia, contains 8 x 10s
propagules/g. There are some soils that do not contain Trichoderma
spp. , and, in these, conidia of this fungus should be introduced at
the beginning of the test.

(b) Analysis of results. Response parameters include CI and
disease incidence (DI) for radishes grown in monoculture, the inoculum
density of R. solani, and the propagule density of Trichoderma spp.
during the course of the test.
-------
114

CI is determined by the following equation:
where A is the number of healthy plants in the noninoculated control
and B is the number of healthy plants in the inoculated treatment.
The limits of the index are 0 to 1:0. If the CI is 0, the soil is
completely suppressive, and if the CI is 1.0, it is completely
conducive. DI is computed similarly:
The difference in CI and DI lies in the experimental design. The
CI is computed when inoculum is introduced into the center of the
container and measures not only inoculum potential but also the
ability of the pathogen to grow in soil. The DI is measured when
inoculum is distributed randomly in soil after mixing and measures
largely inoculum potential. There is no rationale for transformation
of raw data accumulated to propagule densities of R. solani and
Trichoderma spp. A graphic display over time is sufficient.

In most instances, differences (treated below) are so dramatic
that statistics need not be applied to confirm differences. However,
analysis of variance can be used if needed.

I. Criteria for rejection of test results. When the test is
done properly and essential parameters adequately
controlled, results in nontreated controls are predictable.
The initial CI is usually 0.85± in a conducive soil. DI
increases in inoculated treatments and is near 1.0 after the
third replanting. By the fifth replanting, DI is quite low
and may be near 0.1. The inoculum density of R. solani
rises until the third replanting, diminishing subsequently
to undetectable levels by the fifth week if small propagules
are used initially. Trichoderma increases from barely
detectable levels to near 108 propagules/g soil in the
5-week period in inoculated treatments. If these phenomena
do not develop in the controls, the validity of the test is
questionable.

2. Interpretation of results. The test chemical may change
these interactions by modifying the CI or DI, which would
indicate changes in the antagonistic interaction of R.
solani and Trichoderma spp. Changes in these interactions
can be precisely monitored and, thus, readily detected.

3. Generalization to other terrestrial systems. Currently, in
the research area of biological control of plant pathogens
-------
115
in soil, interest is being centered on two groups of
biocontrol agents that apparently contribute substantially
to the level of suppressiveness found in soils. Species of
Trichoderma comprise one of these groups. These fungi are
microparasites and are representative of this type of
antagonistic relationship among microorganisms found in
soil.

(c) Development needs. The test may be validated in the field
to see if a chemical has the impact observed in the laboratory. There
appear to be no major technical questions requiring resolution before
using the process.

7.2.2 Community Properties

There are two types of community properties. The first consists
of properties, such as fungistasis and survival, that represent the
action of the community as a unit on an indicator species. The second
type consists of properties such as diversity, succession, and
relative population levels that indicate the structural state of the
community.

(1) Survival test. After a period of microbial activity in
soil, environmental conditions become less conducive for continued
growth and microorganisms produce propagules. The length of time that
these propagules are capable of surviving is a function of numerous
biotic and abiotic factors. The hypothesis of this test is that a
chemical may influence survival directly, by acting on the introduced
microorganisms, or indirectly by acting on the soil community.

Small containers, each containing 100 g (or less) of nonsterile
soil, can be infested with the test organisms. These are incubated
under standardized conditions of moderate temperature and moisture.
No light is required.

Controls with and without the introduced organism should ensure
that propagule density of the introduced microorganism is being
monitored and not the contaminants. Five replicates should be
sufficient.

The elements that may be found in a typical survival curve are
diagrammed in Figure 7.1.
-------
116
Survival
(per unit)
Increase in propagule density
Introduction of
^microorganism into
soil
Relatively rapid death of
propagules ill adapted to
the soil environment

Long-term survival
of resistant unit
Time Extinction point

Figure 7.1. Typical survival curve.
When a microbial population is introduced into soil, propagule
density may increase because of the availability of substrates in the
soil. Some microorganisms may be introduced with the substrate on
which growth took place, for example, dead tomato stems containing
microsclerotia of VerticiIlium dahlea. In such cases, decay of the
substrate releases the propagule units so that there is an apparent
increase in density when the individual units are counted in assays.

After this period, there appears to be a relatively rapid
increase in death rate because of the different capacities for
survival among the propagules. A proportion of the units may persist
for relatively long periods and are relatively resistant to the
insults inflicted by the soil environment.

Because these resistant propagules represent long-term survival
and are likely to survive in soils in nature, they should be assayed
for persistance. Thus, after the system has reached equilibrium, the
test chemicals should be introduced into the soil at various
concentrations.

Test microorganisms should be selected that are typical members
of soil microbial communities. Also, techniques should be available
for determining propagule densities as aliquots of soil are assayed
over time. We suggest bacteria such as Rhizobium and Agrobacterium
tumefaciens as likely candidates because their cell densities in soil
can be followed quantitatively with the amino fluorescence technique.
Rapid assays, selective for Rhizoctonia solani (a fungus), are also
available. In this case, the persistence of distinct propagule types
(large and small) can be followed. The small propagules would be
particularly susceptible to insults.
-------
117

If the chemical reduces survival, it would be detected as an
increase in death rate over time in comparison with a nontreated
control. The extinction point need not be measured because
interpolations and extrapolations may be obtained when the data are
transformed by methods described later. Although the time required
for the test would depend on the test organism, this obviously
shortens the time to a matter of weeks rather than months or years.

Various transformations have been suggested. The semi log
transformation assumes that propagules die at a logarithmic rate:

In y = yQ + rt

The symbol y equals the number of viable units at a given time
(t), and y equals the number of viable units at 0 time. The rate (r)
is negative because it describes the rate of death.

It may also be assumed that susceptibility of propagules to death
follows normal distribution in time. Thus, the log-probit
transformation may also be a legitimate candidate for mathematical
descriptions of survival. In practice, the proportion of surviving
propagules in probit units is plotted against time on a logarithmic
scale.

Research has yet to be done to establish which of these
transformations is better for data analysis. One study indicated that
they were complementary. In an analysis of data obtained from various
examples in the literature, the log-probit transformation appeared to
give a better performance; that is, it seemed to give TS50 (time
required for 50% of the propagules to die) values and TS10 points much
closer to those observed in nature.

Slopes of transformed curves may be subject to regression
analyses to obtain slope (r) values for comparisons. Test
microorganisms can be selected that are representative of the
microbial community.

(2) Fungi stasis. Fungi stasis is the failure of viable fungal
spores to germinate in soil. Because the fungistatic activity of soil
is removed by sterilization of the soil, anti-fungal activity is
believed to be related to the activities of other microorganisms in
the soil and is probably related to tha ability of the soil microflora
to withstand invasion of alien species. Several mechanisms have been
proposed, but a satisfactory explanation of fungistasis has yet to be
accepted. This test system is designed to determine if potentially
toxic chemicals will alter the interrelationship between fungal
propagules and the rest of the soil microbial community.

The test should be run in the dark at room temperature. The test
chemical is added to the fungistatic soil, which is then placed in a
Petri dish. A suspension of spores from species such as Fusarium
-------
118

sol am or Aspergi11 us flavus (nutrient dependent) is passed through a
0.20-jjm pore size filter so that the spores can be retained on the
filter surface. Final spore density should be ~2000 per mm2. The
filter is then buried in the soil, and the Petri dish is covered.
Plates are incubated at room temperature ~14 d (a time series may be
run), and the percent of fungal spores that have germinated is
determined microscopically after staining the filter with Lactophenol
cotton blue. At least 1000 total spores per filter should be counted.

The following series of control tests should be run to elucidate
the nature of the test chemical's effects:

Steam sterilized soil Fungal spores should germinate.
Potential toxicant may act at
this level and prevent spore
germination.

Chemically sterilized Plus 10 to 100 ul of ethanol/g of
soil soil. All spores should germinate.
(Other mixtures that may be more
reliable, such as amino acid-
carbohydrates, should be tested.)

Non-sterile soil Concentration of the test chemi-
cal equals zero. No, or few,
spores should germinate.

Soil not fungi static, as evidenced by a high degree of spore
germination in non-sterile soil in the absence of the test substance,
would cause rejection of the test. If fungal spores germinate in the
presence of the test substance, then the natural fungi static
properties of the soil have been adversely affected by the chemical.
Because fungi stasis in a soil is strongly correlated with the
activities of the indigenous microflora, the test substance has
interfered with the interactions between these microflora and the
invading fungal spores. Fungi stasis has been found in all natural
soils that have been examined. The degree of fungistasis is limited,
however, in soils high in organic matter such as peats and mucks.
Neither deep subsoils nor very acidic soils are fungistatic.

Laboratory development, interlaboratory testing, and field
validation are necessary. To our knowledge the effect of any toxicant
on fungistasis has yet to be assessed in this manner. The system
needs development to determine adequate spore density per filter,
length of incubation, soil moisture levels, etc.

(3) Population levels. Relative population levels of microbial
species and taxonomic or functional groups could be determined as a
measure of interference of toxicants with normal population balance
mechanisms. Because of the difficulty of enumerating species, the
following ratios are probably most useful.
-------
119
1. Fungus/bacteria ratio. Enumerate bacteria and fungi by
direct counting with acridine orange epifluorescence
counting. This does not indicate viability.

2. Viable fungal propagule/bacteria ratio. Enumerate
fungi and bacteria by viable plate count methods. Use
Saboraud Medium for fungi; Trypticase soy agar for
bacteria.

3. R/S ratio. Enumerate bacteria by direct count or
viable bacteria (plate count) in rhizosphere (R) and in
root-free soil (S).

Measure these ratios in soils and perform ANOVA to determine if
addition of toxicant significantly alters the ratio. It is not
possible to state the ecological significance of these ratios, but
they are generally believed to be important.

(4) Succession. Succession is an important process that can be
measured only at great cost. Significance of deviations of
successional processes often would be difficult or impossible to
evaluate. An exception would occur with preemptive colonization by
microorganisms, such as Lactobacillus, that prevent further
successional events. Preemptive colonization would be easily detected
by monitoring other processes such as respiration. Therefore, it does
not appear reasonable to examine successional events for microbial
populations in soils or on leaf litter. Some early successional
stages can be examined in flow-through systems and should be
considered for aquatic ecosystem testing.

(5) Diversity. Diversity is a community-level parameter that
may be used to measure stress in microbial communities. Stressed
communities often have low diversities that can be quantitatively
assessed by a variety of diversity indices such as the Shannon Index.
Almost any substance can cause a shift in diversity in the microbial
community. Normally, the diversity rapidly returns to its original
level following a minor environmental insult. Exposure to a
persistent toxic substance may result in a prolonged depression in
microbial diversity.

Diversity can be assessed using the techniques of numerical
taxonomy (Kaneko et al. 1977) or perhaps by direct microscopic
observation (Staley et al. 1980). If numerical taxonomy is used,
numerous isolates will be needed. Diversity may be used for
monitoring the long-term major impact of a toxic substance. Measured
changes in diversity can potentially be applied to all microbial
communities, but the consistency of the response between communities
and its validity in the field are unproven.
-------
120

7.3 Results and Discussion - Plant and Microbe Populations

Working Group B considered systems that display interactions
between species of vascular plants and between vascular plants and
microbes. These classes of interactions are closely related because
of the dependence of the competitive ability of a plant on its
relation to microbial symbionts. Because there are no existing test
systems for these classes of interactions, synopses of potential test
systems are presented that are derived from the research experience of
the working group participants. Each proposed system was evaluated
and rated by consensus of Working Group B. The results are presented
in Table 7.2.

7.3.1 Interference

Interference between plant populations includes competition for
space, light, water, and mineral nutrients as well as allelopathy and
other modifications of the medium. The best understood method for
examining this process in the laboratory is to sow seeds of two plant
species into pots. Two designs are possible. In the additive design,
the density of one species is kept constant while different densities
of a second species are added resulting in a variable total density.
In the replacement series, the total density is kept constant while
the proportions of the two species are varied.

Before a pair of species can be used, preliminary studies must be
performed to determine appropriate pot size, nutrient levels, and
total densities. Density selection is based on the following model.

1 •• ' in
Yield
Density

In Phase I, little or no interference occurs and yield is simply
the product of density and age-specific plant weight. In Phase II,
yield is constant, but individual plant weight declines because of
interference. This is the optimum density range for the test system.
In Phase III, yield is constant, but mortality occurs. Controls
should include each species grown alone. To permit allelopathic
interactions, a sandy loam soil should be used, and watering should
not leach the pots.

The following parameters should be considered:

1. Number of individuals for each species.
-------
121

o
t—l
1—
o
o:
UJ
z

r—
z
«J
1
z
1
G_

Q
I
UJ
CO
o
ce
0
t-4
I
I—
1
_J
o.

ce
o
LL.
LU

V)
^»

UJ
(—

LU
O

•z.
o
t—

(—

LU
_j
CO
^
1—

5_ r—
at re
CL u
*-> E
in 0>
0 J=
U W
0)

K
1*- Ol
0 C
i — C
Ol T-
> re
01 S_
—i *->

u
c
01

or
01
Lu

>,
C
o

re
s.
o

re
i.
01
c
H-l

U)
<0

^
/-x
it?
TD
•o
1

3
cr
?!
01
3

re
V-
„_
0
in
U
^J
V)
Q
55

£
re

01
c
O)
^
•i—
•P

in
C
Ol

^
Ol
1*-
in
c
!_

4->

re
u.

..
a.
ai

4->

a>
t—

o
o
o
r^.

1/1
CO

is
< CJ

S
ai u
u in
C 01
ai t-
S. 1
Ol S-
t $
ai o
c "o
i—* "^f

o o o o o o o
o o o o o o o
o in o CM o .-i ID
rH PO CM i-H

00 <&• ^* rH rH rH CM
CM 00 rH CM CM CM ^~

•a ~a
CO UO I/) tO 1/1 l/> tO
oo z s: co oo co co

z> z s: — ' — ' —i — '

^ < < < *t <
> > > > > > 1
o o o o o o i
I i i ill1

U u
^C T* 1C ^ • _j s"

= 3. = 3: z a: s:

£ £ £ z 3: z z

01
ro CM CM CM CM CM CM

01
01 T3
01 Ol E O
re re 3 *>

t- i- t- I -U r— C
3 E i- t- E « r-
4->V>O OI-33Q4J
r— O U UOI'^fi-OlO
•r~ £» ore oc M re ji 4^
(-U-OS. *J O — 01 O O
01— coiuufj: i- o
«£ E UJ i UJ I 0:300:

,3
^

O
•o

in
in
g

o
0

c
0

re
u

CL
D.
re

01
a.

4->

•u
in
Q
D.

c
Ol O

t- 0 0
.C 01 C U
CCT3 » J<
5I3S t
Ol

^
z af -li i» re .

<->
u
Q

01
in
3

C
ai
ai
L.
o>
•o
o
o
01

m
at
in

lt_

u
ai
Q.

in
Ol

U
ai
Q.

fmm

•c

O
C

re
01
re
N

f
U

in
cu
3

r—
3
U

§
0
o
c
•^

c
3
c

•o
c
re
in
'JZ

in
01

•o
ai

•r~
3
o-
01
i.

Ol
c

c
re
c-

it-
o

JZ
Ol

5^
O
re
o
J3
ro
T
ai
c

o
,_
re
•u
c
ai

o
Q.
01
**

c
re

Ol
re

in
i
o
u
1!

3
U
o
c

u
L.
Ol

•""
ai

3
i

Ol
c

.(3
re
a>
iE
ai
-------
122
2. Shoot and root dry weight; shoot weight alone is
frequently adequate.

3. Nitrogen concentration of tissues.

4. Nitrogen fixation rate of legumes.

5. Concentration of other nutrient ions (P, K, Ca, Fe,
Mg).

6. Seed production - if time allows.

From these data, species ratios, mean weight per plant, and mean
weight per pot should be calculated.

The following pairs of plant species are potentially useful in
this type of system:

1. Unstable perennial systems

Trifolium subterraneum - Lotium sp.
Trifolium repens cv Tillman - Festuca arundinacea
cv Kentucky 31;

2. Early successional systems

Helianthus annus - Digitaria sanguinail's
Sorghum halepense - Bromus sp.
Erigeron canadensis - Aster pilosus
Ambrosia sp. - Amaranthus retroflexus;

3. Agricultural systems

Zea mays - Sorghum halepense
Avena sativa - Hordeum sativum
Glycine max - Panicum sp.
Helianthus annuus cv Russian mammoth - Amaranthus
retroflexus;

4. Successional systems (later stages)

Andropogon virginicus - Pinus taeda (seedlings
inoculated with mycorrhizae).

Seed germination is good for these species, except for
Amaranthus, which needs a temperature greater than 25°C; Ambrosia,
which needs a cold period; and Sorghum halepense (Abdul-Wahab and Rice
1967). It helps to remove "seed coats" for consistent germination of
Helianthus.
-------
123
Experiments with these species can be completed within 8 to 10
weeks. Information regarding the above listed combinations are
available in Harper (1977) and Rice (1974 and 1977).

The clover-fescue pair was considered the best single candidate.
Experimental procedures have been developed, and results using
pollutant gases as stressors have been published. A synopsis of the
proposed procedures for this test is presented below:

Conduct the test in a greenhouse with 2- to 2.5-L capacity
plastic pots and a sandy loam-quartz sand mixture at a 3:2
ratio by volume. The temperature range should be 23° to
27°C. Water as needed (use tensiometer check), but water
all plants short of saturation. Fertilize with
VPHF - 5-15-10, a commercially available, water-soluble
fertilizer with micronutrients, or equivalent, at a rate of
15 g/3.8 L H20 in a split application - 100 mL/pot at
seeding and 100 mL/pot 10 d after germination. After
germination thin to eight plants per pot.

Avoid hot summer months. Use supplemental lights (mercury
and sodium vapor) during a 14- to 16-h period to provide 70%
full sunlight; analyze the growing medium prior to the test
(cation exchange capacity, pH, mechanical analysis, organic
matter, ion content, etc.); be sure that the N level is not
so high as to suppress N-fixation by Rhizobium (this also
keeps fescue in check); add Rhizobium inoculum prior to the
test to standardize legume performance; 4 weeks after
germination, add the test compound, then harvest the tops of
the plants 2 weeks later; and determine dry weight of the
total biomass for each species.

7.3.2 Mycorrhizae-Plant Interactions

Test systems are proposed for both of the major classes of
mycorrhizae-endomycorrhizae and ectomycorrhizae.

(1) Endomycorrhizae-grass. This test incorporates effects on
formation of the endomycorrhizal association as well as direct effects
on cereal crops. A synopsis of the proposed test procedure is
presented below:

Several candidate grasses could be used: sorghum (2 to 3 plants
per pot), millet (2 to 3 plants per pot), and corn (1 plant per
pot). This test would be run in the greenhouse with pots,
growing medium, temperature range, lights and water similar to
those specified for the plant interference test. Fertilization
could also be the same except that phosphorus levels must be
maintained below 45 ppm. A multiple species mixture of Glomus
spp. in the form of root fragments (infected) and adhering soil
containing fungus spores can be used as inoculum (Kormanik et al.
-------
124
1977). Add the root-soil inoculum to growing medium (1:10 by
volume) prior to filling pots.

Ten days after germination of the plants, add the test chemical.
Twelve weeks after seeding, terminate the test, remove the roots
and rinse them carefully; remove 2 to 4 random root samples (5
g/sample fresh wt.) from the roots from each pot. Clear and
stain the roots by the method described in Kormanik et al. (in
press). By microscopic examination, determine the frequency and
degree of vesicular/arbuscular colonization in the selected
feeder root segments.

(2) Ectomycorrhizae-conifer. This system is well studied and
mycorrhizal inoculum may soon be commercially available. A synopsis
of the proposed procedure for this test is presented below.

Employ standard tree containers (125 cc capacity each
cavity), loblolly pine seed, and peat-virmiculite growing
medium (1:1 by volume). Prepare a vegetation inoculum of
Pi soli thus tinctorius dried to a bulk density of 350 g/L and
mix to a ratio of 1:15 in the medium. Water as needed.
Fertilize at 3 to 4 week intervals (NPK = 200:20:40).
Greenhouse temperature should be 24 to 27°C. Supplement
light as needed to give a 14- to 16-h day. Use an
uninoculated control. Add the test chemical at seeding.
After 1 or 2 weeks, remove 5 to 10 seedlings per tray and
visually examine the roots for percent ectomycorrhizal
development.

A growing medium for ectomycorrhizae is described in Marx (1969),
inoculation procedures are described in Marx and Bryan (1975), and
container production of inoculated seedlings is described in Ruehle
and Marx (1977).

7.3.3 Rhizobium-Legume Interaction

While the clover-fescue interference test provides a test of the
rhizobium-legume interaction, a test may be desired that does not
include a second plant species to complicate interpretation of results
or that simulates a legume row crop. The proposed procedure for this
test is presented below.

Start with a very low nitrogen soil medium. Rhizobium
inoculum should be added to soil prior to potting.
Watering, growth medium, pot size, light, and greenhouse
temperature can be the same as in the interference test.
Seed four beans (Bush-Blue Lake 290) and thin to one per
pot. The test chemical should be added at planting. The
test should be run with and without N fertilization. Run
the test for 21 d. Measure the above-ground biomass
production, describe visible signs of injury to the plant,
and visually assess nodulation on root systems using broad
categories: < 5%, 45%, 70%.
-------
125
7.3.4 Wheat-Wheat Rust
This system includes a plant (Triticum spp.) and fungal pathogen
(Puccinia graminis var. tritici) that are of major economic
importance. The procedures, developed by the USDA Cooperative Rust
Laboratory at the University of Minnesota in St. Paul, should be used
for culturing, inoculation, and grading the reaction of the wheat to
the rust (Stakman et al. 1962). The combination of wheat and rust
varieties should be selected so that control plants give a moderately
susceptible response to the rust fungus. The chemical should be
applied at planting. Changes in the rust reaction (size of uredia)
should be recorded 10 d after inoculation.

7.3.5 Carrot-Crown Gall

This system provides a simple and compact demonstration of
bacterial plant pathogenesis. The NASA procedure (Wells and Baker
1969; Kleinschuster et al. 1975) for crown gall is recommended.

Cut carrot disks and place them on moistened filter paper in
a Petri dish. Place the bacterial inoculum and the diluted
test chemical on the carrot disk. After 21 d determine the
fresh weight of the galls.

7.3.6 Plant-Nematode Interactions

Nematodes are important components of the soil biota, and the
rootknot nematode is a significant agricultural pest. This test
system would utilize the procedure from the screening for resistance
test developed at North Carolina State University (Taylor and Sasser
1978). The test chemical and nematode egg masses would be added to
the soil at the same time the tomatoes are transplanted. After 5 or 6
weeks, giant cell development, extent of galling and nematode egg
production would be determined. Tomato varieties that have been
identified as moderately resistant (Sasser and Kirby 1979) should be
used.

7.3.7 Agricultural Soil Microcosm

This system represents competition between agricultural and weed
species. The following procedure is recommended for this test.

Collect soil from a field that has been fallow for at least
a year to avoid extremes of fertility and concentrations of
agricultural chemicals. Soil pH should be in the range 5.5
to 6.0. Screen the soil, and mix with seeds of clover,
horseweed, crabgrass, and fescue. Fill wooden flats
approximately 40 cm x 15 cm x 6 cm deep; add the test
chemical; place the flats in a greenhouse; water as needed
but do not fertilize. Measure rate of emergence, survival,
and biomass at termination (4 weeks).
-------
126
7.4 Results and Discussion - Arthropod Interactions

Working Group C developed a set of "Type Arthropod Interactions"
that were considered to have potential for evaluating the impact of
toxic substances on terrestrial population interactions. The
interactions were categorized as follows:

1. Arthropod interactions with plants as phytophagous
feeders

Sucking feeders
Grazing (chewing) feeders

2. Arthropod interactions with biotic mortality factors
(exploiters)

Parasiteid
Predator
Pathogen

3. Interspecific competition

4. Symbiotic interactions

Interspecific symbiosis
Intracellular symbiosis

5. Functional interactions between sucking, grazing
arthropod on a single plant unit

6. Host plant competitive interactions as mediated by a
phytophagous insect.

The best studied sets of species for each of these interactions
were identified and listed. From these lists, species sets were
selected that are either proposed for development (Sect. 7.4.1) or
that show some promise but cannot be recommended at this time (Sect.
7.4.2). Finally, a tentative test protocol is presented for
competition between Tribolium species. Each of the interactions
listed above is included in at least one of these systems.

7.4.1 Proposed Test Systems

The systems described below were designated as having the
greatest potential for evaluating the effects of chemicals on
arthropod population interactions. These systems use relatively
well-studied species and can include more than one of the type
interactions. The systems are evaluated in Table 7.3, and the
constituent interactions are ranked in Table 7.4.
-------
127

to
£
UJ
1 —

to
UJ
1—
a
LU

i
o
Q£
n^

LU
x
l—
u_
0
»
o
NH

^£

i
«c

<
(3
5
I
^

r*.
UJ

S
t—

O> «A
£ -o
10
C O>
•i- at
C I.
•i- OJ
S*
t— s^
a>
03
a,.^
3 re
U- O

in
u

^j
tn

^j
(O

£>
-p-
.J
+J
•f"
tf)
c
a>
CO

^
t.
o
.s t~
ID ai
fi- H-
O IS>
J3 C
a re
r— i.
(- -U
0)

*J
^
JO
re
u

f—
CL
ai
ce

c
re
ce

S
ai

(A
3?

in o
^0 VD O O OO Csj
LD ID O^ rorO r-HOO^"
II 1 II 1 CO 00 1
CSJCSI OO^-O O Or-HiHO
GO ro i— t CM co icvvro
VI
•°.£ . . ....
re u>\fc/>i to tot/>t/>to
to to ... i . ....
co£ xtocoi x xcacom
m

33 =3=>_l=> _J =>=23— 1

i
CD
L) O>

iniA lAinin O)C IAIA in
aia> a>oia>i ma) o>a>i o>
t— I— 1— 1— 1— i oiin t— 1— II—
II 1 1 1 1 S- J3 1 1 1 1
(—1- h- 1— 1- Q. < t-(- t-

X => — 1 £ =3 X X ££3 X

XX X X X £ X XXX X

XX XXX£ X X -1 — 1 X

r-l CM CO «*• in ID

•a
§.r-
o in
••- *> T3
ai IA o
•u re *>
0 S. -r-
§ja -F- a. re
1 O 1 t-
•*- t.ino> +J>)>>rea)
in a>3-o -i-i— r- a i—
t- o i- o lAi-v-ire
ere j=->--i-> ro S 5 T3 u
reo Ecu>re t. o o -r- in
OlC t. i- -r~ E re •— •— J=
ja uj o o i- a. ID a. JD ja a *>
ii sot— z c E iii re n-
i— r— •— tj e s s s -t- ,— —,— ,-« «
U-4-H- 01 S-(.S.t- i— l<-it-lt-t-
OIO)Q) OOOO O 0)0)0)0) r*~ C
4J4J+J C22jt2 J3 lAiAininre jt
•^•t-.|- (.UI.S.C -r- 3333<*- O
J=J=JT orererere t- 0000-— s-
S S S o uJ LU uj uj i— xxxx< m

ID l£> ID VO
in in in m
iiii
co oo co co
CM CM CM CM
l/i to l/> to
£ £ £ SI
^^ ^^ ^^ ^.

Irt l/> tO t/)
CO CO CO CO

—1 —1 —J —1

X £X£

X X X X

X £ X X

•o

'5 1-
<-> 0
0) a> m re
•— i— re -a
re re s- o>
u u re t-
IA in Q. Q.
i i i i
*> 0) 0) 01
C i— i— i—
i— U O U
o- 10 to to

T3
01
re
u

•o
c

c
o
•£
re
u
3
•o
0)

a>
^

c
o
>»
J3
0)
E
re
^j

•a
0)
.^!

re
u
01
CL
in

0)
.,_
3
er
01

in
1

0)
a> IA i—
*J C 0) 0
re S j= o
S- O *> J=
J= 0) C U
a>-o jt j^ n- ui
•r- O O C O
X £ -J => J=
^- 0)
< X
x £ -i =3 re ja
-------
128

re
to
o

u
re
a> s-

>,
£
re

01
c
01
o

c
o

u
re

a>
c
*-H

CO rH CM *S" CO rH f— 1 rH CM ^f *tf° CO rH CO CM rH

1-HrHrHCOrHCMCMl-H
rH^-rHrO rH^J'^'^'rHrOCO^- CMCVJCOrH

CO rH CM ° 3 -"2 •!- £ r-
s- •— v>o-<-(-oo(-in t-

> xt-rere> ••-•>-•>->(- o
•r- aireEt-inmrerec •—
JD ia.airoa.(oreTJo.o ^
t- ai i c D.Z s- s- ai i -r- IE
ai>> j- >, c i i ID ro i- >> +J >>3
JZ^E Or-EEEQ-C- O.C- •- •— -1-
IH-4- >«-(-i-i-IIIO- *JlBUI«t-i—
^j oio^joi '^ aioop^oioioi 01 "001010

re '»-s»jcre t. •p-^CCjzreres s jcre3*r-
•— J= re o. u aij=rerereo.uuo o 0.001-

w
O
u
•-
>>
•r- il
> ai

•i- 'o
W r—
c a.
01 X
in ai
•o ai
C L>
re o

^ IS
•r- fc
1 — 01
re -c
£.
ai <•-
c o
01
01 >,
^ •!-
01 >
-C 4->
1 '^
1 l/>
01 C
i— Ol
.a in
ra
(- ai
0 £
> *->
re

ai
01 •i-'
i- re
re u

rH "5
C
0 "~
c
in o
Ol •!-
C -U
•i- O
•u re
re t-
re J3
-------
129

(1) PIant-whitefly-paras1toid. The greenhouse whitefly is a
sucking herbivore that can use tomatoes, cotton, and a variety of
other domestic plants, but beans are recommended for laboratory
testing. It is exploited by the parasitoid Encarsia formosans, which
is available commercially. Both insects have 21-d life cycles at
20°C. The system is documented in Burnett (1967).

(2) Corn-earworm-exploiters. The corn earworm, a lepidopteran,
chewing herbivore, is an important pest of corn. Earworm exploiters
that can be used in the laboratory include the egg parasitoid
Trichograma, the pathogens Bacillus thuringensis and nuclear
polyhedrosis virus, and the nematode Neoapplectana carpocapsal. The
nematode is also the vector for another pathogen, Achromobacter
nematophilus. The system is documented in Starks and McMillian 1967.

(3) Alfalfa-aphid-parasitoid. The alfalfa aphid, Therioaphis
trifolii (T. maculata) is a sucking pest of leguminous crops. It is
exploited by the parasitoids Praon exsoletum (P. palitans), Trioxys
complanatus (T. utilis), and Aphelinus asychis (A. semiflavusT!This
system presents the possibility of testing the effects of chemicals on
competition between the parasitoids as well as on the
plant-herbivore-parasitoid food chain. These insect species are not
available from stock cultures. The system is described in Force and
Messenger (1964a, 1964b, and 1965).

(4) Plant-brown scale-exploiters. The brown soft scale (Coccus
hesperidum) is a sucking herbivore that can be raised on numerous
plant genera including Coleus, Begonia, Ficus, and Ilex. It has more
than 35 hymenopterous parasitoids and several coccinellid predators
including Chilocorus stigma. Advantages of scale insects as test
organisms include:

1. Their sessile nature provides for manipulation and
quantification of many population parameters.

2. They share an intimate spatial and chemical
relationship with their host plant including a high
sensitivity to host chemistry such as concentrations of
nitrogen and pesticides.

3. They leave a permanent record of survival and
parasitism.

4. They have numerous predators and parasites and engage
in intense intraspecific and interspecific competition.

5. They are easily reared in the greenhouse and produce
several generations per year.

The major disadvantages of this system include the relatively
small amount of work that has been done with the system and the
-------
130
relatively high level of specialized training required in identifying
and manipulating the insects.

(5) Housefly-blowfly-parasitoid. These systems include
parasitism of the housefly (Musca domestica) by a wasp (Nasonia
vitripennis), competition between the housefly and blowfly [Phaenicia
(Lucilia) sericata Meig.], and competition in the presence of the
parasitoid. These systems require little equipment and have
well-defined media; the physical parameters have not been studied in
detail as they have for Tribolium but they are probably somewhat less
important. Although the population system is fairly simple to run and
requires little training, it would require a fair amount of labor.
Differences among fly strains used could influence results because
flies vary in their sensitivity to parasite attack and can evolve
defensive mechanisms. Both fly species are commercially available.
Documentation for the fly-parasitoid system is found in Chabora and
Pimentel (1970); for the fly-fly system in Pimentel et al. (1965); and
for the fly-fly-parasitoid system in Cornell and Pimentel (1978).

(6) Flour beetle competition. Competition between Tribolium
castaneum and Tribolium confusum is one of the best studied systems in
population ecology. Under certain well-defined conditions, the
outcome of the competition is indeterminate. Therefore, the system is
thought to be very sensitive. A provisional protocol for this system
is presented in Section 7.5. Procedures are available to rid cultures
of the pathogen Adelina, but it may be intentionally included as an
additional interaction.

7.4.2 Promising Systems

Several arthropod interactive systems were enumerated that have
good promise, but that are either limited by the unavailability of
documentation for standardization or present potential problems for
implementation. These systems are discussed briefly in the following
sections.

(1) PI ant-herbivore-exploiters.

Hemlock scales. This system includes forest ecosystem organisms
rather than agricultural pests. Several interactions are possible
involving the insect, plant, natural enemy, and competitors. The
sessile nature of scales renders them amenable to laboratory testing.
The system is documented by McClure (1979a and 1979b).

Gypsy moth. This insect is well documented because of its
periodic pest status. It is a forest ecosystem, chewing herbivore.
Methods for culturing and manipulation are documented, but there are
possible quarantine problems (Campbell and Podgwaite 1971; Capinera
and Barbosa 1977; Odell and Rollison 1966; Hugh and Pimentel 1978).
-------
131
Grasshopper-grass. This system provides a laboratory test for
plant-chewing herbivore interactions from a grassland ecosystem. The
test procedure is described in Dyer and Bokhari (1976).

Corn rootworm-grass. This system provides a laboratory test for
interactions between a plant and a soil dwelling, herbivorous insect.
It is also possible to add an insect pathogen to the system. This
system is well documented (Branson 1971; Ortman and Branson 1976).

Plant-Japanese beetle-pathogen. This is one of the best docu-
mented insect-pathogen systems and provides the advantages of a
soil-dwelling insect. Documentation for this system is old but
voluminous (Fleming 1963; Hawley 1952; Hadley and Fleming 1952).

Cactus (Opuntia) - moth or scale-exploiters. This system in-
cludes arid ecosystem organisms and allows several interactions
including the use of both chewing and sucking insects on the same
plant. Possible herbivores include Cactoblastis cactorum (Lepi-
doptera: Phycitidae), Dactylopius opuntiae (Homoptera: Dactylo-
piidae), Archlagocheirus funestus (Coleoptera: Cerambyeidae), and
Metamasius spinolae (Coleoptera: Curculionidae). Documentation of
laboratory procedures may be found in Gunn (1974); Hoffman (1977);
Lindley (1978); Moran and Annecke (1978).

(2) Predator competition. This interaction is not included in
the proposed tests, and we know of no documented systems. One
possible system would include checkered beetles (Cleridae) and bark
gnawing beetles (Ostomidae), which are both polyphagous predators
living on trees or in the bark. Another possible system would include
ladybird beetles (Coccinellidae) and green lacewings (Chrysopidae),
which both feed on aphids.

(3) Mutualism. No examples of mutualism are included in the
proposed tests. A good example might be the cultivation and
consumption of the fungus Fusarium by the ambrosia beetle (Xyleborus
ferrugineous) (Morris and Baker 1967 and 1968; Norris and Chu 1970).

(4) Plant competition mediated by insects. This system of a
greenbug on small grains includes a class of interactions not included
in the other tests. (The plant interactions group pointed out that
herbivorous insects can easily be introduced into plant competition
experiments by neglecting to fumigate the greenhouse.) It should be
easy to extend this test to include other herbivores or natural
enemies of the herbivores. Procedures are described in Windle (1979).

(5) Insect-pathogen. Dermestid beetles can be cheaply and
easily reared. The pathogen Gregornia is known to have chronic
effects on population parameters such as fecundity and longevity.

Procedures for this system are described in Schwalbe et al.
(1973a and b) and Schwalbe et al. (1974).
-------
132

(6) Interspecific competition. Competition between Drosophila
species is as well studied as Tripoli urn competition and is also
probably a good candidate system. However, we are not sufficiently
familiar with the system to evaluate it in detail or to propose its
development.

7.4.3 Protocol Development (Tribolium competition)

(1) Test descriptions. The only system for which a tentative
protocol could be developed during the workshop was Tribolium
competition. The Tribolium experimental model has been used
extensively to study competition. The system has both competitive and
predatory interactions because adult beetles feed on eggs and larvae.

A hypothesis to be tested by this protocol includes - What is the
impact of toxic substances on competitive/predatory interactions?
Though the system envisioned has only two species, the effects of
toxic substances on community diversity, trophic structure, and
stability may be inferred from the results.

Size must be considered when designing the system. Under some
environmental conditions competitive interactions between Tribolium
confusum and T. castaneum are indeterminate. In other words, the
winner cannot be predicted; rather each species wins a given percent
of the time. This has been demonstrated in studies by Park (1948,
1954) and Mertz et al. (1976). In standard media (Park et al. 1965),
at 29°C, about 25% relative humidity and constant darkness, the two
species are eventually matched. In approximately 50% of the cultures
T. confusum will win, while in the remaining cultures, T. castaneum
will win.

A description of the recommended test procedure is provided
below:

Treatments (a) 10 T. confusum and 10 T. castaneum
(b) 20 T. confusum
(c) 20 T. castaneum

Replication. For (a) treatments, a minimum of 20 replicates
are needed. For the (b) and (c) treatments, 10 should be
sufficient. Replicability can be increased by the use of
block design.

Introduction of test chemicals. It would be easy to add the
test chemical to the flour medium. Since the medium is
usually changed every 30 d when a census is taken, the
exposure could be short- or long-term. Exposures of less
than 30 d are also possible.

Measurement of effects. Censuses of adults will be taken
every 30 d. Immatures can also be speciated but with less
-------
133
accuracy. Counting immatures is also much more
time-consuming.

Outcome of competition. In the control (a) treatment, com-
petitive indeterminism should be found, whereas, in the
experimental (a) treatments, deviations from the control
treatment are a measure of the impact of the test chemical.
For example, one species might win in all of the cultures
(determinate competition). Time to extinction of one com-
petitor could also be a response criterion for treatment
(a).

(2) Analysis of results. There are numerous statistical tests
for analyzing results by comparing a control and an experimental
treatment (e.g., Dunnett's Test and Duncan's Test) and for comparing
several treatments (e.g., Tukey's Test or Scheffe's Test) (see Steele
and Torre 1980; Brown 1965). The above test could be used for the (b)
and (c) treatments, where each species is alone.

For comparing the (a) treatments, possible goodness of fit test
would be appropriate, for example, the G-statistic (Sokal and Rohlf).
The control treatment could be used to generate expectations.

If in control (b) and (c) treatments, one species always becomes
extinct or if there is a lack of indeterminacy in the control (a)
treatment, results should be rejected.

The Tribolium experimental model is a "laboratory model system."
As such, it is simpler than nature, though it is by no means simple.
The results of experiments utilizing the Tribolium model may be used
to indicate the possible effects of a test chemical on natural
communities.

There are several major questions that need to be resolved. For
example, how quickly can you predict the outcome of a competition
culture? By the third census (90 d), the outcome can be predicted
with perhaps 80% accuracy by using numerical superiority as the
prediction criterion; by day 150, predictions are perhaps 90%
accurate. Exactly how accurate are these estimates?

It might be very useful to use this test system with a test
chemical whose "effects" are well known. This would give a better
understanding of the ability to generalize results to other
terrestrial systems.

Numerous review articles are available, several of which deal
specifically with competition: Sokoloff (1972, 1975, and 1978); King
and Dawson (1971); Mertz (1972); Park (1948, 1954); Mertz et al.
(1976); and Neyman et al. (1956).
-------
134

7.5 Conclusions

The potential test systems for microbial population interactions
and community properties are not highly recommended for the following
reasons:

I. Population interactions and community properties of
terrestrial microbes are difficult to measure because
of interference by the soil. These properties can be
far more easily measured in aquatic systems if they are
deemed to be of interest.

2. The primary importance of terrestrial microbes is their
functional role in the ecosystem. Functional responses
are also more easily measured (Section 5).

3. Because the relationship between population and
community level properties of soil microbes and their
functional dynamics in ecosystems are poorly
understood, the results of these tests would have
little explanatory or predictive power.

The two best potential test systems for microbial population
interactions appear to be lichen mutualism and the antagonism between
Trichoderma and Rhizoctonia. These systems are well understood and
could be easily tested.

Of the plant and microbe systems evaluated by Working Group B,
the clover-fescue interference system received the highest rating
because it combines interactions between plant populations with
interactions between plants and microbial symbionts; it is a
well-developed system of some importance and it appears to be
sensitive. The agricultural soil microcosm is rated relatively low
because it is untried. The remaining systems that include
interactions between single plant species and their mutualistic
symbionts or pathogens were all thought to hold high promise.

Working Group C (arthropod interactions) concluded that, in
general, systems involving Homoptera will be more sensitive than those
involving lepidopterous larvae. Systems involving more than one
interaction are presumably more realistic but may require much more
time, cost, development, etc. There is no clear perception that one
or a few particular types of interaction (e.g., plant herbivore;
natural enemy-prey) is superior to the others. The major common
parameters for all interactions are fecundity, survival, and
development time. These should be determined for any test to ensure
that effects are detected. However, as systems are developed it may
occur that one or more parameters are impractical. For example,
measurement of fecundity and development time for nematodes in the
earworm-nematode-bacterium system would seriously complicate use of
the system.
-------
135
Some of the serious questions that were left unresolved are:

1. What is the most appropriate method for applying the
test chemical?

2. What are the criteria for validating these systems in
the field?

3. What magnitude of effect is significant?
-------
136

7.6 References

Abdul-Wahab, A. S. , and E. L. Rice. 1976. Bull. Torrey Bot. Club
94:486-497.

Bransen, T. F. 1971. Resistance in the grass tribe Maydeae to larvae
of the western corn rootworm. Ann. Entomol. Soc. Am. 64:861-863.

Bryant, M. P., E. A. Wolin, M. J. Wolin, and R. S. Wolfe. 1967.
Menthanobaci11 us omelianskii, a symbiotic association of two
bacteria. Arch. Microbiol. 59:20-31.

Burnett, T. 1967. Aspects of the interaction between a chaleid
parasite and its Aleurodid host. Can. J. Zool. 45:539-578.

Campbell, R. W., and J. D. Podgwaite. 1971. The disease complex of
the gypsy moth. 1. Major components. J. Invertebr. Pathol.
18:101-7.

Capinera, J. L. , and P. Barbosa. 1977. Influences of natural diets
and density on gypsy moth egg mass characteristics. Can.
Entomol. 109:1313-8.

Chabora, P. C. , and D. Pimentel. 1970. Patterns of evolution in
parasite-host systems. Ann. Entomol. Soc. Am. 63:479-486.

Cornell, H. , and D. Pimentel. 1978. Switching in the parasitoid
Nasonia vitripennis and its effect on host competition. Ecology
59:297-308.

Dyer, M. I., and U. G. Bokhari. 1976. Plant-animal interactions;
Studies of the effects of grasshopper grazing on blue grama
grass. Ecology 57:762-772.

Fleming, W. E. 1963. The Japanese beetle in the United States.
USDA-, Handbook N. 236. 30 p. ill us.

Force, D. C. , and P. S. Messenger. 1964a. Fecundity, reproductive
rates and innate capacity for increase of three parasites of
Therioaphis maculata (Buckton). Ecology 45:706-715.

Force, D. C. , and P. S. Messenger. 1964b. Duration of development,
generation time, and longevity of three hymenopterous parasites
of Therioaphis maculata (Buckton) reared at various constant
temperatures. Ann. Entomol. Soc. Am. 57:405-413.

Force, D. C. , and P. S. Messenger. 1965. Laboratory studies on
competition among three parasites of the spotted alfalfa aphid
Therioaphis maculata (Buckton). Ecology 46:853-859.
-------
137
Gunn, B. H. 1974. Laboratory investigations into the dispersal of
crawlers of Dactylopius austrinus De Lotto (Homoptera:
Dactylopiidae). Unpublished report.

Hadley, C. H., and W. E. Fleming. 1952. The Japanese Beetle, pp.
567-573. In: USDA Yearbook of Agriculture, 1952.

Harper, J. L. 1977. Population Biology of Plants. Academic Press,
New York.

Hawley, I. M. 1952. Milky diseases of beetles, pp. 394-401.
In: USDA Yearbook of Agriculture, 1952.

Healy, J. B., and L. Y. Young. 1979. Anaerobic biodegradation of
eleven aromatic compounds to methane. Appl. Environ. Microbiol.
38:84-89.

Henis, Y. , A. Ghaffar, and R. Baker. 1978. Integrated control of
Rhizoctorria solani dumping-off of radish: Effect of successive
plantings, PCNB, and Trichoderma harzianum on pathogen and
disease. Phytopathology 68:900-907.

Henis, Y., A. Ghaffar, and R. Baker. 1979. Factors affecting
suppressiveness to Rhfzoctonia solani in soil. Phytopathology
69:1164-1169.

Hoffman, J. H. , and V. C. Moran. 1977. Pre-release studies on
Tucumania tapiacola Dyar (Lepidoptera: Pyralidae), a potential
biocontrol agent against jointed cactus. J. Entomol. Soc. South
Afr. 40:205-209.

Hough, J. A., and D. Pimentel. 1978. Influence of host foliage on
development, survival and fecundity of the gypsy moth. Environ.
Entomol. 7:97-102.

Kaneko, T., R. M. Atlas, and M. Krichevsky. 1977. Diversity of
bacterial populations in the Beaufort Sea. Nature 270:596-599.

King, C. E. , and P. S. Dawson. 1971. Population biology and the
Tribolium model. Evol. Biol. 5:133-227.

Kleinschuster, S. J. , B. L. Baker, and R. Baker. 1975. Responses of
crown gall tissue to gravity compensation. Phytopathology
65:931-935.

Kormanik, P. P., W. C. Bryan, and R. C. Schultz. 1977. Influence of
endomycorrhizae on growth of sweetgum seedlings from eight mother
trees. For. Sci. 23:500-506.

Kormanik, P. P., W. C. Bryan, and R. C. Schultz. 1980 (in press).
Procedures and equipment for staining large numbers of plant root
samples for endomycorrhizal assay. Can. J. Microbio.
-------
138

Marx, D. H. 1969. Antagonism of mycorrhizal fungi to root pathogenic
fungi and soil bacteria. Phytopathology 59:153-163.

Marx, D. H., and W. C. Bryan. 1975. Growth and ectomycorrhizal
development of loblolly pine seedlings in fumigated soil infested
with the fungal symbiont Pi soli thus tinctorius. For. Sci.
22:245-254.

McClure, M. S. 1979a. Spatial and seasonal distribution of hemlock
scales. Environ. Entomol. 8(5):869-873.

McClure, M. S. 1979b. Self-regulation in hemlock scale populations:
Role of food quantity and quality. Entomol. Soc. Am. Misc. Publ.
Vol. II, No. 3, 49 pp.

Mertz, D. M. 1972. The Tribolium model and the mathematics of
population growth. Ann. Rev. Ecol. Syst. 3:51-78.

Mertz, D. M., D. A. Cawthon, and T. Park. 1976. An experimental
analysis of competitive indeterminacy in Tribolium. Proc. Nat.
Acad. Sci. 73:1368-1372.

Miller, T. L., and M. J. Wolin. 1974. A serum bottle modification of
the Hungate technique for cultivating obligate anaerobes. Appl.
Microbiol. 27:985-987.

Moran, V. C., and D. P. Annecke. 1978. Critical reviews of biological
pest control in South Africa. 2. The Prickly pear, Opuntia
ficus - indica (L.) Miller. J. Entomol. Soc. South Afr.
41:161-188.

Neyman, J., T. Park, and E. L. Scott. 1956. Struggle for existence.
The Tribolium model biological and statistical aspects. In:
Proc. Berkeley Symp. Math. Statist. Probab. , 3rd ed., J. Neyman,
ed. 4:41-79. U. California Press, Berkeley.

Norris, D. M. , and J. K. Baker. 1967. Symbiosis: Effects of a
mutualistic fungus upon growth and reproduction of Xyleborus
ferrungeneus. Science 156:1120-1122.

Norris, D. M. and J. K. Baker. 1968. A minimal nutritional substrate
required by Fusarium sol am' to fulfill its mutual i stic
relationship with Xyteborus ferrugeneus. Ann. Entomol. Soc. Am.
61:1473-1475.

Norris, D. M., and H. M. Chu. 1970. Nutrition of Xyloborus ferruginous
II. A holidic diet for the asymbiotic insect. Ann. Entomol.
Soc. Am. 63:1142-1145.

Odell, T. M., and W. D. Rollinson. 1966. A technique for rearing the
gypsy moth, Portheria dispar (L) on artificial diet. J. Econ.
Entomol. 59:741-742.
-------
139

Ortman, E. E., and T. F. Branson. 1976. Growth pouches for studies of
host plant resistance to larvae of corn rootworm. J. Econ. Ent.
69:380-382.

Pimentel, P., E. H. Feinberg, P. W. Wood, and J. T. Hayes. 1968.
Selection, spatial distribution and the coexistance of competing
fly species. Am. Nat. 99:97-109.

Park, T. 1948. Experimental studies of interspecies competition. I.
Competition between populations of flour beetles, Tribolium
confusum Duval and Tribolium castaneum Herbst. Ecol. Monogr.
18:265-308.

Park, T. 1954. Experimental studies of interspecies competition. II.
Temperature, humidity and competition in two species of
Tribolium. Physiol. Zool. 27:177-238.

Rice, E. L. 1974. Allelopathy. Academic Press, New York.

Rice, E. L. 1979. Allelopathy - an update. Bot. Rev. 45:15-109.

Roper, M. M., and K. C. Marshall. 1978. Effects of clay mineral on
microbial predation and parasitism of Escherichia coli. Microb.
Ecol. 4:279-289.

Rouse, D. I., and R. Baker. 1978. Modeling and quantitative analysis
of biological control mechanisms. Phytopathology 68:1297-1302.

Ruchle, J. L., and D. H. Marx. 1977. Developing ectomycorrhizae on
containerized pine seedlings. USDA Forest Service Research Note
SE-242. Southeast Forest Experiment Station, Asheville, North
Carolina.

Sasser, J. N. , and M. F. Kirby. 1979. Crop cultivars resistent to
root-knot nematodes, Meloidogyne species. North Carolina State
University Graphics, Raleigh, North Carolina.

Schwalbe, C. P., G. M. Boush, and W. E. Burkholder. 1973a. Factors
influencing the pathogenicity and development of Mattesia
trogodermae infecting Trogoderma glabrum larvae. J. Invertebr.
Pathol. 21:176-182.

Schwalbe, C. P., G. M. Boush, and W. E. Burkholder. 1973b. Physical
and physiological characteristics of.Trogoderma glabrum infected
with the schizogregarine pathogen Mattesia trogodermae. J.
Invertebr. Pathol. 22:153-160.

Schwalbe, C. P., W. E. Burkholder, and G. M. Boush. 1974. Mattesia
trogodermae infection rates as influenced by mode of
transmission, dosage and host species. J. Stored Prod. Res.
10:161-166.
-------
140
Sokoloff, A. 1972, 1975, and 1978. The Biology of Triboll urn With
Special Emphasis on Genetic Aspects (in 3 volumes).Oxford
University Press, London.

Stakman, E. C., D. M. Stewart, and W. Q. Loegering. 1962.
Identification of physiologic races of Puccini a graminis var.
tritici. E617. USDA, Agricultural Research Service, St. Paul,
Minnesota.

Staley, J. T., K. C. Marshall, and U. B. D. Skerman. 1980. Budding and
prosthecate bacteria from freshwater habitats of various trophic
states. Microb. Ecol. 5:245-251.

Starks, K. J., and W. W. McMillan. 1967. Resistance in corn to the
corn earworm and fall armyworm. II: Types of field resistance to
the corn earworm. J. Econ. Entomol. 60:920-923.

Taylor, A. L., and J. N. Sasser. 1978. Biology, identification and
control of root-knot nematodes (Meloidogyne species). North
Carolina State University Graphics, Raleigh, North Carolina.

Wells, T. R., and R. Baker. 1969. Gravity compensation and crown gall
development. Nature 223:734-735.

Wjetunga, C., and R. Baker. 1979. Modeling of phenomena associated
with soil suppressive to Rhizoctonia solani. Phytopathology
69:1287-1293.

Windle, P. N., and E. H. Franz. 1979. The effects of insect parasitism
on plant competition: Greenbugs and barley. Ecology 60:521-529.
-------
METHODS FOR MEASURING EFFECTS OF

CHEMICALS ON AQUATIC POPULATION INTERACTIONS
March 18 and 19, 1980
Jeffrey M. Giddings

Environmental Sciences Division
Oak Ridge National Laboratory
-------
142
PARTICIPANTS

J. M. Giddings, Chairman
Oak Ridge National Laboratory
PREDATION GROUP
Charles Ashton C. R. Cripe
University of West Florida University of West Florida

G. J. Atchison Geraldine Cripe
Iowa State University U.S. Environmental Protection Agency

C. C. Coutant, Group Leader R. E. Millemann
Oak Ridge National Laboratory Oak Ridge National Laboratory

J. F. Sullivan
EG&G Idaho, Inc.
COMPETITION AND MULTISPECIES GROUP
B. G. Blaylock Stephen Hansen
Oak Ridge National Laboratory U.S. Environmental Protection
Agency

A. S. Bradshaw Andrew Kindig
Environmental Sciences Division University of Washington

J. D. Cooney Larry Klotz
University of Tennessee State University of New York

Stephen Gough F. B. Taub, Group Leader
Oak Ridge National Laboratory University of Washington
OBSERVERS
J. V. Nabholz A. S. Mammons
U.S. Environmental Protection Oak Ridge National Laboratory
Agency
-------
143

SECTION 8

METHODS FOR MEASURING EFFECTS OF CHEMICALS

ON AQUATIC POPULATION INTERACTIONS
8.1 Introduction

This workshop was convened to discuss laboratory test systems
that address interactions among populations of two or more aquatic
species. The interactions of greatest concern were predation
(especially among fish) and competition (especially among algae),
because these phenomena have received the most attention in previous
research. Simple laboratory systems containing algae, grazers, and
decomposers (referred to hereafter as "multispecies systems" or as
"microcosms") were also discussed; there is no sharp distinction
between these systems and the mixed flask cultures considered in
Section 6. Other population interactions (parasitism, grazing, and
symbiosis) have been largely neglected in environmental toxicology and
have, therefore, received little attention in this section.

The workshop objectives were: (1) to ensure that the critical
review conducted by Oak Ridge National Laboratory (ORNL) of existing
and potential methods for testing chemicals for effects on aquatic
population interactions was as complete and comprehensive as possible;
and (2) to gather information and ideas on the practical applications
of such research for hazard assessment under the Toxic Substances
Control Act (TSCA) of 1976.

8.2 Results and Discussion

8.2.1 Evaluations of Test Methods

Two working groups were formed, one to discuss predation
experiments, and the other to discuss competition experiments and
multispecies culture systems. Each group evaluated selected test
methods in terms of the criteria suggested for hazard assessment
protocols. These evaluations (Appendix A) aided in the preparation of
the ORNL review (Giddings, 1981). Summaries of the evaluations are
presented in Tables 8.1, 8.2, and 8.3. The criteria and rating scales
were as follows:

1. Replicability. How similar are the results of
replicates from any given experimental run? G = Good,
F = Fair; P = Poor; U = Unknown.

2. Reproducibility. How well can an experiment be
repeated to give the same results? G = Good; F = Fair;
P = Poor; U = Unknown.
-------
144
TABLE 8.1. EVALUATION OF PREDATION TESTS3

^
•,_
-(->

U)
c
0)
to
u
H
H
H
H
L
H
U
H
U
H
U
U
H
01
5
o-
(U
S-

• r-
•"
D
H
W
W
D
D
D
H
D
H
D
H
H
D

i ^
in
o
CJ
H
M
M
H
L
H
L
L
M
L
L
L
L
L
in (ft
-------
145
TABLE 8.2. EVALUATION OF COMPETITION TESTS3

"^QJ
U
c
O)
£-
0)
il-
CU
o:
3
13
14
18
23
24
25
26
28
30
31
12,38
46
47
»
•f—
J3
(0
U

'o.
^
p
U
F
P
G
G
F
F
F
U
U
G
U
G
.a

u
3
•a
0
S-
Q.
O)
Q£
P
U
u
u
u
u
u
u
u
u
u
u
u
u
c
o
-p
to
N

•5
IB
•o
c
m
.p

p
G
P
P
G
G
P
P
G
G
P
F
G
G
•P
•^
>
•r~
•p
•r—
l/>
C
at
to
L
H
U
L
U
H
H
U
U
H
U
H
U
U
•a
0)
i.
3
CT
0)

E
•r—
1—
M
W
M
W-M
W
D
D
W
M
D
W
W
W
W

0
0
H
L
L
M
L
L
L
L
L
L
M
L
M
L

-------
146
TABLE 8.3. EVALUATION OF MULTISPECIES CULTURE SYSTEMS3

_Q
u
0)
U
4-
0)
Q£
21,22
32
33
35
36,37
43
44,45
48
$
•—
P
m
u
•r—
'o.
O»
QC
P
U
P-F
U
P
G
G
U
•r—
-Q
U
3
•a
o
a.
ai
ce.
U
U
U
U
U
U
U
U
c
o
•r—
fO

•o
Standar
P
U
F
P
P
U
G
G

-P
•r-
>
Sensiti
L
U
U
U
U
U
H
U
•o
0)
•r—
13
cr
CD
0)
•r-
M
M
M
M
M
W-M
W-M
D

•P
U)
o
CJ
M-H
L
L
L
H
L
L
M

U) (A

It)
r—
O
t
+*
X
LU
U
U
P
U
U
F
F
P

aSee text for explanation of criteria and symbols.

bSee Section 8.4.
-------
147
3. Standardization. How readily can the procedure be
standardized for use by other laboratories? G = Good;
F = Fair; P = Poor; U = Unknown.

4. Sensitivity. Does the test system show effects at low
concentrations of chemical? H = High (more sensitive
than acute LC5o); L = Low (less sensitive than acute
LC5o); U = Unknown

5. Time required. How long does the experiment last? D =
1 to 10 d; W = m to 8 weeks; M = 2 or more months.

6. Cost. What is the cost of setting up one experimental
system? Once the system is in place, what is the cost
per chemical tested? H = High; M = Moderate; L = Low.

7. Special facilities and skills. Does the procedure
require any special equipment, techniques, or training?
Y = Yes; N = No.

8. Extrapolation. How well can the results of the
laboratory experiments be used to predict chemical
effects or other ecological phenomena in natural
ecosystems? G = Good; F = Fair; P = Poor; U = Unknown.

In general, predator-prey tests appear to be more replicable and
more readily standardized than competition tests or multispecies
culture systems (Tables 8.1, 8.2, and 8.3). Reproducibility is
virtually unknown for all systems reviewed. Predator-prey tests are
more sensitive to chemicals than acute single-species bioassays in
many cases; the sensitivity of competition tests and multispecies
cultures is largely unknown. Predator-prey tests are the most rapid,
usually lasting from less than an hour to several days, whereas
competition tests take weeks and most multispecies culture experiments
last for months. Few of the test systems in any category are rated
highly expensive, but absolute costs per test are generally not known.
All tests require some special facilities or skills (see Appendixes
for details). Extrapolation to nature is still a matter of guesswork
for most systems, although laboratory-field comparisons have been made
in a few instances. Overall, predator-prey systems are in a more
advanced stage of development as chemical effects tests than are the
other two categories. It should be remembered, however, that the
different categories of tests were evaluated by different groups of
researchers with different backgrounds and biases.

8.2.2 Group Discussion

Each working group was asked to (1) identify major issues in
multispecies toxicity testing, (2) identify problems in predicting
ecological effects of toxic chemicals from results of laboratory
tests, and (3) outline research that is needed. The two groups took
-------
148
rather different approaches. The Predation Group concentrated on
formulating guidelines for the design and analysis of predator-prey
experiments. The Competition/Multispecies System Group addressed some
of the more general problems of using complex experimental systems to
test chemicals for ecological effects. The following summaries were
adapted from the statements of Dr. Charles Coutant, who led the
Predation Group, and Dr. Frieda Taub, head of the
Competition/Multispecies System Group, that were presented during the
final session of the workshop.

(1) Predation experiments. The scope of this session included
the relatively simple experiments that have been conducted with one
predator and one prey species. The more complex systems that were
discussed by the Competition/Multispecies System Group were excluded.
Predator-prey tests are seen as a step towards ecological realism.
Predator-prey tests are often more sensitive than conventional acute
toxicity tests. Finally, predator-prey tests provide data for impact
assessment questions related to the survival of prey and the
energetics of predator populations.

These tests are not to be performed in a vacuum, but in
conjunction with conventional methods. Results must be related to
acute or chronic mortality. In a sense, predator-prey experiments are
a test of the application factor concept. Predator-prey tests might
be an alternative to chronic toxicity tests, because chronic exposure
tests are often long and expensive while some predation tests can be
completed in a short time.

Where in the testing hierarchy, then, do predator-prey tests fit?
Acute LC50 tests should be conducted first to provide background
information, particularly for predators and prey that might be used in
predation experiments. The purposes of the acute tests are: (1) to
characterize the relative sensitivity of species; (2) to look for
behavioral clues as to what effects might occur in a predator-prey
situation; (3) to examine water quality effects (temperature,
salinity, turbidity, etc.) that are difficult to include in a
predation test but that require investigation; (4) to decide which
organisms to use in a predator-prey test; and (5) to determine the
necessity for a predator-prey test for a given chemical.
Predator-prey tests might be followed by the multispecies tests
discussed in Section 8.2.2(2). Results could be used for an impact
analysis in terms of the population dynamics of the prey and the
bioenergetics of the predator. Ultimately, ecosystem studies will be
needed for verification of predicted impacts.

One category of available tests includes single-species tests
that are based on the premise that the observed responses might
influence predator-prey interactions. Such responses include:
swimming speed (burst and stamina), maneuverability, activity,
burrowing rate, reaction time, reaction distance, feeding orientation,
schooling behavior, aggressive behavior, and learning. At some point
-------
149
in the development of such tests, the effects should be correlated
with the presumed predator-prey effects. Another category of tests
includes tests with two species, many of which are described in the
literature (Appendix A). Usually the prey are stressed separately
from the predator, and the predator and prey are then placed together
to test the differential selection of the prey by the predator (e.g.,
Coutant 1973). In a few cases, simultaneous stresses are given to the
predator and the prey (e.g., Weltering et al. 1978). We are not aware
of any tests in which only predators were stressed. A third category
of test involves multiple predators and prey (e.g., Farr 1978).

(a) Design of predator-prey experiments. Criteria that should
be used in setting up the optimal system for testing predator-prey
relationships include: (1) criteria for the test organisms; (2)
criteria for the test systems; and (3) criteria for the test protocols
themselves.

The criteria for test organisms are:

1. The organisms must be readily available. Wild strains
are desirable, although cultured stocks are more
practical for routine testing. The history of cultures
should be known.

2. The organisms must survive under laboratory conditions.
Wild strains should be given time to acclimate to the
laboratory.

3. The predator should be a good feeder under experimental
conditions.

4. Short generation times are desirable for organisms
cultured in the laboratory.

5. Relatively small animals are more convenient to work
with.

6. The organisms should have ecological, economic, or
social significance.

7. The organisms should have a realistic potential for
exposure to the chemical (t,g. , aquatic insects may be
unintended targets of arthropod toxicants).

8. There should be a natural relationship between predator
and prey. The vulnerability of the prey to the
predator should be known.

9. The natural behavior of the organisms should be known.
-------
150
10. Responses to selected standard chemicals (LC5o and
chronic toxicity) should be known.

11. The levels of disease and parasitism in the test
populations should be as low as possible, unless these
factors are to be studied as part of the test (Butler
and Millemann 1971).

The criteria for test systems include:

1. The system should be matched to the natural
predator-prey interaction. The system must be
appropriate to the behavior of the organisms in terms
for example, of shelter, substrate, light regime,
relative sizes of predator and prey, and ratio of
predators to prey.

2. The test should be applicable to a wide range of
chemicals.

3. The system should be amenable to a range of
concentrations and exposure times.

4. The dosing system should be realistic. Flow-through
dosing systems are necessary for long-term tests. The
use of carriers for test chemicals should be minimized.

5. Concentrations of the test chemical and its derivatives
should be measured periodically during the test.
Experience with many chemicals indicates that
concentrations are often not constant, especially in
static systems.

6. There should be minimum contamination of the untreated
portions of the system. For instance, treated prey
should not contaminate the water and thereby cause the
predator to be exposed to the chemical. This may be
difficult to achieve in static systems.

7. The organisms should be behaviorally isolated and free
from outside noise and distraction.

8. The system should be replicable.

The criteria for test protocols include:

1. Temperature, salinity, hardness, and other experimental
conditions must be controlled. If the toxicity of a
chemical is suspected to be greatly dependent on
certain physical or chemical variables, it may be
desirable to run the predation test under a range of
-------
151
conditions. Water quality parameters should be
measured throughout the test.

2. Manipulations of the test organisms should be
minimized. Test organisms must be allowed sufficient
time to acclimate to the experimental system. Handling
stress should be equalized between predator and prey,
if possible. When it is necessary to handle one
species more than the other, there should be less
handling of the species that has been exposed to the
chemical. For example, if one is attempting to
quantify effects of the chemical on the predator, the
predator should be acclimated to the experimental
system before prey are added. This will minimize
synergistic effects of handling and the chemical.

3. Internal controls are desirable. The predator should
be allowed to feed on stressed and unstressed prey
simultaneously. This requires a differential marking
technique, which may be difficult with small organisms.

4. Separate prey controls without predators or chemicals
should be run simultaneously to correct for normal
mortality.

5. Chemical concentrations should range from a no-effect
level to one that produces clear effects. Failure to
do this has been a deficiency of many papers in the
literature (Coutant et al. 1979).

6. Effects of satiation of the predator should be
accounted for. The feeding regime must be standardized
so that the organisms are neither overfed nor starved.
This has to be tailored to the organisms used.

7. Predators and prey should not be reused in successive
tests. Learning and accumulation of the test chemical
are two ways in which reusing test animals may
influence the results (Ginetz and Larkin 1976). Tissue
residue analysis should be conducted after long
exposure tests where significant accumulation could
occur.

8. The endpoint of the test should be more sensitive than
an acute LC5oi and no less sensitive than a chronic
test, for most chemicals.

9. Standard statistics should be sufficient for data
analysis. Controls should be compared among
experiments to determine the inherent variability of
the predator-prey response.
-------
152
(b) Role of predator-prey tests in the hazard evaluation
sequence. Predator-prey tests can be sequenced to provide the most
information to the impact analyst. The relative sensitivities of the
organisms can be determined from LC50 tests. The next step would be a
predator-prey test, exposing either the prey, the predator, or both to
the chemical. The purpose of this test would be to determine if the
predator-prey interaction is likely to be affected at levels below the
Effects on survivorship and mortality of the prey can become
inputs to population dynamics models. Here, the analyst must consider
possible compensatory mechanisms operating on the prey. Some increase
in prey vulnerability might be offset by other factors without
affecting the dynamics of the population, so that the no-effect level
measured in the experiment would be lower than the true no-effect
level in the ecosystem. Predator-prey switching could be another
compensatory mechanism (Farr 1978).

Changes in the predator's feeding behavior could be incorporated
into a predator growth and energetics model. If the predator is
affected, there may also be increased survival of prey. A decision
must then be made as to which effects are of real concern. Some
predator-prey tests might be performed because of interest in the
prey; in other tests the predator may be of greater interest.

Ultimately, the validity of predator-prey test systems must be
verified in studies on streams, ponds, or other whole ecosystems.

(c) Potential difficulties. One difficulty with any laboratory
test is determining whether the laboratory derived no-effect level is
really significant in nature. For example, most test systems are
designed for one predator and one prey, but in reality predators have
a suite of prey available. Once predators eliminate one prey species,
they can simply switch to another. This leads to the problem of
predicting community dynamics. There are changing levels of prey
availability and vulnerability in any natural system, which we are not
able to simulate in any simple test.

Another difficulty in predator-prey systems is accumulation of
the chemical in the predator. If a predator is perpetually selecting
the most contaminated prey, its body burden may continually increase.
What will be the effects of the body burden? Clearly, there has been
little experience using predator-prey systems to test the effects of
organic chemicals.

(2) Competition experiments and multispecies cultures. This
group addressed the general problem of predicting ecological effects
from laboratory results. Very simple systems, with relatively few
organisms, are amenable to mechanistic understanding and are,
therefore, powerful research tools. The limitations of simple systems
stem from their incomplete connectiveness. The importance of
-------
153
connect!veness in determining the effects of disturbance on an
ecosystem, especially a simple system, is illustrated by a
hypothetical example from a recent paper by May et al. (1979). In
this example, harvesting of one species results in the extinction of a
competing species as a consequence of competitive interactions at a
lower trophic level. The situation is unique because the result
depends on the hypothesized feeding and competitive interactions of
the four species involved. If a natural system were controlled by
this kind of connectiveness and one tested a chemical on a laboratory
system with a different connectiveness, then it would be impossible to
extrapolate from the laboratory to nature. It was suggested that
natural systems may have a multitude of possible connective
relationships and that natural systems may be more like one another
than are simplified systems. This problem, although an important one
in the long run, is not easy to deal with.

(a) Complexity. Comparisons of simple and complex laboratory
systems with natural systems are major research priorities. If
simpler systems are amenable to mechanistic understanding, clear-cut
demonstrations of this fact are needed. For example, there are only a
few examples in the literature of reversals of competitive dominance
due to selective toxicants (Fielding and Russell 1976; Fisher et al.
1974). If these simple systems are sensitive to chemicals, data are
needed to prove it.

Ecological complexity in laboratory systems might be achieved in
two ways. Synthetic (or gnotobiotic) systems can be made more complex
by adding more species. Other systems may be complex by virtue of
being naturally-derived. A difficulty with naturally-derived systems,
at least with many of those that have been used in the past, is that
one cannot always analyze or document the connective relationships
within the systems. Each of these approaches (simple and complex) has
its advantages as well as its disadvantages. Simpler systems are
easier to analyze; complex systems may be more realistic.

(b) Sensitivity. The relative sensitivities of different
laboratory systems is another important issue for research in the
near-term. (The relative sensitivity of laboratory systems vs.
natural environments is the ultimate question, but not one that we are
ready to approach yet.) It is important to recognize that sensitivity
is partially a function of the experimental conditions; this must be
taken into account when comparisons are made among different
laboratory systems. Moreover, there are a variety of parameters that
can be measured in a multispecies system. Many researchers measure
photosynthesis and respiration and examine production/respiration
ratios; others enumerate populations; others measure nutrient uptake
rates. The sensitivity of the system will be a function of which
parameters are selected for measurement.

(c) Representativeness. The results of a multispecies test are,
to some extent, specific to the organisms in the system and to the
-------
154

experimental conditions. The problem of selecting test organisms that
are representative of natural ecosystems affects the design of multi-
species systems just as it affects the design of single-species tests.
Likewise, one must choose a temperature, pH, hardness, etc., that are
typical of the natural environment of interest. Unless a laboratory
system has been tested over a range of experimental conditions, one
does not know how situation-specific the results of any particular
test may be. The effects of system design and experimental conditions
should be studied by each researcher for his or her own system.

(d) Chemical exposure. Another important issue is the
possibility of transformation of a chemical in the test system. For
example, the state of the chemical may change with Eh, pH, or oxygen.
Test chemicals may also be transformed by organisms within the system.
This is one instance in which multispecies systems can be more
realistic than single-species tests. For example, a chemical that is
determined to be quite toxic in a single-species test may be readily
degraded by other organisms and produce virtually no effects in a
natural ecosystem or a multispecies system.

The exposure of organisms in a multispecies system to a test
chemical can be influenced by the presence of other organisms and
abiotic components. Sediments may absorb the chemical and reduce the
exposure to the biota. Herbivores and predators may be exposed to the
chemical through the food chain. Interactions like this between the
chemical and the various components of the test system contribute to
the greater realism of multispecies systems as compared to
single-species tests.

(e) Replicability and reproducibility. According to Dr. Taub,
the ability to replicate multispecies systems within an experiment is
fairly good. Occasionally several replicates in a group will develop
differently from the rest and will be omitted from an experiment.
Initially, the systems undergo fluctuations with large amplitudes,
occurring synchronously in all replicates. Later in an experiment,
the amplitudes decrease but replicates become asynchronous, resulting
in high apparent variability among replicates. This is especially
true of species enumerations; most chemical factors (e.g., phosphate
and pH) tend to be fairly consistent.

When streptomycin was tested in two separate experiments, many of
the effects were reproduced, though not necessarily on the same day in
the two tests. For example, Anabaena was reduced by the streptomycin;
this effect was significant from day 7 to .day 28 in one experiment,
and from day 11 to day 25 in the other experiment.

(f) Other issues. Two other subjects discussed briefly were
interactions among chemicals in ecosystems receiving pollution from
several sources and problems associated with carrier solvents
necessary for testing certain types of chemicals. The carrier problem
may be especially serious in systems containing decomposers; some
-------
155
observed effects may be caused by utilization of the carrier as a
carbon source by the microorganisms.

(g) Role of multispeci'es tests in the hazard assessment
sequence. Multispecies tests are considered most useful in the
intermediate stages of hazard assessment. They should be preceded by
acute toxicity tests with single species. Multispecies tests might be
supplements or alternatives to some single-species chronic toxicity
tests, because the former are less expensive and easier to perform.
If chemicals are expected to be transformed in the environment, a
multispecies test should precede extensive testing of the
transformation products. If a chemical is not highly toxic in acute
screening tests, but produces unexpected mortalities in a microcosm,
then the formation of toxic transformation products would be suspected
and should be investigated. Certainly, multispecies tests should
precede impact analysis. They would, therefore, be used in
approximately the same position as predator-prey tests.

8.3 Conclusions

Perhaps the major problem to be resolved before predation,
competition, and other population interaction tests can be
incorporated into a hazard assessment process is extrapolation or
generalization of experimental results to predict effects in natural
ecosystems. To a certain extent this extrapolation can be facilitated
by selecting organisms and experimental conditions that are
representative of the natural systems of interest. However, the
system-specific interrelationships among populations in nature are too
complex to reproduce in their entirety in the laboratory. The degree
to which the necessary simplicity of laboratory test systems may
distort chemical effects on population interactions is not yet known.
There is a need for research to (1) compare the sensitivity of
simplified laboratory systems to that of natural ecosystems, (2)
relate the ecological complexity of laboratory systems (number of taxa
or number of functional groups) to their responses to chemicals, and
(3) develop models or other analytical approaches to linking
laboratory results to predictions about chemical effects in nature.

A hazard assessment sequence should begin with single-species
toxicity screening tests before multispecies tests are undertaken.
The screening tests are necessary to identify chemicals that are
likely to produce ecological effects, to determine the relative
sensitivities of different organisms, and to aid in the interpretation
of multispecies test results.

The effects of a chemical in an ecosystem depend in part on
whether the chemical is degraded, transformed, sorbed by inorganic or
organic substrates, bioaccumulated, etc. An ecologically realistic
hazard assessment is possible only in conjunction with a realistic
exposure assessment. Fate and effects of chemicals are inextricably
linked.
-------
156
Laboratory systems involving predation, competition, and multiple
population interactions are available for development as hazard
assessment test protocols. Few of these systems have been used for
chemical testing, however, and further experimentation is needed on
all types of tests before any particular technique can be selected for
immediate use.
-------
157

8.4 References

1. Boyce, N. P., and S. B. Yamada. 1977. Effects of a parasite,
Eubothrium salve!ini (Cestoda: Pseudophyllidea), on the resistance
of juvenile sockeye salmon, Oncorhynchus nerka, to zinc. J. Fish.
Res. Board Can. 34:706-709.

2. Butler, J. A., and R. E. Millemann. 1971. Effect of the "salmon
poisoning" trematode, Nanophyetus salmincola, on the swimming
ability of juvenile salmonid fishes. J. Parasitol. 57:860-865.

3. Confer, J. L. 1972. Interrelations among plankton, attached
algae, and the phosphorus cycle in open artificial systems. Ecol.
Monogr. 42:1-23.

4. Confer, J. L., and P. I. Blades. 1975. Reaction distance to zoo-
plankton by Lepomis gibbosus. Verh. int. Verein. Limnol. 19:
2493-2497.

5. Confer, J. L., G. L. Howick, M. H. Corzette, S. L. Kramer,
S. Fitzgibbon, and R. Landesberg. 1978. Visual predation by
planktivores. Oikos 31:27-37.

6. Coutant, C. C. 1973. Effect of thermal shock on vulnerability
of juvenile salmonids to predation. J. Fish. Res. Board Can.
30:965-973.

7. Coutant, C. C., H. M. Ducharme, Jr., and J. R. Fisher. 1974.
Effects of cold shock on vulnerability of juvenile channel catfish
(Ictalurus punctatus) and largemouth bass (Micropterus salmoides)
to predation. J. Fish. Res. Board Can. 31:351-354.

8. Coutant, C. C., R. B. McLean, and D. L. DeAngelis. 1979. Influences
of physical and chemical alterations on predator-prey interactions.
IN H. Clepper (ed.), Predator-Prey Systems in Fisheries Management,
Washington, D.C.: Sport Fishing Inst. pp. 57-68.

9. Draggan, S. J. 1977. Interactive effect of chromium compounds
and a fungal parasite on carp eggs. Bull. Environ. Contam. Toxicol.
17:653-659.

10. Farr, J. A. 1977. Impairment of antipredator behavior in
Palaemonetes pugio by exposure to sublethal doses of Parathion.
Trans. Am. Fish. Soc. 106:287-290.

11. Farr, J. A. 1978. The effect of methyl parathion on predator
choice of two estuarine prey species. Trans. Am. Fish. Soc.
107:87-91.

12. Fielding, A. H., and G. Russell. 1976. The effect of copper on
competition between marine algae. J. Ecol. 64:871-876.
-------
158
13. Fisher, N. S., E. J. Carpenter, C. C. Remsen, and C. F. Wurster.
1974. Effects of PCB on interspecific competition in natural and
gnotobiotic phytoplankton communities in continuous and batch
cultures. Microb. Ecol. 1:39-50.

14. Frank, P. W. 1957. Coactions in laboratory populations of two
species of Daphnia. Ecol. 38:510-519.

15. Giddings, J. M. 1981. Chemical effects on aquatic population
interactions and ecosystem properties. IN A. S. Mammons (ed.),
Methods for Ecological Toxicology: A Critical Review of Laboratory
Multispecies Tests. ORNL-5708; EPA 560/11-80-026, Oak Ridge
National Laboratory, Oak Ridge, Tennessee.

16. Ginetz, R. M., and P. A. Larkin. 1975. Factors affecting rainbow
trout (Salmo gairdneri) on migrant fry of sockeye salmon (Onco-
rhynchus nerkaTiJ. Fish. Res. Board Can. 33:19-24.

17. Goodyear, C. P. 1972. A simple technique for detecting effects
of toxicants or other stresses on a predator-prey interaction.
Trans. Am. Fish. Soc. 101:367-370.

18. Goulden, C. E., and L. L. Hornig. 1980. Population oscillations
and energy reserves in planktonic Cladocera and their consequences
to competition. Proc. Nat. Acad. Sci. (in press).

19. Hatfield, C. T., and J. M. Anderson. 1972. Effects of two insectv
cides on the vulnerability of Atlantic salmon (Salmo salar) prey
to brook trout (Salvelinus fontinalis) predation. J. Fish. Res.
Board Can. 29:27-29.

20. Kania, H. J., and J. O'Hara. 1974. Behavioral alterations in a
simple predator-prey system due to sublethal exposure to mercury.
Trans. Am. Fish. Soc. 103:134-136.

21. Kersting, K. 1975. The use of microsystems for the evaluation of
the effects of toxicants. Hydrobiol. Bull. 9:102-108.

22. Kersting, K. 1978. Experiments with dichlobenil in a micro-
ecosystem. Proc. EWRS 5th Symp. on Aquatic Weeds.

23. Kindig, A. 1979. Investigations for streptomycin-induced algal
competitive dominance reversals. Experimental Rept. ME25,
FDA Contract No. 223-76-8348, University of Washington.

24. Klotz, R. L., J. R. Cain, and F. R. Trainor. 1976. Algal
competition in an epilithic flora. J. Phycol. 12:363-368.

25. Kricher, J. C., C. L. Bayer, and D. A. Martin. 1979. Effects of
two Aroclor fractions on the productivity and diversity of algae
from two lentic ecosystems. Int. J. Environ. Stud. 13:159-167.
-------
159

26. Lange, W. 1974. Competitive exclusion among three planktom'c
blue-green algal species. J. Phycol. 10:411-414.

27. Li, J. L., and H. W. Li. 1979. Species-specific factors affecting
predator-prey interactions of the copepod Acanthocyclops vernal is
with its natural prey. Limnol. Oceanogr. 24:613-626.

28. Marshall, J. S. 1969. Competition between Daphnia pulex and £.
magna as modified by radiation stress. Ecol. Soc. Amer. Annual
Meeting, University of Vermont, August 17-22, 1969.

29. May, R. M., J. R. Beddington, C. W. Clark, S. J. Holt, and R. M.
Laws. 1979. Management of multispecies fisheries. Science 205:
267-276.

30. Mosser, J. L., N. S. Fisher, and C. F. Wurster. 1972. Poly-
chlorinated biphenyls and DDT alter species composition in mixed
cultures of algae. Science 176:533-535.

31. Muller, W. A., and J. J. Lee. 1977. Biological interactions and
the realized niche of Euplotes vannus from the salt marsh aufwuchs.
J. Protozool. 24:523-527.

32. Neill, W. E. 1975. Experimental studies of microcrustacean com-
petition, community composition, and efficiency of resource
utilization. Ecol. 56:809-826.

33. Nixon, S. W. 1969. A synthetic microcosm. Limnol. Oceanogr.
14:142-145.

34. Pippy, 0. H. C., and G. M. Hare. 1969. Relationship of river
pollution to bacterial infection in salmon (Salmo salar) and
suckers (Catastomus commersoni). Trans. Am. Fish. Soc. 98:685-690.

35. Reed, C. C. 1976. Species diversity in aquatic microecosysterns.
Ph.D. dissertation, U. Northern Colorado.

36. Ringelberg, J. 1977. Properties of an aquatic microecosystern.
II. Steady-state phenomena in the autotrophic subsystem. Helgol.
Wiss. Meersunters. 30:134-143.

37. Ringelberg, J., and K. Kersting. 1978. Properties of an aquatic
microecosystern. I. General introduction to the prototypes. Arch.
Hydrobiol. 83:47-68.

38. Russell, G., and A. H. Fielding. 1974. The competitive properties
of marine algae in culture. J. Ecol. 62:689-698.
-------
160
39. Sullivan, J. F., and G. J. Atchison. 1978. Predator-prey behavior
of fathead minnows, Pimepnales promelas. and largemouth bass,
Micropterus salmoides, in a model ecosystem. J. Fish. Biol.
13:249-253.

40. Sylvester, J. R. 1972. Effect of thermal stress on predator
avoidance in sockeye salmon. J. Fish. Res. Board Can. 29:601-603.

41. Sylvester, J. R. 1973. Effect of light and heat stress on vulner-
ability of sockeye salmon to predation by coho salmon. Trans. Am.
Fish. Soc. 102:139-142.

42. Tagatz, M. W. 1976. Effect of Mi rex on predator-prey interactions
in an experimental estuarine ecosystem. Trans. Am. Fish. Soc.
105:546-549.

43. Taub, F. B. 1969. A biological model of a freshwater community: A
gnotobiotic ecosystem. Limnol. Oceanogr. 14:136-142.

44. Taub, F. B., and M. E. Crow. 1978. Loss of a critical species in
a model (laboratory) ecosystem. Verh. int. Verein. Limnol. 20:
1270-1276.

45. Taub, F. B., and M. E. Crow. 1980. Synthesizing aquatic microcosms.
IN J. P. Giesy (ed.), Microcosms in Ecological Research (in press).

46. Tilman, D. 1977. Resource competition between planktonic algae:
An experimental and theoretical approach. Ecol. 58:338-348.

47. Titman, D. 1976. Ecological competition between algae: Experimental
confirmation of resource-based competition theory. Science 192:
463-465.

48. Tsuchiya, H. M., J. F. Drake, J. L. Jost, and A. G. Frederickson.
1972. Predator-prey interactions of Dictyostelium discoideum and
Escherichia coli in continuous culture. J. Bacteriol. 110:1147-1153.

49. Vinyard, G. L., and W. J. O'Brien. 1975. Dorsal light response as
an index of prey preference in bluegill (Lepomis macrochirus).
J. Fish. Res. Board Can. 32:1860-1863.

50. Vinyard, G. L., and W. J. O'Brien. 1976. Effects of light and
turbidity on the reactive distance of bluegill (Lepomis macrochirus).
J. Fish. Res. Board Can. 33:2845-2849.

51. Weltering, D. M., J. L. Hedtke, and L. J. Weber. 1978. Predator-
prey interactions of fishes under the influence of ammonia. Trans.
Am. Fish. Soc. 107:500-504.
-------
161
APPENDIX A

EVALUATIONS OF SELECTED TESTS FOR

EFFECTS ON AQUATIC POPULATION INTERACTIONS
-------
163

APPENDIX A

A.I Predation Tests

A.1.1 Butler and Millemann 1971

Effect of parasites on swimming ability of salmonids. Single-
species of test with implications for predation.

Replicability: Very good. Standardization: Very good. Sensi-
tivity: Unknown. Time required: One to several days. Cost: $6000
to $10,000 (set up); $200 to $400 per chemical tested. Special
facilities: Swimming tube. Extrapolation: Very good if the host-
parasite system is natural. Comments: This approach determines the
effects of a dual stress, one of which is natural (parasite or
disease) and the other of human origin (a chemical) or an animal
(vertebrate or invertebrate). The results can show synergism,
antagonism, or additive effects between two stresses, e.g., activation
of a latent infection by the chemical (Pippy and Hare 1969); increased
susceptibility to the chemical by an existing infection (Boyce and
Yamada 1977); or protection against the parasite by low concentrations
of the chemical (Draggan 1977). In the case of swimming ability
tests, the criterion of effect is impairment of swimming ability with
susceptibility to predation enhanced.

A.1.2 Coutant 1973; Coutant et al. 1974, 1979

Effect of temperature stress on susceptibility of fish to un-
stressed predators.

Replicability: Good. Standardization: Good; has been used for
both heat shock and cold shock with some variation. Sensitivity:
Better than acute LC50. Time required: For chemicals, time will be
determined by predator satiation; several weeks to train predators to
feed in laboratory; actual test takes minutes or hours. Cost: Mainly
for tank set up, which will depend on predator and prey used. Cost
per chemical depends on manpower and organism rearing. Special
facilities: Holding and exposure tanks for fish. Special
skills: Fish care. Extrapolation: Good for cases where only prey
are likely to be exposed to chemical; artificial when both predator
and prey are likely to be exposed together. Comments: This method
has certain disadvantages: (1) low dosages can cause stimulation and
thus enhanced survivorship of prey; and (2) some behaviors are protec-
tive; stressed fish sink out of sight and escape predation. Either of
these phenomena can confuse the interpretation of results.

A.1.3 Farr 1977

Effects of chemicals on susceptibility of grass shrimp to kiHi-
fi sh predation.
-------
164
Replicability: Good. 18 to 20 replicates per concentration per
compound. Standardization: Probably not high because of time.
Sensitivity: High because of high arthropod sensitivity to organo-
phosphorus compounds used in test. Time required: High because of
number of replicates needed. Cost: Moderately high cost per chemical
because of time required. Extrapolation: No shelter was provided for
the prey. Comments: The same predators were used throughout the
test. With many compounds that accumulate this could cause a problem.
By dumping the shrimp into the tank the possible added toxicant stress
may produce a synergism resulting in behavior that could not be
properly eliminated with the control. The necessary number of repli-
cates to avoid the predator affecting the outcome increases the time
and cost factor. Using marked prey (treated and control) in the same
tank might improve this. The effects of the toxicant on the fish
cannot be determined. Farr makes a good point that this test may have
limited significance with stenophagic predators—they will eat the
shrimp they need whether they have been affected by dosing or not.
His follow-up, two-prey study addressed the implications with oppor-
tunistic feeders (see Farr 1978).

A.1.4 Farr 1978

Effects of chemicals on predator choice between two prey species.

Replicability: Good. Standardization: Easy. Sensitivity:
Better than acute LC50. Time required: Large amount of time and
effort. Cost: High. Special facilities: Enough space for 26
55-gal. tanks. Extrapolation: Demonstrates predator switching to
behaviorally impaired prey. Good correlation to natural systems.
Comments: Very good test to show predator switching. Demonstrated
sublethal effects on most sensitive prey. Too much work involved for
a standardized test. Fish-fish predation tests are interesting, but
both predator and prey may be affected by toxicants at the same con-
centration. Freshwater predation tests should be developed with the
more sensitive crustaceans as prey.

A.1.5 Goodyear 1972

Effects of stress on susceptibility of fish to predation; refuge
provided for prey.

Replicability: Good. Standardization: Good. Sensitivity:
Good. Time required: 7~d acclimation; 10~d test. Cost: Low.
Special facilities: Holding and exposure tanks. Special
skills: Fish care. Extrapolation: The absolute refuge concept cannot
be extrapolated to a natural environment. Once off the shelf, prey
were consumed very quickly. Comments: Mosquito fish are easily
obtained and reared. Largemouth bass are easily obtained and trained
as predators. Test geared specifically to strong shelter-seeking
behavior of the prey. The basic concept is sound but this study dealt
with only quantitative data (i.e., percent survival); alterations in
-------
165
behavior responsible for the increased prey vulnerability were not
dealt with. Quick, inexpensive. Intraspecific interactions between
prey organisms were not discussed; for example, does the treatment
and/or crowding cause the prey to leave the refuge?

A.1.6 Hatfield and Anderson 1972

Effect of chemicals on susceptibility of fish to predation.

Replicability: Possible. Standardization: Good. Sensitivity:
Effects at same level as LC50- Time required: 24-h exposure; 24-h
clean holding water; 24-h predation test. Cost: Higher than
Goodyear1s because of size. Extrapolation: (1) Natural light regime
used. (2) Natural temperature regime for time of test. (3) The use
of the same predator throughout the series of tests raises the
question of how "natural" the response of the predator becomes after
more than one trial. (4) In the field, the predator may not often be
completely prohibited from pursuing the prey. (5) Prey were
frequently caught by being forced against the side of the pond.
Comments: Same concept as Goodyear1s, but the system does not offer
any advantages and the cost is higher. Behavioral alterations are not
well discussed. The test with the compound Sumithion® did not indi-
cate a more sensitive effect than the 96-h LC50. However, it still
may be of value in that those fish left at the LC50 concentration will
themselves be more susceptible to predation than clean fish. Note:
Although there is only a weak correlation, a swimming stamina test
with the same compound indicated a 35% decrease in critical swimming
speed of the brook trout at a similar concentration (0.5 ppm). The
comparison of these two types of tests may be of some value.

A.1.7 Kania and O'Hara 1974

Effect of chemicals on susceptibility of fish to predation;
refuge for prey.

Replicability: Possible. Standardization: Good. Sensitivity:
Increased prey vulnerability at sublethal exposure concentrations.
Time required: 24-h treatment to prey; 24-h acclimation; 60-h test.
Cost: Low. Extrapolation: see Goodyear 1972.

A.1.8 Li and Li 1979 (and others dealing with zooplankton predation)

Survival of zooplankton prey in the presence of a zooplankton
predator.

Replicability: Unknown. Standardization: Should be possible.
Sensitivity: Unknown, but if species are differentially sensitive to
a toxicant then this system may be sensitive to effects. Time
required: Short because interactions are rapid. Cost: Very inexpen-
sive. Special skills: Species identification. Extrapolation:
Unclear. Comments: An area that deserves work but for which little
-------
166
background work has been completed. Perhaps a late level screening
tool as opposed to a quick first level screen.

A.1.9 Sullivan and Atchison 1978

Effect of chemicals on susceptibility of fish to predation.

Replicability: Good. Standardization: Good. Sensitivity:
Exposure levels investigated down to the "no effect" level. Time
required: 48-h exposure (acute) or 21-d exposure (chronic), 48-h
acclimation, 168-h test. Cost: Initial set up low, labor high
because of behavioral observations made. Comments: Statistics used
to analyze the data seem to accurately deal with the probability that
the next fish eaten will be a treated fish, given the absolute density
at the time. Much more appropriate than the standard Chi-square for
this type of data.

A.1.10 Sylvester 1972, 1973

Effect of stress on susceptibility of fish to predation.

Replicability: Good. Standardization: Good. Time required:
Short acclimation, exposure less than 1 min. Cost: Very low.
Comments: Predators were preacclimated while prey were not.
Predators were starved for 7 d before testing—this is unrealistic.
No cover was provided for prey—also unrealistic. Survival time of
prey was measured in seconds; this does not reflect behavioral effects
of the thermal stress on the prey, but stress from initially being
dumped into the tank.

A.1.11 Tagatz 1976

Effect of chemicals on susceptibility of grass shrimp to
predation.

Replicability: Fair. Standardization: Good potential except
for possible variability of plant survival. Perhaps artificial plants
would be better. Sensitivity: Based on an invertebrate, which
probably had a lower LC50 than the fish. This compound (Mirex®) often
produces delayed mortality necessitating a longer testing time.
Time required: Moderate amount of setup time because of necessary
replication. Cost: Relatively low setup and cost per chemical.

Extrapolation: Good in terms of availability of toxicant to prey
through plants, water, sediment—analyses of all these components were
performed. See comments. Comments: The test was designed to observe
the effects on the prey and not on the predator. Because no chronic
or acute data were included on grass shrimp sensitivity, it is diffi-
cult to determine the overall sensitivity compared with a standard
test, except that essentially sublethal levels were tested.
-------
167
A.1.12 Vinyard and O'Brien 1975

"Tilt box" to measure predator's interest in potential prey item.

Replicability: High. Standardization: High. Sensitivity:
Unknown. Time required: Seconds. Cost: Relatively inexpensive.
Extrapolation: Unclear. A very simple system that may be useful for
early screening.

A.1.13 Vinyard and O'Brien 1976; Confer et al. 1978; Confer and
Blades 1975

Measurement of reactive distance in fish (distance between prey
and predator at the point where predator first orients towards prey).

Replicability: Results to date indicate quite good replicability
if controlled for prey size and predator size. Standardization:
Easy. Sensitivity: Unknown. Time required: Relatively rapid.
Cost: Low. Extrapolation: Unclear, although reactive distance
controls search volume, which is clearly of significance under low
prey densities. Comments: Although this test has not been used in
toxicity testing, it has been used with light intensity and turbidity
as experimental variables. Atchison is currently beginning to examine
the impact of copper on reactive distance of bluegill to mosquito
larvae.
A.1.14 Weltering et al. 1978
Effect of chemicals on fish predator-prey system with both
predator and prey exposed.

Replicability: Good. Standardization: Good. Sensitivity:
Significant impacts at sublethal levels of ammonia. Time required:
10-d exposure. Extrapolation: See comments. Comments: The feature
of simultaneously and continuously exposing both the predator and prey
makes sense from an ecological perspective. In this experiment, the
bass was actually more sensitive than the Gambusia, and a density
dependent impact was also assessed. This approach likely makes more
sense than exposing only the prey, especially in the case of toxic
compounds rather than thermal stress.

A.2 Competition Tests

A.2.1 Confer 1972

Competition between phytoplankton and attached algae in con-
tinuous flow aquaria.

Replicability: Poor. Reproducibility: Low. Standardization:
Very low; standard species not used. Sensitivity: Low. Time
required: 5 to 9 months. Cost: Fairly high setup cost; extremely
-------
168
high cost per chemical because of length of experiment, number of
parameters measured. Special facilities: Greenhouse or other large
area with plumbing. Special skills: Radiotracer technique, taxonomic
expertise. Extrapolation: Examines relation of attached algae and
open water phosphorus concentration to phytoplankton. Similar to
shallow ecosystems with heavy bottom growth. Tracer studies could
indicate alteration of energy flow because of toxic stress. Extrapo-
lation depends on similarity to natural system (e.g., surface to
volume ratio).

A.2.2 Fisher et al. 1974

Effect of chemical on algal competition in two-species batch and
continuous cultures.

Replicability: Not examined (no replicates run); should be
fairly good. Standardization: High. Sensitivity: Good—competition
effects in continuous culture observed with PCB at 0.1 ppb (close to
ambient in some rivers). Time required: 2-3 weeks. Cost: Low.
Special facilities: Constant temperature area; microscope. Special
skills: Algal culture. Extrapolation: No data; systems very
simplified.

A.2.3 Frank 1957

Competition between two Daphnia species.

Replicability: Began with 8 to 14 replicates; contamination
problem over 18-month experiment. Standardization: Complicated.
Time required: 70 d. Cost: Low, comparatively. Extrapolation:
Species naturally occur in similar habitats but seemingly do not occur
together; extrapolation "only at considerable risk of proving wrong,"
according to the author. Comments: Evaluating the effects of a toxic
substance on a multispecies test system seems to be needlessly compli-
cated by introducing the factor of competition between two closely
related (congeneric) species.

A.2.4 Goulden and Hornig 1980

Competition between two zooplankton species.

Replicability: Low; static system varied with change of media;
low food every 2 d, high food every 4 d. Standardization: Low;
oscillations between replicates very high with different conclusion
drawn as to best competitor over time. Sensitivity: Low? Time
required: 1 to 3 months. Cost: Moderate (mainly personnel costs).
Special skills: Familiarity with cladoceran life history. Extrapola-
tion: Difficult to assess because variability is high; however, the
energy reserves may be important when the food (algae) are diminished
either by grazing or toxicants. Comments: Variation between repli-
cates was large—population densities of Daphnia were twice as high in
one sample as in a replicate.
-------
169
A.2.5 Kindig 1979

Algal competition in batch culture.

Replicability: Sufficient to show significant effects with
ANOVA. Standardization: Good. Sensitivity: Unknown. Time
required: Test runs 7 weeks, but effects observable within 2 weeks.
Cost: Very low. Special facilities: Constant temperature room.
Special skills: Algal culture. Extrapolation: Unknown (see
comments). Comments: (1) Results verified predictions from single
species assays to the resistance of Scenedesmus. (2) Test indicated
competitive ability of Scenedesmus relative to other species tested
and demonstrated competitive reversal. (3) Results from this experi-
ment predicted Scenedesmus increases in complex microcosms with strep-
tomycin treatments correctly. It did not indicate that dominance
would occur late in microcosm development (7 weeks). Extrapolation to
more complex (i.e., natural) systems is open to speculation. Probably
could only be used with extreme caution.

A.2.6 Klotz et al. 1976

Algal competition in batch cultures.

Replicability: Good. Standardi zati on: Good. Sensitivity:
Lowest concentration of sewage tested (20%) gave results. Time
required: 5 to 7 d. Cost: Low. Special facilities: Side-arm
shaker. Extrapolation: Field biomass data verified laboratory
findings. Other rivers receiving municipal sewage in the northeast
have similar situations with Chlorella occurring in the effluent
plume, diatoms outside the plume. Comments: (1) It is necessary to
shake the cultures so the diatom does not attach to the side of the
flask. (2) The diatom must be placed in a blender to disperse clumps
for accurate cell counts.

A.2.7 Kricher et al. 1979

Effects of chemicals on productivity and diversity in natural
algal communities.

Replicability: Fair. Standardization: Poor (inoculum from
natural ecosystem). Sensitivity: Good; 1 ppm significantly reduced
carbon fixation and decreased the total number of organisms. Time
required: 24 h. Cost: Low. Special facilities: 14C. Special"
skills: Algal taxonomy. Extrapolation: Possible.

A. 2.8 Lange 1974

Algal competition in batch cultures.

Replicability: Fair. Time required: 15 d or less. Cost: Low.
Extrapolation: Possible. Comments: Quantification of blue-green
-------
170
filaments is difficult. The accuracy with which this is done may
limit usefulness.

A.2.9 Marshall 1969

Competition among zooplankton species.

Replicability: Fair. Standardization: Good. Sensitivity:
Unknown for chemicals. Time required: 100 weeks. Cost: Low.
Extrapolation: Possible. Comments: System would need to be tested
with a chemical before consideration as a test system.

A.2.10 Mosser et al. 1972.

Effect of chemicals on algal competition in batch cultures.

Replicability: Unknown. Standardization: Should be good.
Sensitivity: Better than single-species tests. Time required: 4 d.
Cost: Low. Special facilities: Coulter counter. Special skills:
Algal culture. Extrapolation: Unknown; simple systems may vary with
grazing and sediment.

A.2.11 Muller and Lee 1977

Competition among a ciliate, a foramaniferan, and a nematode (all
herbivores).

Replicability: Unknown. Standardization: The organisms' habitat
seems too ephemeral to promote standardization. Sensitivity:
Demonstrated for nutrient quantity. Time required: 42 d. Cost:
Moderate? Special skills: Sophisticated culturing techniques.
Extrapolation: Authors emphasize they have limited their approach to
nutrient factor and population growth results.

A.2.12 Russell and Fielding 1974; Fielding and Russell 1976

Algal competition in batch cultures.

Replicability: Good. Standardization: Moderate. Sensitivity:
More sensitive to copper in dual culture than unialgal; changes in
competition detected from 10 to 500 ppb Cu. Time required: 35 d.
Cost: Low. Special facilities: Constant temperature room.
Extrapolation: Was not done, but suggest it would be possible in case
of accidental industrial spillage. Comments: (1) Quantification of
growth of filaments (as used here) is more difficult than that of uni-
cellular algae. (2) This method compares the performance of a species
against one competitor with the performance of the species against
another competitor. By this triangular method you obtain more infor-
mation than by just studying performance of species alone vs. per-
formance against one competitor.
-------
171
A.2.13 Tilman 1977

Effect of nutrient regimes on algal competition in semi-
continuous cultures.

Replicability: Unknown. Standardization: Good. Sensitivity:
Unknown. Time required: 30-40 d. Cost: Moderate. Special skills:
Bacteria-free algal culture. Extrapolation: Apparently good. Data
inserted into Monod model accounted for 70% of the variance of two
species along a natural silicate/phosphate gradient in Lake Michigan
(even though the cells used were cloned from a different lake).
Comments: The fact that the clones used in the test, even though not
from Lake Michigan, accounted for much of the variability in the
abundance of those species is indicative of the extrapolative appeal
of this approach. It remains to be demonstrated whether changes in
resource utilization with toxicant exposure will extrapolate as well.

A.2.14 Titman 1976

Effects of nutrient regimes on algal competition in continuous
cultures.

Replicability: Appears good. Standardization: High. Sensi-
tivity: Unknown. Time required: 3 weeks. Cost: Minimal. Special
facilities: Temperature control area. Special skills: Algal
culture. Extrapolation: Resource utilization analysis appears to
predict boundaries of coexistence quite well. Intuitively, it seems
that imposed stresses should alter the competitive balance between
species and become evident with this testing approach. Comments:
Establishment of new nutrient gradient boundaries allowing coexistence
of species in the presence of a toxicant could be valuable in pre-
dicting the outcome in natural systems where nutrient concentrations
are known.

A.3 Algae-Grazer-Decomposer Systems

A.3.1 Harrass and Taub (FDA Contract #223-76-8348)

Mixed culture; algae, grazers, protozoa, fish.

Replicability: Tested; good. Standardization: Fair; used
filtered lake water, enriched with nutrients, inoculated with
organisms—other lake waters may be different. Sensitivity: Yes, at
low doses; high doses were more similar to control. Time required: 1
to 3 months. Cost: $10 per replicate, not counting cost of main-
taining stock cultures. Special skills: 14C, chlorophyll extrac-
tion. Extrapolation: Would yield bioaccumulation and consequences;
species diversity shifts; may not predict which species in a natural
community would predominate.
-------
172
A.3.2 Kersting 1975, 1978

Compartmentalized autotroph, grazer, decomposer system.

Replicabili'ty: Appears poor. Standardization: Poor because of
complexity. Sensitivity: Effects observed appear identical to those
achieved with single species assays, or even less sensitive in some
trials. Time required: 8 to 9 weeks. Cost: Initial cost high;
moderate thereafter. Special skills: Algal culturing. Extrapola-
tion: see Ringelberg 1977, 1978. Comments: This system appears less
vulnerable to species invasion than the system detailed by Ringelberg
but it is still highly complex.

A.3.3 Kindig (FDA Contract #223-76-8348)

One or two algal species grazed by zooplankton.

Replicability: High. Standardization: Good. Sensitivity:
Unknown—established for baseline feeding preference only. Time
required: 24 h. Cost: Low. Special facilities: Coulter counter.
Special skills: Algal culture. Extrapolation: Low, unless species
very similar to these are found in the natural system. Would have to
work by "ecological analogy." See comments. Comments: The utility
of this experiment in toxicity testing is difficult to assess. The
fact that the presence of preferred food species increases feeding on
the nonpreferred food species by Daphnia magna was a surprise, but
this, in itself, was sufficient to allow extrapolation of results.
Attempts at extrapolation would require knowledge of the competitive
relationship of preferred and nonpreferred algal species.

A.3.4 Neill 1975

Mixed culture of microorganisms, naturally derived.

Replicability: Unknown. Sensitivity: Not tested. Time
required: Months. Cost: Low. Extrapolation: Possible. Comments:
Good potential for combining ecological effects and fate, except that
the fish (which is admitted to the microcosm for brief feeding
periods) could remove the toxicant.

A.3.5 Nixon 1969

Mixed culture including algae, bacteria, and grazer (marine).

Replicability: Moderate for bacteria and algae, poor for grazer.
Standardization: Moderate. Sensitivity: Not tested. Time required:
Steady state after 2 months, experiment ran for 5 months. Cost: Low.
Special skills: Bacterial and algal culture. Extrapolation: Low, no
grazers survived. Comments: Bacteria and algae stabilized after 50 d
but all attempts at maintaining a grazer population failed. This
suggests some modification of the system would be desirable before
further testing.
-------
173
A.3.6 Reed 1976

Mixed cultures with algae, bacteria, and grazers.
Standardization:
Replicability: Unknown.
Unknown. Time required: 20 weeks. Cost: Low.
Taxonomic. Extrapolation: Possible. Comments: Systems were not
used to test toxicants.
Poor. Sensitivity:
Special skills:
A.3.7 Ringelberg 1977; Ringelberg and Kersting 1978

Compartmentalized autotroph-grazer-decomposer system.

Replicability: Probably low. Standardization: Potential for
invading species high. Expect difficulties in standardizing decom-
poser chamber flowrates. Sensitivity: Unknown—see comments. Time
required: Systems were set up with intent to examine stability; ran 3
years. Time required for a chemical test unknown. Cost: Initial
costs high. Cost per chemical depends on length of run required,
difficulty of cleaning system from previous run, and reestablishing or
reexamining baseline performance. Special skills: Bacterial and algal
culture. Extrapolation: See comments. Comments: (1) These systems
are unique because they enable separate evaluation of autotrophy/
heterotrophy/ decomposer units. This obviously increases ease of
sampling and establishment of chemical fate or manipulation of the
system. (2) Increased complexity of the system and its vulnerability
to species invasion makes replication and reproducibility quite diffi-
cult to achieve, as well as expensive to attempt. (3) Use of toxi-
cants in system is feasible, but reuse of systems afterwards would be
questionable. (4) The value of this system is for investigations of
nutrient enrichment or nutrient ratio studies on the herbivore-
autotroph-decomposer relationship.

A. 3.8 Taub 1969

Mixed culture including alga, grazer, bacteria; gnotobiotic.

Replicability: Very good. Standardization: Should be
excellent, but not tested. Sensitivity: Not tested. Time required:
1 to 6 months. Cost: Low. Special skills: Sterile technique.
Extrapolation: Can extrapolate major trophic level effects to the
extent that these organisms are typical of natural organisms; i.e.,
about the same problems as single species bioassays. Comments:
Should be most able to be standardized among laboratories because
contaminant organisms are excluded. This test is limited to a few
organisms that can be cultured in bacteria free culture or with
defined bacterial flora. Could combine chemical fate and ecological
effects in one test.
-------
174
A.3.9 Taub and Crow 1978 and 1980 (in press)

Mixed culture, including algae, grazers, protozoa, and bacteria.

Replicability: Usually good agreement among 4 to 6 replicates.
Standardization: High. Sensitivity: Fairly good. Time required: 1
to 3 months. Cost: Less than $10 per replicate, excluding cost of
maintaining stock cultures. Special skills: Culture techniques, 14C,
taxonomic. Extrapolation: Can predict trophic level interactions
(e.g., effect on primary producers, or change in dominance); probably
cannot predict which species would become dominant in a specific
natural system. Comments: Especially useful for chemicals that are
transformed by some organisms to metabolites that have lesser or
greater toxicities, or that become more available via bioaccumulation.
Also useful for showing indirect effects such as algal blooms if
grazer trophic level is destroyed, or altered grazer relationships if
algal community is altered. Most "ecosystem level" variables can be
measured (e.g., P/R ratio, chlorophyll a, species abundance, species
diversity).

A.3.10 Tsuchiya et al. 1972

Grazing by protozoa on bacteria in continuous culture.

Replicability: Unknown. Standardization: High. Sensitivity:
Unknown. Time required: Days. Cost: Several $100 initial costs.
Special skills: Sterile technique. Extrapolation: Difficult.
Comments: Could be used in conjunction with mathematical models to
predict effects.
-------
175
APPENDIX B

ALPHABETICAL LIST OF PARTICIPANTS
-------
177

APPENDIX B

ALPHABETICAL LIST OF PARTICIPANTS
S. M. Adams
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
R. T. Belly
Eastman Kodak Company
Research Labs - Building 82
Rochester, New York 14607
John All
Department of Entomology
University of Georgia
Athens, Georgia 30602
B. G. Blaylock
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
R. V. Anderson
Department of Biological Sciences
Western Illinois University
Macomb, Illinois 61455
Udo Blum
Department of Botany
North Carolina State University
Raleigh, North Carolina 27607
Charles Ashton
University of West Florida
c/o U.S. Environmental
Protection Agency
Sabine Island
Gulf Breeze, Florida 32561

G. J. Atchison
Department of Animal Ecology
Iowa State University
Ames, Iowa 50011

R. M. Atlas
Department of Biology
University of Louisville
Louisville, Kentucky 40208

Ralph Baker
Department of Botany and Plant
Pathology
Colorado State University
Fort Collins, Colorado 80522

L. W. Barnthouse
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
John Bowling
Savannah River Ecology Laboratory
Drawer E
Aiken, South Carolina 29801

A. S. Bradshaw
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

Willard Chappell
Environmental Trace Substance
Program
Department of Colorado
Boulder, Colorado 80309

J. D. Cooney
Graduate Program in Ecology
University of Tennessee
Knoxville, Tennessee 37916

C. F. Cooper
Biology Department
San Diego State University
San Diego, California 92182
-------
178
C. C. Coutant
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

David Craig
Department of Biomedical Science
P.O. Box 4348
University of Illinois
Chicago Circle
Chicago, Illinois 60680

C. R. Cripe
University of West Florida
c/o U.S. Environmental Protection
Agency
Sabine Island
Gulf Breeze, Florida 32561

Geraldine Cripe
U.S. Environmental Protection
Agency
Sabine Island
Gulf Breeze, Florida 32561

D. L. DeAngelis
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

Sidney Draggan
Division of Policy Research and
Analysis
National Science Foundation
1800 G Street, NW
Washington, D.C. 20550

N. T. Edwards
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
J. W. Elwood
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

W. R. Emanuel
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

Farley Fisher
National Science Foundation
1800 G Street, SW
Room 1130 D
Washington, D.C. 20550

D. C. Force
Department of Biology
California State Polytechnic
University
Pomona, California 91768

R. H. Gardner
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

J. M. Giddings
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

J. D. Gile
Terrestrial Ecology Branch
EPA, Corvallis Environmental
Research Laboratory
Corvallis, Oregon 97330
-------
179
J. W. Gillett
Terrestrial Ecology Branch
Corvallis Environmental Research
Laboratory
200 SW 35th Street
Corvallis, Oregon 97330

Daniel Goodman
A-024
Scripps Institute of Oceanography
La Jolla, California 92093
D. R. Jackson
Ecology and Ecosystems Analysis
Section
Battelle Columbus Laboratories
Columbus, Ohio 43201

D. W. Johnson
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
Stephen Gough
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
Eugene Kenaga
9008 Building
Health and Environmental Sciences
Dow Chemical Company
Midland, Michigan 48640
T. G. Hal lam
Department of Mathematics
University of Tennessee
Knoxville, Tennessee 37916
Andrew Kindig
College of Fisheries WH-10
University of Washington
Seattle, Washington 98195
A. S. Hammons
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
D. A. Klein
Department of Microbiology
Colorado State University
Fort Collins, Colorado 80523
Stephen Hansen
U.S. Environmental Protection
Agency
200 SW 35th Street
Corvallis, Oregon 97330
Larry Klotz
Department of Biological
Sciences
State University of New York
Cortland, New York 13045
A. S. Heagle
Department of Plant Pathology
North Carolina State University
P.O. Box 5397
Raleigh, North Carolina 27650

James Hill
U.S.E.P.A. Environmental
Research Laboratory
College Station Road
Athens, Georgia 30605
J. W. Leffler
Department of Biology
Ferrum College
Ferrum, Virginia 24088

Richard Levins
Department of Population Sciences
Harvard School of Public Health
Boston, Massachusetts 02100
-------
180
Donald Levy
Lawrence Berkeley Laboratory
University of California
Berkeley, California 94720

Marc Lorenzen
Tetra Tech, Inc.
1900 116th Avenue, N.E.
Suite 200
Bellvue, Washington 98004

Alan Maki
Environmental Safety Department
Ivorydale Technical Center
Proctor and Gamble
Cincinnati, Ohio 45217

M. S. McClure
Department of Entomology
Connecticut Agricultural
Experiment Station
P.O. Box 1106
New Haven, Connecticut 06504

Allen Medine
Department of Civil and
Environmental Engineering U-37
University of Connecticut
Storrs, Connecticut 06268

R. E. Millemann
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

P. J. Mulholland
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

R. J. Mulholland
School of Electrical Engineering
Engineering South 202
Oklahoma State University
Stillwater, Oklahoma 74074
J. V. Nabholz
Office of Toxic Substances
U.S. Environmental Protection
Agency
401 M Street, S.W.
Washington, D.C. 20460

D. M. Nafus
United States Department of
Agriculture
Building 177A
BARC-East
Beltsville, Maryland 70705

J. D. Newbold
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

R. V. O'Neill
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

Richard Park
Center for Ecological Modeling
MRC-238
Rensselaer Polytechnic Institute
Troy, New York 12181

Hap Pritchard
U.S. Environmental Protection
Agency
Environmental Research Laboratory
Sabine Island
Gulf Breeze, Florida 32561

J. C. Randolph
Indiana University
The Poplars Building
Room 433
400 East 7th Street
Bloomington, Indiana 47405
-------
181
J. E. Richey
Fisheries Research Institute
College of Fisheries
University of Washington
Seattle, Washington 98195
R. F. Strayer
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
John Rodgers
Institute of Applied Science
North Texas State University
N. T. Box 13078
Denton, Texas 76203

J. L. Ruehle
Institute of Mycorrhizal
Research and Development
Forestry Sciences Laboratory
Carl ton Street
Athens, Georgia 30602

B. W. Rust
Computer Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

F. S. Sanders
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

D. S. Shriner
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

H. H. Shugart
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
J. F. Sullivan
EG&G Idaho, Inc.
Idaho Falls, Idaho
83401
G. L. Swartzman
Center for Quantitative Science
College of Fisheries
University of Washington
Seattle, Washington 98195

F. B. Taub
College of Fisheries WH-10
University of Washington
Seattle, Washington 98195

Peter Van Voris
Ecology and Ecosystems Analysis
Section
Battene Columbus Laboratories
Columbus, Ohio 43201

J. B. Waide
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830

W. J. Webb
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
B. P. Spalding
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, Tennessee 37830
-------
183
TECHNICAL REPORT DATA
(f 'lease read Instructions on the reverse before completing)
1. REPORT NO.
EPA-560/6-81-004
^RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Ecotoxicological Test Systems: Proceedings of a
Series of Workshops
5. REPORT DATE
June 1981
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
A. S. Mammons (editor)
8. PERFORMING ORGANIZATION REPORT NO

ORNL-5709
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Environmental Sciences Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee 37830
10. PROGRAM ELEMENT NO.

B2BL2S
11. CONTRACT/GRANT NO.
IAG No. EPA-78-D-X0387
12. SPONSORING AGENCY NAME AND ADDRESS
Office of Toxic Substances
U.S. Environmental Protection Agency
Washington, D.C. 20460
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
Six workshops, held from November 1979 through March 1980, at Oak Ridge, Tennessee,
were designed to identify laboratory methods and data evaluation techniques for
predicting the environmental effects of chemical substances. Participants
discussed assessment and policy requirements of multispecies toxicology test
procedures, mathematical models useful in hazard and risk assessments, and methods
for measuring effects of chemicals on terrestrial and aquatic population
interactions and ecosystem properties.

Methods were evaluated for their potential for standardization and for use in the
ecological hazard and risk assessment processes under the Toxic Substances Control
Act. Results from the workshops were used in preparing a critical review of
Methods for Ecological Toxicology (EPA 560/11-80-026; ORNL 5708). The workshops
were primarily used as a mechanism to collect information about research in
progress by bringing together investigators presently working with laboratory test
systems and data evaluation techniques.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS C. COSATI I icId/Group
Aquatic ecology
Aquatic organisms
Assessment
Biological systems
Ecology
Environnental tests
Environments
Hazard]
Invertebrates
Laboratory tot*
Hethododogy
Plants(Botany)
Terrestrial Ecology
Test methods
Toxicology
Vertebrates
Water
Aquatic microcosms
Community structur*
Ecosystea function
Ecotoxicology
Interspecific
interaction
Model ecosystem
Terrestrial microcosms
Testing protocols
06/T
18. DISTRIBUTION STATEMENT
Release unlimited
19. SECURITY CLASS (This Report)
Unclassified
11. NO. OF PAGES

196
20 SECURITY CLASS 'Thtspafcl
• Unclassified
22. PRICE
SPA Form 22W-1 ,'«,> 4-"'',
I T ION IS OBSOLETE
4U.S. GOVERNMENT PRINTING OFFICE: 1981-740-062/120
-------
I
u
CD
O
0)'

-n
0
on c
8 i
pr •
0
m
CO
tn
DO
m > TJ n
T3 (D ^ 3
>CD ° <
w •-* ~
; , 2 to -i
S^ o o o
CO ^ iJ. 3
01 03
3 CB
3
Q)

/**""""
1
Ok
-------