AQUATIC MESOCOSM TESTS TO SUPPORT
PESTICIDE REGISTRATIONS
Leslie W. Touart
U. S. Environmental Protection Agency
EEB/HED/OPP TS-769
401 M Street S. W.
Washington, DC 20460
-------
PREFACE
The Federal Insecticide, Fungicide, and Rodenticide Act (7
U.S.C. 136 et seq. ) as amended (P. L. 92-516, 94-140, and 95-396)
requires that pesticides be supported by data sufficient to allow
an assessment of potential adverse environmental effects. The
data requirements considered sufficient are delineated in federal
regulations (40 CFR Part 158). The purpose of this document is
to outline a procedure for performing an aquatic test to address
the ecotoxicity of a pesticide demonstrated to be toxic in
laboratory tests relative to expected exposures from the proposed
pesticide use or uses.
The guidance criteria proposed are the result of efforts by
several individuals. These criteria are an outgrowth of a
workshop held at George Mason University, Fairfax, VA April 8-9,
1986. Participants at this workshop are congratulated and
thanked for successfully wrestling a perplexing problem of
aquatic field testing into a viable solution. These participants
were as follows:
Howard Alexander
Dow Chemical Co.
Wesley Birge
University of Kentucky
Terence P. Boyle
National Park Service
Arthur Buikema
Virginia Polytechnic Institute
and State University
James Clark
US-EPA
David Coppage
US-EPA
Peter deFur
George Mason University
Frank deNoyelles, Jr.
University of Kansas
Reinhard Fischer
Hoechst-Germany
Jeffrey Giddings
Springborn Bionomics, Inc.
John P. Giesy, Jr.
Michigan state University
Francis Heliotis
George Mason University
Paul Hendley
ICI Americas, Inc.
Ian Hill
ICI Americas, Inc.
Robert Hoist
US-EPA
Jim Huckins
US-Fish and Wildlife
Service
Robert Huggett
Virginia Institute of
Marine Sciences
Robert Jonas
George Mason University
R. Christian Jones
George Mason University
Donald Kelso
George Mason University
-------
Thomas W. LaPoint
US-Fish and Wildlife Service
John Leffler
Ferrum College
Suzanne Levine
Cornell University
Derek Muir
Canada Department of
Fisheries and Oceans
Carolyn Offut
US-EPA
Ken Perez
US-EPA
James R. Pratt
Murray state University
Gary Rand
FMC Corporation
Richard Siefert
US-EPA
Michael Slimak
US-EPA
Keith Solomon
University of Guelph
Frieda Taub
University of Washington
Douglas Urban
US-EPA
J. Reese Voshell, Jr.
Virginia Polytechnic
Institute and State
University
Cliff Webber
Auburn University
In addition, advice and comments from the following were
invaluable in completing the criteria as presented: Richard
Anderson, Tom Armitage, John Cairns, Ken Dickson, Stuart
Hurlbert, Curt Hutchinson, Richard Lee, Mark W. Luckenbach,
Robert Pilsucki, Ann Stavola, and Zigfridas Vaituzis.
A special acknowledgement is extended to Harold Bergman,
Wendell Kilgore, Jim Tiedje, Tom Clarkson and Jim Swenberg, all
members of the Federal Insecticide, Fungicide and Rodenticide
Act Scientific Advisory Panel. Their advice and comments were
especially useful for clarifying controversial aspects of the
proposed test criteria.
ii
-------
TABLE OF CONTENTS
INTRODUCTION 1
OBJECTIVES 5
PROPOSED DESIGN CRITERIA 6
Physical Description 6
Experimental design 6
Mesocosm number 6
Mesocosm size 6
Mesocosm features 7
Mesocosm biota 8
Mesocosm treatment 9
MEASURED PARAMETERS 9
Chemical/Physical Properties 9
Biological Structure 10
Residue Analysis 11
Meteorological Conditions 12
RATIONALE 12
INTERPRETATION OF RESULTS 16
LITERATURE CITED 20
APPENDIX - Aquatic Mesocosm Workshop summaries 24
Group 1 - Mesocosm size and composition 24
Group 2 - Observational parameters 27
Group 3 - Treatment, duration and design 30
Group 4 - Interpretation of results 31
Comments and response 32
iii
-------
INTRODUCTION
Decision makers, or risk managers, face a difficult task
when making decisions concerning the registration or approval
to market of pesticide products. These decisions require sound
information on the potential risks resulting from pesticide uses
to determine if unwanted impacts are tolerable recognizing that
pesticides can benefit society. Aquatic ecosystems such as
ponds, lakes, streams, and estuaries are the ultimate
depositories of most outdoor-use pesticides. Aquatic ecosystems
receive pesticide contamination directly from certain pesticide
uses (i.e., mosquito larvicides, aquatic herbicides, etc.) and
indirectly through spray drift, surface runoff and deposition of
volatilized compounds from pesticide products applied to land.
Risk assessments are traditionally made by combining exposure
information and toxicity information to determine likelihood of
an adverse effect. Exposure can be predicted by knowing the
environmental fate of a compound and the use conditions, albeit
not as readily as stated. Toxicity is usually predicted by
employing extrapolations from single-species laboratory tests on
select representative species. Although this laboratory testing
has been a useful tool for risk managers, ecologists and aquatic
toxicologists have recognized the weaknesses of using single-
species tests alone for assessing potential ecosystem impacts
(Cairns, 1981; Pimentel and Edwards, 1982; Cairns, 1984; Levin e_£
ai. , 1984; Odum, 1984; Kimball and Levin, 1985).
Aquatic toxicologists have had limited success in obtaining
predictive information from field investigations. Field testing,
-------
when used to obviate concerns of impact to aquatic organisms, can
take a variety of forms. Most common are actual use field
studies where representative sites such as farm ponds are exposed
to typical or exaggerated (representing worst case) use practices
to determine residues and/or biological effects. These actual
use tests have been preferred for risk management decisions
because residue or biological information collected is expected
to accurately depict hazard. Since the pesticide is exposed to
natural chemical, physical and biological conditions which can
alter or mitigate its toxic potential even if these conditions
are unknown, the actual use field test has been perceived as the
best choice to unequivocally demonstrate safety. However, due to
the complexity, inherent resilience and lack of replicability of
these natural or semi-natural farm pond ecosystems, limited tests
over one or two years may not be adequate for use in hazard
assessment, especially where biological observations are limited
to a few structural parameters on highly variable populations.
Parameters most often investigated in an actual use field
study include survival, abundance, diversity and pathology.
These parameters as generally investigated are limited to just a
few of the most dominant populations. Interrelationships among
several populations within a community are seldom considered.
Functional parameters like production (increase in biomass/unit
of area/unit of time) and assimilation (production/respiration)
have been neglected.
Natural environments may not be adequately safeguarded by
protecting only a few populations. Consideration needs to be
given to important aspects of both ecosystem structure and
-------
function (Cairns, et al.. 1972). Natural ecosystems are dynamic
and cannot be effectively replicated for unequivocal cause and
effect determinations, which are desired for effective risk
management. Catastrophic events can usually be detected in an
actual use field investigation, but subtle effects which may
slowly degrade or negatively alter a system are not easily
identified. Ecologists who have recognized this deficiency have
developed physical models (i.e., simulated ecosystems,
microcosms, mesocosms) for aquatic ecosystems which allow the
necessary control and replicability to detect ecosystem-level
effects (Witherspoon et al.. 1976; Metcalf, 1977; Giddings,
1980). Mesocosms (experimental ponds and in situ enclosures) may
offer the greatest promise for providing the requisite
information for risk managers. The use of aquatic mesocosms most
likely began with the experimental ponds of Swingle (1947, 1950)
to determine the role of nutrient enrichment for increasing fish
production, long before the term "mesocosm" came into common
usage (Grice and Reeve, 1982; Odum, 1983). Several investigators
have employed mesocosms for assessing effects due to chemical
contaminants (Jones and Moyle, 1963; Hurlbert et al.. 1972;
Mclntosh and Kevern, 1974; Shindler et al., 1975; Mauck ££. al. ,
1976; Menzel and Case, 1977; Tucker and Boyd, 1978; Klassen and
Kadoum, 1979; Boyle, 1980; Kettle et al.. 1980; Papst et al.,
1980; Solomon et al. . 1980; Crossland, 1982; deNoyelles et al. .
1982; Giddings et al. . 1984; Boyle et al. . 1985; Crossland and
Wolff, 1985; Kaushik e± al. , 1985).
-------
Pond-like mesocosms are a good choice for investigating
ecotoxicity of pesticides in aquatic ecosystems. These systems
simulate ponds, shallow lakes, riverine embayments, backwaters,
etc., which form habitats for many important aquatic species
(Giddings and Franco, 1985). When containing assemblages of
organisms together with appropriate substrates and sub-systems
which are as complex as in natural communities, mesocosms should
respond to chemical perturbation in a similar fashion as
naturally occuring systems. As suggested by deNoyelles and
Kettle (1985), "one ecosystem (the experimental pond) in the
field that can be controlled, manipulated and replicated is being
used to simulate the responses of another in the field that
cannot (natural ponds and lakes)."
For a mesocosm study to be truly effective in supporting
regulatory requirements, it must address parameters which are
meaningful to risk managers. In addition, the study must be
scientifically credible, performed with appropriate methods,
verifiably accurate with a reasonable confidence of repeatability
and applicable to predicting pesticide impacts. The purpose of
this paper is to describe the criteria for an acceptable aquatic
mesocosm study to be used in an ecological risk assessment of a
pesticide. The rationale for these criteria will be presented as
will a discussion of how the results of such studies may be
interpreted. This document is not intended to detail specific
test methods, but only to provide a flexible framework for
developing an acceptable aquatic mesocosm study protocol.
-------
OBJECTIVES
An aquatic mesocosm test will serve two regulatory
objectives. First, it will provide a pesticide registrant
supportable means for negating presumptions of unacceptable risks
to aquatic organisms for their product. Such presumptions of
risk are initially based on comparisons between single-species
laboratory data and exposure information. Only those pesticides
which are presumed hazardous to aquatic organisms from such data
are required to initiate a field-level test such as an aquatic
mesocosm study. The question of whether concentrations of a
given pesticide, which result in adverse effects to aquatic
organisms under laboratory conditions, will adversely impact
aquatic organisms under field exposure conditions is addressed by
the mesocosm test. And second, it will provide risk managers
descriptive information on the extent of adverse impacts, both in
duration and magnitude, likely to occur in aquatic systems which
can then be evaluated in risk-benefit analyses.
Because an aquatic mesocosm study is an ecosystem-level
test, many parameters of both ecosystem structure and function
are compared between untreated and treated systems to ascertain
differences. Such differences must be quantitatively and
qualitatively analyzed for significance. Risk managers must know
how expected exposures of a potentially hazardous pesticide
impact populations, community structure or ecosystem function in
a representative aquatic system before making regulatory
decisions. An aquatic mesocosm study can readily address these
questions.
-------
PROPOSED DESIGN CRITERIA
Physical Description
Experimental Design —
One acceptable design is a minimum of four (4) experimental
treatments consisting of a control which receives no test
compound, an »X' treatment level representing expected exposures,
an »X+' treatment level representing an upper bound, and an »x-»
treatment level representing a lower bound. At least three
replicates per treatment level are minimally needed to provide
the requisite resolution of effects and probability of their
occurrence. However, it is recommended that the replicate number
be dictated as a function of the parameters of interest and the
sensitivity of their analysis.
Alternative designs which emphasize regression analysis and
utilize more treatment levels with fewer or no replicates may
also be appropriate. Regression designs are most useful for
determining maximum exposure conditions which provide no
significant impacts or a specified level of effect in test
systems.
Mesocosm Number
A minimum of twelve (12) mesocosms are required with
additional mesocosms added as replicates or treatments when
needed to increase the sensitivity of analysis for specific
parameters.
Mesocosm Size —
Dimensions of a mesocosm must be large enough to accommodate
a viable finfish population. Depth should be sufficient to
provide a representative open water area, and sloped sides should
-------
provide a littoral area for macrophyte growth and finfish
reproduction. An acceptable design would occupy approximately
0.1 acre surface area with a volume of at least 300 cubic meters
and a maximum depth of 2 meters. Sides of the mesocosm should be
sloped approximately 1 unit of drop per 2-3 units of linear
distance.
Mesocosm Features —
Mesocosms can be constructed as dug-out ponds or enclosures
of existing impoundments. The mesocosms should be lined with an
impervious material of known adsorption for the test compound.
The sediment used should be well defined and representative in
composition (% clay, silt and sand; % organic carbon; % organic
nitrogen; ion exchange capacity) to pond sediments in the
intended use area of the pesticide. The sediment depth at the
bottom of the systems should be a minimum of 15 cm. Sediments
may consist of natural pond sediment or top soil. If top soil is
used, the complete mesocosm should be 'seasoned' for one year
prior to experimental use. This time is necessary to develop
benthic biota. If pond sediments are used, a shorter 'seasoning'
(e.g., 6 months) period is adequate. Organic content of the top
soil should be at least 2%.
A means of interchange (circulation, fill-drain-refill,
etc.) of the water between the systems during initial
establishment is desirable to ensure even distribution of biota
among the mesocosms. Once the systems have become established or
at initiation of a test the circulation should be stopped and
each system kept separate from all other systems. The required
-------
precautions to ensure no cross contamination from pond overflow
during rainstorms, leakage in the circulation system, etc.,
should be taken from the outset.
Mesocosm Biota —
The mesocosms must contain a 'representative' pond biota.
It is recommended that an established pond with diverse biota
will act as a parent pond. The water in the mesocosm should be
equivalent to the water of the parent pond and biota collected
from the parent pond will be evenly distributed to each mesocosm
to act as a starter base. Biota from other sources may be used to
augment a natural assemblage to ensure adequate representation of
important taxa.
Phytoplankton are expected to reach a concentration
consistent with the nutrient levels of the system prior to
introduction of macroinvertebrates. Nutrient levels should be
within a mesotrophic classification. The macroinvertebrate fauna
should include representatives of the rotifers, annelids,
copepods, cladocerans, amphipods, aquatic insects and gastropods.
Introduced macroinvertebrates, if necessary to augment naturally
colonized populations, should not exceed 10 g wet weight/cubic
meter and finfish should not be introduced at more than 2 g wet
weight/cubic meter. Fish species used in the test must be of
known sensitivity to the test compound (determined from acute
toxicity tests) and appropriate to small pond enclosures.
Finfish species used must be native North American species
(bluegill sunfish alone or in combination with largemouth bass
are recommended).
8
-------
Mesocosm Treatment —
Treatment levels of the mesocosms will be based on exposure
models and residue monitoring data if available. In a three
replicate by four treatment design, the three experimental
treatments will be separated into a low, intermediate and high
treatment (dosed) and a control treatment (undosed). The
intermediate treatment will approximate the estimated
environmental concentration determined through modeling and
experiential data for the intended pesticide use. It is
recommended that the low treatment should be 1/10 the
intermediate and the high treatment should be 10 times the
intermediate. Regression designs should bracket expected
exposures and expected response concentrations. Loading of
pesticide into the mesocosms will be by direct overspray to
simulate drift and aerial deposition and with a sediment/water
slurry channeled into the system at predetermined points to
simulate runoff. Model predictions with available monitoring
data will dictate the timing, frequency and mode of introduction
of the test material.
Measured Parameters
Chemical/Physical Properties —
Mesocosm water will be monitored for pH, temperature,
transparency (turbidity), dissolved oxygen, alkalinity, total
nitrogen, total phosphorus, conductivity (total hardness) and
particulate and dissolved organic carbon at appropriate intervals
(e.g., biweekly). Observations will be made at several locations
throughout the mesocosm (which will be dictated by the physical
-------
design of the mesocosm) and at appropriate depths to allow
quantification of vertical and horizontal variations. A complete
water analysis should be' conducted at the test initiation and
termination and at significant periods during the test (i.e.,
pesticide inputs, substantial changes in other observed
parameters, etc.). Temperature, pH and dissolved oxygen should
be monitored on a continuous basis for 24 hrs. on a biweekly
schedule and at significant periods during the test to provide an
estimate of gross production and community respiration.
Mesocosm sediment must be analyzed for pesticide content,
particle size, cation exchange capacity, organic content and pH
at the initiation of the test.
Biological Structure —
Biota will be identified to species or lowest taxonomic unit
practical. The schedule for sampling and collection of
biological samples will depend on the design and composition of
the mesocosm and must be determined prior to the initiation of
the test. Collections should not be so frequent as to disrupt
the system.
Phytoplankton will be collected from the water column,
dominant species identified, and biomass determined by measuring
chorophyll a and phaeophytin. All samples should be preserved
for archival reference. Periphyton will be collected from
glass slide substrates placed in the mesocosm and exposed for a
minimum of 2 weeks. Periphyton will be analyzed for chorophyll a
and ash free weight. Macrophytes will be identified to species,
biomass determined by dry weight and per cent cover of the
mesocosm determined.
10
-------
Zooplankton will be collected weekly with tube cores of the
water column and vertical net tows. All samples will be archived
for future reference. Zooplankton samples will be analyzed
biweekly by enumerating and identifying dominant species.
Cladocerans should be identified to genus and differentiated by
size (e.g., measured for length of muon). Macroinvertebrates,
at a minimum, will be collected from emergent insect traps and
artificial substrates. Sampling of sediment directly (e.g.,
Ekman dredge), should be employed cautiously, if necessary for
tracking benthic community parameters, to minimize disruption to
the benthic community. Samples will be enumerated, identified to
lowest practical taxon and archived.
Finfish will be identified to species, enumerated, sexed
(when possible) and measured in length and weight (wet) at
introduction into the mesocosms and at test termination. Also at
test termination, females will be assessed for fecundity and all
collected fish will be examined for gross pathology. Spawning
substrates will be placed in the systems and periodically
surveyed for number of deposited eggs.
Toxicity testing and bioassays with indigenous fauna on-site
and in the laboratory may be used to assist in confirming cause-
and-effect relationships.
Residue Analysis —
Residues of the test material and major degradates/
metabolites will be analyzed at appropriate intervals to the
environmental properties of the compound in the water, sediments
11
-------
and biota at a sensitivity consistent with concentrations of
concern.
Meteorological Conditions —
Continuous monitoring of air temperature, wind velocity,
precipitation, evaporation and solar radiation are required
within 1 mile of the mesocosm test facility.
RATIONALE
Critical features of a mesocosm test design include size of
the mesocosm, its composition, duration of the test and measured
parameters. The test design presented is not intended to be
overly restrictive, and some flexibility is allowed to adjust to
specific questions of a pesticide hazard. The primary intent of
these studies is to allow potential pesticide registrants an
opportunity to demonstrate the environmental safety of their
product under conditions closely approximating those encountered
in naturally occurring systems. Risk managers require a high
level of confidence that the test employed is sensitive in
detecting adverse effects if they are likely to occur. Note,
pesticides tested in a mesocosm study should always have
demonstrated toxicity under laboratory conditions at exposure
concentrations expected to be encountered under typical uses.
Because of this indicated risk, the requirement for conducting
these studies is justified. Registrants want to demonstrate that
complex systems will mitigate the exposure or toxicity indicated
from laboratory data, regulators need information on ecosystem
level responses to evaluate in risk/benefit analysis.
12
-------
Ecotoxicity testing with aquatic mesocosms may ideally serve both
goals.
The size of an aquatic mesocosm is critical to the overall
study design. Distorting influences of large predators (e.g.,
finfish) and an inverse relationship between mesocosm size and
available surface area for periphyton growth must be balanced
with informational needs and practicality for adequate
replication and sampling. Finfish are important as integrators of
the systems and to provide the requisite end-points for risk
management decisions. The inclusion of an integrating finfish
»
population in the systems dictates a relatively large-scale
enclosure. Ponds of 300 cubic meters or larger should provide
sufficient volume for reproducing populations of a species such
as the bluegill sunfish, Lepomis macrochirus. Many reasons exist
to recommend the bluegill as the finfish species of choice. The
bluegill sunfish is a native North American species and is easily
obtainable in most areas of the United States. The species is
cultured throughout the country which provides easy availability
of healthy stocks. In optimal conditions, the bluegill will reach
reproductive maturity in 4 months (Breder and Rosen, 1966). As a
preferred laboratory test species, the bluegill is considered
reasonably sensitive with a large amount of toxicology data
available for interpretational and comparative purposes. Also, a
great deal is known about its biological requirements in small
impoundments. Finally, and more importantly, the bluegill is
planktivorous as a juvenile and insectivorous as an adult
allowing a single finfish species to cover two important trophic
roles. Where finfish predator/prey relationships are important,
13
-------
piscivorous species (e.g., largemouth bass, Micropterus
salmoides) may be included in systems of 1000 cubic meters or
larger. Although smaller systems could support the piscivorous
trophic level, the large systems will provide larger populations
which should allow more precise tracking of system perturbations.
The composition of the experimental mesocosms should include
naturally derived biota, organisms obtained from established
natural systems free from chemical contamination. It is assumed
that a natural asssemblage from most origins would contain a
collection of organisms diverse and complex enough to adequately
represent pond-like systems. Pond flora and invertebrate fauna
are relatively consistent from one area to the other at gross
taxonomic levels, and differences at the species level may be
trivial where genera and families are represented. Finfish
species other than the bluegill should be chosen for specific
purposes, and should include species appropriate for the confines
of a mesocosm and with known sensitivities for the test compound.
The duration of the test depends in part on the
environmental half-life of the test compound and its chronicity.
It is expected that, ponds can be established in early spring or
the preceding fall or winter, treated at representative times in
the year corresponding to the expected use, as indicated by
exposure models, and terminated in late fall or early winter.
When treatment timing does not allow determination of effects on
finfish reproduction or if system recovery is of interest,
studies may have to be continued over winter and through an
additional season. Treatments in the second season are optional
14
-------
depending on the registration intent and could be required if the
temporal dynamics of the pesticide in the system warrant. Tests
conducted in outdoor ponds must have at least 1/3 of the pond
area greater than 1.5 meters in depth if the test is continued
over winter to prevent complete freezing.
The parameters chosen for measurement are considered minimal
for investigating structure and function of the systems. A
biweekly sampling regime is considered adequate for tracking
potential perturbations except for zooplankton and finfish.
Zooplankton are expected to serve as sentinels of system distress
and should be sampled weekly. It is sufficent to analyze
zooplankton data biweekly, but weekly samples should be collected
and archived since they may be required to explain disturbances
if these occur. Finfish in these relatively small systems cannot
be routinely sampled and since the summation of their production
is an important end-point, sampling of the finfish population
should occur only at test termination. The intent of a mesocosm
test is to determine how a contaminant perturbs an aquatic system
and the trajectory, magnitude and duration of the peturbation if
it occurs. Protocols specific to a pesticide of concern should
be discussed with the Environmental Protection Agency on a case-
by-case basis when the study will be used to support a product
registration.
The experimental design of four (4) treatment groups
including controls with three (3) replicates in each group was
chosen for several reasons. Successful studies by deNoyelles et
al. (1982) and Boyle et al. (1985) utilized this same general
design. Differences in treatment groups as little as 15% from
15
-------
controls for fish survival and recruitment were detected as
significant at p=0.0001. Also, twelve (12) mesocosms 0.1 acre in
size appear to be logistically manageable for the observational
requirements associated in a test. The design additionally
allows the expected exposures resulting from the pesticide use to
be bounded by higher and lower treatment groups.
INTERPRETATION OF RESULTS
The objective interpretation by qualitative and quantitative
methods is defined here as analysis. Subjective interpretation
and regulatory implications will be termed evaluation. Analysis
may unequivocally demonstrate a change in a treatment pond in
comparison to control yet evaluation could determine such change
to be trivial. Risk management decisions involve much more than
simple analysis and evaluation of ecotoxicity tests such as
economic and/or social considerations, but such tests could
contribute strongly to the ultimate decisions.
A mesocosm test involves many levels of biological
organization and effects at any and all of these levels will be
analyzed and evaluated for significance. Ecosystem stress is
manifest through changes in nutrient cycling, productivity, the
size of dominant species, species diversity and a shift in
species dominance to opportunistic shorter-lived forms (Rapport
et al. . 1985). Harte et al. (1981) identify direct chemical
threats to drinking water quality, impairment of sports-fish
populations, aesthetic loss from increasing turbidity or
eutrophication, enhanced odor-producing biological activity and
increased likelihood of disease-bearing vectors and pathogens as
16
-------
the water quality issues of most concern to the public, and these
intuitively would be of most concern to risk managers. Of these,
impairment of sports-fish populations would be a very important
attribute to be affected, since it is the one most likely to
express an unwanted change in any other attribute as well.
Therefore, interpretation of ecotoxicity tests with aquatic
mesocosms will have a bottom-line assessment of the potential for
adverse effects to finfish populations. Finfish occupy the
higher trophic levels of aquatic ecosystems and therefore are the
summation of much of the activity at lower levels. Impacts at
lower levels should be expressed at all higher levels which are
dependent upon them. This is not to imply that impacts to
invertebrates or phytoplankton are not important, just that
effects at the lower levels will be interpreted for their impact
to the finfish level. Notable exceptions would include pesticide
uses likely to expose habitat of commercial shellfish importance
or where endangered/threatened aquatic species are present. In
any event, effects on organisms other than finfish will be
included in an ecological risk assessment for completeness and to
keep risk managers informed of all effects which may result from
a pesticide residue in aquatic environments.
To better exemplify the above discussion, consider two
responses which may occur in a mesocosm test. In one, fish
production is demonstrably reduced by treatment, while in the
other, fish production is unchanged between control and dosed
systems. If in the first instance, transient reductions to
macroinvertebrates were responsible for the fish production
17
-------
decrease and the decrease is assessed as temporary, then such an
impact may be considered minor. On the other hand, if the second
response is the result of permanent changes in another component
(e.g., zooplankton) of the system which will not allow fish
production to be sustained, then such a response may be
considered major. In both cases, fish production parameters were
used in making an assessment of the responses and, in both cases,
other system component parameters were necessary to interpret the
findings.
Three general outcomes of a mesocosm treatment level are
possible: (1) no discernible effect to any measured parameter;
(2) a marginal effect to one or more parameters such that the
system can compensate for the perturbation and fully recover, and
there is no substantial alteration seen or expected for finfish
populations; and (3) a substantial effect to one or more
parameters such that the system cannot accommodate the stress and
fails to recover or finfish populations are markedly reduced.
These three outcomes are possible in each of three treatment
levels. By bounding an intermediate level approximating expected
exposure concentrations with a lower and higher level by an order
of magnitude, evaluation of the results allows some inference of
probability for its occurrence. For outcome (1) at all treatment
levels, environmental concerns are sufficiently obviated to allow
pesticide registration to proceed accordingly. For outcome (3)
at all treatment levels, serious hazard is indicated and
registration is warranted only with substantial benefits and
heavy use restrictions. For other outcomes in between these
extremes, registrations will depend on risk/benefit analysis.
18
-------
Registrations could be allowed to proceed where only marginal
impacts are observed at the low and intermediate levels, but a
condition of field testing/monitoring may be imposed for the
initial years of the registration to ensure no unreasonable
impacts occur. Additionally, where marginal effects are
suggestive of high risk then additional testing (field and/or
laboratory) may be needed prior to registration.
The power of the mesocosm experimental design is expected to
be greater than the statistical sensitivity of hypothesis testing
such as ANOVA would initially suggest. The design is somewhat
constrained by practical limitations such that, at a minimum,
only twelve mesocosms are required for conducting the test. By
bounding an intermediate exposure level with a high and low level
additional statistical strength can be gained by way of
regression analysis where appropriate. From a regulatory
viewpoint, if there is no discernible effect either qualitatively
(e.g., change in dominant species) or statistically (e.g., no
significant difference in an observational parameter at a
confidence level appropriate to that parameter) between the high
treatment, which represents an exposure concentration 10 times
greater than generally expected under actual use conditions, and
control systems, given that the systems are representative of
aquatic ecosystems, then one can be reasonably confident that the
product tested will not threaten aquatic resources.
19
-------
LITERATURE CITED
Boyle, T. P. (1980) Effects of the aquatic herbicide 2,4-D on the
ecology of experimental ponds. Environmental Pollution
(Series A) 21:35-50.
Boyle, T. P. , S. E. Finger, R. L. Paulson and C. F. Rabeni
(1985) Comparison of laboratory and field assessment of
fluorene — Part II:Effects on the ecological structure and
function of experimental pond ecosystems. pp. 134-151 in
Boyle, T. P., ed. Validation and Predictability of
Laboratory Methods for Assessing the Fate and Effects of
Contaminants in Aquatic Ecosystems. ASTM STP 865, American
Society for Testing and Materials, Philadelphia, PA.
Breder, C. M. , Jr. and D. E. Rosen (1966) Modes of reproduction
in fishes. Natural History Press, Garden City, N.Y. 941 pp.
Burns, L. A., D. M. Cline and R. R. Lassiter (1981) Exposure
analysis modeling systems (EXAMS) user manual and systems
documentation report. EPA-600/3-82-023. U. S.
Environmental Protection Agency, Athens, GA.
Cairns, J. , Jr., G. R. Lanza and B. C. Parker (1972) Pollution
related structural and functional changes in aquatic
communities with emphasis on freshwater algae and protozoa.
Proc. Acad. Natl. Sci. Phil. 124:79-127.
Cairns, J. , Jr. (1984) Are single species toxicity tests alone
adequate for estimating environmental hazard? Environmental
Monitoring and Assessment 4:259-273.
Cairns, J., Jr. (Chairman), Committee to Review Methods for
Ecotoxicology, National Research Council (1981) Testing for
Effects of Chemicals on Ecosystems. National Academy Press,
Washington, DC. 103 pp.
Crossland, N. O. (1982) Aquatic toxicology of cypermethrin: II.
Fate and biological effects in pond experiments. Aquatic
Toxicol. 2:205-222.
Crossland, N. O. and C. J. M. Wolff (1985) Fate and biological
effects of pentachlorophenol in outdoor ponds. Environm.
Toxicol. Chem. 4:73-86.
deNoyelles, F. , Jr. and W. D. Kettle (1985) Experimental ponds
for evaluating bioassay predictions. pp. 91-103 in Boyle,
T. P. , ed. Validation and Predictability of Laboratory
Methods for Assessing the Fate and Effects of Contaminants
in Aquatic Ecosystems. ASTM STP 865, American Society for
Testing and Materials, Philadelphia, PA.
20
-------
deNoyelles, F. Jr., W. D. Kettle, and D. E. Sinn (1982) The
responses of plankton communities in experimental ponds to
atrazine the most heavily used pesticide in the USA. Ecology
63(5):1285-1293.
Giddings, J. M. (1980) Types of aquatic microcosms and their
research applications, pp. 248-266 in Giesy, J. P. Jr., ed.
Microcosms in Ecological Research. CONF-781101. National
Technical Information Service, Springfield, VA.
Giddings, J. M. and P. J. Franco (1985) Calibration of laboratory
bioassays with results from microcosms and ponds, pp. 104-
119 in Boyle, T. P., ed. Validation and Predictability of
Laboratory Methods for Assessing the Fate and Effects of
Contaminants in Aquatic Ecosystems. ASTM STP 865, American
Society for Testing and Materials, Philadelphia, PA.
Giddings, J. M. , P. J. Franco, R. M. Cushman, L. A. Hook, G. R.
Southworth and A. J. Stewart (1984) Effects of chronic
exposure to coal-derived oil on freshwater ecosystems: II.
Experimental ponds. Environm. Toxicol. and Chem. 3:465-488.
Grice, G. D. and M. R. Reeve, eds. (1982) Marine Mesocosms:
Biological and Chemical Research in Experimental Ecosystems.
Springer-Verlag, New York.
Harte, J. , D. Levy, J. Rees and E. Saegebarth (1980) Making
microcosms an effective assessment tool. pp. 105-137 in
Giesy, J. P. , Jr., ed. Microcosms in Ecological Research.
CONF-781101. National Technical Information Service,
Springfield, VA.
Harte, J., D. Levy, J. Rees and E. Saegebarth (1981) Assessment
of optimum aquatic microcosm design for pollution impact
studies. Rep. EA-1989. Electric Power Res. Inst., Palo
Alto, CA.
Hurlbert, S. H., M. S. Mulla and H. R. Willson (1972) Effects of
an organophosphorous insecticide on the phytoplankton,
zooplankton and insect populations of fresh-water ponds.
Ecological Monographs 42(3):269-299.
Jones, B. R. and F. B. Moyle (1963) Populations of plankton
animals and residual chlorinated hydrocarbons in soils of
six Minnesota ponds treated for control of mosquito larvae.
Trans. Am. Fish. Soc. 92:211-215.
Kaushik, N. K. , G. L. Stephenson, K. R. Solomon and K. E. Day
(1985) Impact of permethrin on zooplankton communities in
limnocorrals. Can. J. Fish. Aquat. Sci. 42(l):77-85.
Kettle, W. D. , F. deNoyelles, Jr., and C-H. Lei (1980) Oxygen
consumption of zooplankton as affected by laboratory and
field cadmium exposures. Bull. Environm. Contain. Toxicol.
25:547-553.
21
-------
Kimball, K. D. and S. A. Levin (1985) Limitations of laboratory
bioassays: The need for ecosystem-level testing. BioScience
35(3):165-171.
Klassen, H. E. and A. M. Kadoum (1979) Distribution and
retention of atrazine and carbofuran in farm pond
ecosystems. Arch. Environm. Contain. Toxicol. 8:345-353.
Levin, S. A., K. D. Kimball, W. H. McDowell and S. F. Kimball
eds. (1984) New perspectives in ecotoxicology.
Environmental Management 8(5):375-442.
Mauck, W. L. , F. L. Mayer, Jr. and D. D. Holz (1976) Simazine
residue dynamics in small ponds. Bull. Environm. Contain.
Toxicol. 16(1):1-8.
Mclntosh, A. W. and N. R. Kevern (1974) Toxicity of copper to
zooplankton. Journal of Environmental Quality 3:166-170.
Menzel, D. W. and J. Case (1977) concept and design: Controlled
ecosytem pollution experiment. Bull. Mar. science 27(1):l-
7.
Metcalf, R. L. (1977) Model ecosystem approach to insecticide
degradation: A critique. Annu. Rev. Entomol. 22:241-261.
Odum, E. P. (1983) Basic Ecology. Saunders Publ. Co. ,
Philadelphia. PA.
Odum, E. P. (1984) The mesocosm. BioScience 34(9):558-562.
Papst, M. H. and M. G. Boyer (1980) Effects of two
organophosphorous insecticides on the chlorophyll a and
pheopigment concentrationsof standing ponds. Hydrobiologia
69(3):245-250.
Pimentel, D. and C. A. Edwards (1982) Pesticides and ecosystems.
BioScience 32(7):595-600.
Rapport, D. J. , H. A. Regier and T. C. Hutchinson (1985)
Ecosystem behavior under stress. Am. Naturalist 125(5):617-
640.
Shindler, D. B. , B. F. Scott and D. B. Carlisle (1975)
Effect of crude oil on populations of bacteria and algae in
artificial ponds subject to winter ice formation.
Internationale Verenigung fur Theoretische und Angewandte
Limnologie 19:2138-2144.
Solomon, K. R. , K. Smith, G. Guest, J. Y. Yoo and N. K. Kaushik
(1980) Use of limnocorrals in studying the effects of
pesticides in the aquatic ecosystem. Can. Tech. Rep. Fish.
Aquat. Sci. 975:1-9.
22
-------
Swingle, H. S. (1947) Experiments on pond fertilization.
Bulletin No. 264, Alabama Polytechnical Institute,
Agricultural Experiment Station. 34 pp.
Swingle, H. S. (1950) Relationships and Dynamics of Balanced and
Unbalanced Fish Populations. Bulletin No. 274, Alabama
Polytechnical Institute, Agricultural Experiment Station.
Tucker, C. S. and C. E. Boyd (1978) Consequences of periodic
applications of copper sulfate and simazine for
phytoplankton control in catfish ponds. Trans. Am. Fish.
SOC. 107(2):316-320.
Witherspoon, J. P., E. A. Bondietti, S. Draggan, F. B. Taub, N.
Pearson and J. R. Trabalka (1976) State-of-the-art and
proposed testing for environmental transport of toxic
substances. Report ORNL/EPA-l, Oak Ridge National
Laboratory. National Technical Information Service,
Springfield, VA.
23
-------
APPENDIX
AQUATIC MESOCOSM WORKSHOP
A two-day workshop was held April 8 and 9, 1986 at George
Mason University in Fairfax, Virginia. Participants at the
workshop were from EPA, academia and the National Agricultural
Chemicals Association. Organization of the workshop was such
that participants were subdivided into four working groups of
approximately ten members. Each working group addressed
technical questions related to mesocosm testing in only one of
four areas — 1) mesocosm size and composition, 2) observational
parameters, 3) treatment, replication and duration, and 4)
interpretation or results. All participants attended a general
discussion on the last day which culminated the workshop.
Recommendations from participants were taken from the summaries
of the working group chairmen and from specific comments on a
circulated draft mesocosm test document. The summaries from the
working group chairmen are as follows:
Working Group 1 — Mesocosm Size and Composition
1. Finfish should be included in the mesocosm experimental
design. Fish are needed to evaluate potential adverse
effects at that trophic level. Fish are needed as a
component of the mesocosm system to complete the model
ecosystem design and, thus, allow the development of
potential secondary effects due to the presence of a
consumer species.
2. Finfish in the mesocosm should be planktivorous as juveniles
and insectivorous as adults. Top predators require too
large an enclosure and too long a time to develop a stable
population.
24
-------
3. Two size classes of finfish should be introduced — young,
sexually mature fish which will reproduce during the
experimental period and immature fish. Two size classes
provide a minimal model of natural finfish population
distribution. Mature fish can be used within one annual
season for reproduction evaluation.
4. Population density of the finfish should be adjusted to
simulate natural population densities for the chosen fish
species in a similar (to the mesocosm) habitat. Densities
within the range of 2-5 g/cubic meter are appropriate. Care
must be taken due to potential serious disruption of
individual systems if loss of fish occurs. Sufficient
numbers of fish are needed to reduce effects of small
losses.
5. Volume of the mesocosm should be at least 300 cubic meters.
Maximum depth should be 2 meters. Sides of the mesocosm
should be sloped approximately 1 unit of drop per 2-3 units
of linear distance. Dimensions must be large enough to
accomodate a reliable finfish population. Depth should be
sufficient to provide a representative open water area, and
sloped sides should provide a littoral area for macrophyte
growth and finfish reproduction. The design described
should occupy approximately 0.1 acre surface area.
6. Mesocosm structure should normally be a dug pond, but
naturally occurring or 'old' ponds which meet all other
criteria will not be excluded. The required size and shape
is not amenable to prefabricated, container construction.
Provided sufficient land is available, the mesocosms can be
machine dug at relatively low cost.
7. Larger ponds may be subdivided to act as replicate mesocosms
provided that each subdivision meets all other criteria.
Subdivisions should all be to the same size, shape and
dimensions. Design of subdivisions should be such that no
leakage occurs between adjacent mesocosms. Materials used
for subdivision should not introduce any toxic substance
which could significantly affect the test.
8. Bottoms of the dug ponds should be lined with impervious
material (clay or plastic sheet) to avoid interconnection
with groundwater.
9. Sediments should be added (over the impervious layer) to a
depth of at least 10 cm. Sediments may consist of natural
or aged pond sediment or top soil. If top soil is used the
complete mesocosm should be 'seasoned' for one year prior to
experimental use. This time is necessary to develop benthic
biota. If pond sediments are used, a shorter 'seasoning'
(e.g. , 6 months) period is adequate. Organic content of the
top soil should be at least 10%.
25
-------
10. Ponds may be reused without sediment removal if water
quality parameters and biota are shown to be similar among
all mesocosms. Sediment removal and/or other cleaning to
assure uniform starting conditions may be carried out as
needed.
11. Sampling methods should not significantly disturb the
sediments or water column.
12. An even distribution of biota among all test mesocosms is an
overriding objective. Various options may be employed to
achieve this goal. For example, all mesocosms could be
drained and refilled from a common source immediately before
the start of a test cycle, or a mixing system among all
mesocosms may be employed provided an even distribution of
the biota can be achieved.
13. Establishment of a 'representative' pond biota is a major
objective. This may be achieved by addition of biota from
other sources including natural ponds, aged systems or
cultured material.
14. Treatment time should coincide with applicable use patterns.
If established 'aged' pond water is used to fill the
mesocosms, they should be filled and sampled for at least 4
weeks prior to introduction of the test compound. If water
from another source is used, sufficient time must be allowed
for development of a representative pond flora and fauna.
15. All mesocosms should contain representative macrophytes
evenly distributed with regard to biomass and species
composition. At least submerged macrophytes should be
represented. Prior to the start of an experiment,
macrophyte population structure should be mechanically
adjusted to meet the above criteria. Macrophyte populations
should not be manipulated during the experimental period.
Mesocosm design criteria are intended to provide some areas
which are macrophyte free.
16. Minimal fertilizer addition can be used to avoid a pond
model which would be highly oligotrophic. This item is not
to be interpreted to mean that the mesocosm should be
eutrophic. The objective is to model a 'representative'
biologically active ecosystem.
17. Water level within the mesocosm systems may be adjusted to
account for evaporation or rainfall inputs. If the levels
change from the starting conditions by 5 cm either over or
under, they should be adjusted to the initial levels. Water
added to the systems during the experiment should be
filtered through a sand filter (e.g., swimming pool type
filter) to remove most plankton. All experimental mesocosms
should be adjusted to the same level simultaneously.
Particular care must be taken in subdivided systems
26
-------
curtained with flexible materials so that water levels in
adjacent enclosures remain the same.
18. It is suggested that a mesocosm test in support of the
registration of a particular compound be conducted in the
major use region for that compound.
19. Because of the experimental design and controlled nature of
the experiments it is felt that results from a test in one
geographic region could be extrapolated to other regions.
20. One set (multitreatment, replicate mesocosm) of mesocosm
tests could be sufficient for regulatory purposes, either
acceptance for registration or denial of registration.
Should the result imply significant, long-term hazard in one
geographical location further mesocosm testing in other
areas under different conditions could be used to achieve
limited compound registration for similar regions.
Working Group 2 — Observational Parameters
1. Physico-chemical parameters
A. Water column
1)
2)
3)
Parameter
PH
transparency
alkalinity
total nitrogen
total phosphorus
conductivity or
total hardness
dissolved organic
carbon
particulate organic
carbon
dissolved oxygen
temperature
PH
Frequency
biweekly
biweekly
temperature
biweekly
(maximum and
minimum for
2 weeks prior)
Location of sample
integrated sample
(from sufficient
samples to include
vertical and
horizontal
variations)
subsurface and
10 cm above the
bottom at
sufficient points
to include all
variations needed
to characterize
the mesocosm
0.5 m below
the surface
27
-------
4) target pesticide
5)
contaminants
at regular intervals
appropriate to the
compound
beginning and end
of test
Considered but rejected: BOD, COD, free C02, redox potential,
forms of N, forms of P, sulfur (except in cases involving
arsenicals), and inorganic carbon.
B. Sediment
Parameter
1) particle size
cation exchange
capacity
organic content
PH
pesticide scan
2) target pesticide
Frequency
at beginning of
test
Location of sample
at regular intervals
appropriate to the compound
Considered but rejected: dissolved oxygen, depth of redox
potential discontinuity (RPD) layer, oxygen uptake, ammonia
efflux, sulfur, temperature.
C. Atmosphere
Parameter
air temperature
wind velocity
precipitation
evaporation
solar radiation
Frequency
continuously
Location of sample
at a meteorological
station within l mile
of the mesocosm
2. Biota
A. Biotic component
1) Phytoplankton
Frequency Parameter
and Technique of Measured
Sampling
biweekly measurements
from at least 3
integrated samples
chlorophyll a and
phaeophytin;
identification of
dominant species;
preserve all
samples for
archival reference
28
-------
2) Periphyton
3) Mac rophytes
4) Zooplankton
5) Macroinvertebrates
6)
Finfish
biweekly collections
of glass slide sub-
strates exposed for
2 weeks
sampling as
appropriate and at
end of test
weekly collections,
biweekly counts;
tube cores of the
water column and
vertical net tows
biweekly collections
from emergent insect
traps and artificial
substrates
sampling at beginning
and end of test (test
should be long enough
to span reproductive
cycle)
chlorophyll a and
ash free weight
percentage cover
of each species;
dry weight
archive all
samples; in
alternate weeks
count dominants
to species or
genus and note
length of muon
of cladocerans
count to lowest
practical taxon;
archive all
samples
count to species,
measure length &
wet weight; note
pathologic
conditions; target
pesticide concen-
tration in fish
tissue
Considered but rejected: sampling of microbiota and epipelon.
B. Bioassays
Bioassays were considered using caged organisms in the
mesocosm or toxicity testing of organisms removed from the
mesocosm, but both were rejected as being redundant.
C. Ecosystem function
While these measures (production at each trophic level) were
thought to be valuable, it was felt that the measurments of
structure of the mesocosm ecosystem over time were
sufficient to determine these parameters.
29
-------
Working Group 3 — Treatment, Duration and Experimental Design
1. A minimum of 4 treatments consisting of a control, an 'X'
treatment level representing expected exposures, an 'X+'
representing an upper bound, and an 'X-' representing a
lower bound. Expected exposures would be predicted from
appropriate drift and runoff models and/or relevant
empirical studies.
2. A minimum of 3 - 5 replicates per treatment as a function of
parameters of interest, mesocosm design, etc. Existing data
sets (and new data as it becomes available) should be
consulted for establishing minimum replicate numbers.
3. Treatment frequency is based on the number of applications
permitted on the pesticide label for aerial drift simulation
and on established runoff models (e.g., SWRRB) for runoff
simulation.
4. Aerial treatment simulations should be by spray on surface
with pesticide finish spray. Runoff simulations may be by
uniformly distributing a subsurface slurry of pesticide in
water and sediment. Problems are expected in determining
the method for creating the slurry, dealing with compound
alterations while on soil, determining the appropriate
particle size, and so on.
5. Test ponds must be aged at least one year if using
established pond sediments. Use of established pond
sediments in mesocosms is a trade off between pond stability
and similarity of replicates. Test ponds using topsoil will
require study to determine an adequate pretreatment interval.
The duration of a mesocosm test is at least 1 full growing
season and longer for persistent compounds, but not without
added problems.
6. Macrophytes will cause increased heterogeneity and variance
between replicates in addition to other sampling problems.
No macrophytes or intensely controlled macrophytes should be
used in the test ponds. Two species of fish (i.e., bluegill
sunfish and largemouth bass) are recommended rather than two
age classes of one species. The ponds should have fish,
sediment, invertebrates and other orgainisms seeded from an
established pond.
30
-------
Working Group 4 — Interpretation of Results
1. Finfish are of primary importance for detrmining significant
effects and meaningful endpoints from conduct of a mesocosm
study. Fish number, biomass, growth, fecundity (estimated
from nesting substrates), survival, and residues in tissues
(e.g., gravid gonads) should be determined.
2. Zooplankton community structure determined by observations
on total biomass and species dominance. Size may be used
instead of detailed taxonomy. Changes to zooplankton should
provide the first indications of system impact.
3. Benthic community structure should be studied to determine
major taxonomic shifts. Such shifts will give clues on
changes of detrital processes.
4. Plankton and periphyton should be studied for changes in
chloropyll a and dominance over time.
5. Community metabolism is tracked by monitoring pH, dark
respiration and light oxygen evolution, preferably by
automated continuous recorders.
6. The use of diversity or absolute ratio indices should be
avoided.
7. Geographic extrapolations of mesocosm results may be derived
from available fate models. Biological extrapolations from
mesocosm investigations are limited without linking field
investigations.
8. Fish are the integrators of the experimental ecosystems.
The systems should not be overloaded by fish biomass and the
fish should not require external feeding.
9. Sediment is expected to be somewhat of a "black hole" for
certain compounds. The sediment requires characterization
for the sorption nature, rates, etc. The biomass/sediment
ratio should be considered an important design parameter.
10. Bioassays with selscted organisms are important before,
during and after treatment of the mesocosms.
11. In the pesticide registration and hazard evaluation process,
there is a continued role for microcosm and full-scale field
investigations.
31
-------
Substantive comments received from participants of the
April, 1986 workshop not fully resolved within the document are
discussed here:
1. A strong reservation exists among some that reliance on fish
data may be precipitous due to inherent variability expected
between ponds. It is suggested that credibility of an approach
which relies on fish population parameters would be enhanced by
referencing successful studies and detailing the sensitivity of
statistical differences, specifically if three replicates would
be sufficient to separate treatment differences. It is further
suggested that such data could be used in defining performance
criteria for use in evaluating the acceptability of tests
dependent on these parameters.
Comment — The Agency agrees that documentation of successful
studies is needed to allow implementation of the mesocosm test
philosophy without reservation. However, as a new initiative in
this arena few mesocosm studies with pesticides exist and none of
these tests are fully consistent with the proposed design.
Because of this lack of empirical data to support the proposed
design, the criteria presented here and Agency position are that
this document represents only a proposed framework for fulfilling
aquatic organism simulated and/or actual field test requirements
in support of pesticide registrations. The Agency is convinced
that the proposed design adequately represents a consensus among
the scientific community, that reliance on fish data are
consistent with the available data base, and the Agency fully
expects tests conducted in accordance with these criteria will
allow acceptable hazard evaluations of pesticide registrations to
aquatic ecosystems. It is anticipated that as an empirical
base of these tests is developed the Agency will be in a position
to evaluate its position and further specify the limits of
acceptability for these tests. In the interim, the Agency will
use the criteria presented as guidance in evaluating simulated
aquatic field investigations.
2. Some confusion is still present on how an Estimated Exposure
Concentration (EEC) will be used in specifying treatment levels.
Comment — An EEC is used when empirical data on expected
exposure concentrations are absent or insufficient for specifying
treatment levels. Agency calculated EECs are based on empiricaly
derived models and manipulated over a range of likely scenarios.
Treatment levels should be designed to span this range such that
a low level would be at or below the lowest expected exposure
concentration, a high level at or above the maximum or worst-case
exposure concentration and a median level which roughly
approximates a typical exposure concentration consistent with the
supported pesticide label. The exposures in the test systems
should simulate as nearly as practical the modelled exposure.
That is, the water column concentration in the test system should
be equivalent to the modelled water column concentration for that
treatment level, the sediment-bound concentration in the test
32
-------
system should be equivalent to the modelled sediment-bound
concentration for that treatment level, etc. Therefore,
sufficient chemical needs to be added to a test system to ensure
that the water concentration is equivalent to the modelled value
after chemico-physical partitioning. In general, order of
magnitude differences between treatment levels in a '3-dose by 3-
rep' mesocosm test design will suffice, such that a median
treatment level (x) is bracketed by a low level (1/10 x) and a
high level (10 x). Again, the median level represents a typcial
exposure condition expected for the given pesticide use under
investigation.
3. Several participants have suggested experimental design
changes to provide better dose-response information and thereby
improve result interpretation (i.e., '5-dose by 2 rep','11-dose
by no rep», etc.).
Comment — The Agency has no objections to modifications in
experimental design so long as a minimum of 12 mesocosms are
utilized and the minimum and maximum expected exposures are
bounded by dose levels employed. A preference is given to a '3-
dose by 3-rep' design because, as the Agency believes, it can
effectively span the expected exposure range and provide adequate
replication to separate obfuscating observations which result
from individual differences between the mesocosms. The Agency
will rely solely on established differences in interpreting the
results from these tests and true effects which go undetected
would, therefore, be considered to pose minimal threat to aquatic
ecosystems. Designs with less than 3-reps/level may not provide
equivalent separation.
4. A few participants objected to size limitations of the
mesocosm and preference given to bluegills as the fish component.
It is felt that other fish species or even invertebrate species
could provide the requisite information and certain of these
species could be tested in systems of less than the 300 cubic
meters.
Comment — The Agency dependence of native North American
species for use in performing ecological effects studies has been
well established. Mesocosm tests must include finfish for both
direct and indirect effect evaluations. Effects seen on
invertebrates alone cannot provide these data. Direct responses
to a toxicant may not be fully exhibited by finfish in the
laboratory. Alterations of behavior, for example, could affect
feeding, reproduction, etc., in the field which would not be
easily discernible from laboratory data. Indirect responses of
finfish to a toxicant may not be explained by changes in
invertebrate production alone. Alterations of diet, for
example, could impact a fish differently when it is stressed by a
pesticide. Bluegills are perceived as the most appropriate
species when the mesocosm includes only a single fish species.
The same rationale as used for preferring it in acute laboratory
tests can be given together with, its life history
(planktivore/insectivore and reproductive maturation in as little
33
-------
as 4 months) and extensive culture experience in small ponds.
Many of the mesocosm studies referenced successfully included
bluegills. Certainly other species of fish may be used but the
Agency would be hesitant to approve protocols with different
species without sufficient justification. Since bluegills are
preferred, 300 cubic meter systems would be the minimally
accepted mesocosm size consistent with the species requirements.
5. Some participants felt that the list of measured parameters
were too extensive and emphasized system structure without
consideration of system function.
Comment — The list of recommended parameters to be measured
during a mesocosm test is almost exactly those recommended by the
sub-group dealing on the issue at the April workshop. This list
was pared down from a more extensive list and determined to be
the minimum data set sufficiently comprehensive to charaterize
pesticide effects. Additional emphasis may be needed for some
chemicals and can only be determined after scrutiny of their
database. The sub-group also determined that the listed measures
of structure over time can provide sufficient information on
ecosystem function to draw meaningful conclusions.
6. A few participants commented on macroinvertebrate sampling
and stocking in the test mesocosms. It is felt that insufficient
emphasis is given to benthic invertebrates and aquatic insects.
Comment -- Potential disruption of benthic sediments is the
overriding consideration in limiting macroinvertebrate sampling
to artificial substrates and emergent traps. However, this
limitation is a "minimal" limitation in that additional sampling
of system substrate and shoreline is acceptable. Note that
increased sampling of the systems will follow the law of
diminishing returns. As more sampling takes place the greater
the likelihood that the system will be disrupted. The point was
well taken by some that stocking of aquatic insects into the
systems may be wasted effort as natural colonization should be
more than adequate. The Agency intent here was to ensure a
reasonably representative macroinvertebrate fauna in systems
which may have had limited time for natural development.
7. Some confusion exists on whether mesocosm tests are designed
to evaluate ecosystem effects or to evaluate indirect effects on
fish.
Comment — The Agency wants to emphasize that the purpose of the
mesocosm test philosophy is to evaluate ecosystem effects.
However, interpretations of the significance, if any, of these
effects will be directed towards socially meaningful ends (e.g.,
finfish). Finfish parameters are perceived as the ones which
will be less variable and an integration (for some species of
fish) of other system parameters, therefore more sensitive from
an analysis viewpoint. The Agency understands that other
34
-------
ecosystem attributes are important and evaluations will include
all such attributes impacted. Pragmatically, effects to finfish
will, in most instances, carry the overriding emphasis of
potential regulatory outcomes.
35
------- |