xvEPA
United States
Environmental Protection
Agency
Environmental Research
Laboratory
Athens GA 30613
Research and Development
EPA-600/S3-81-036 July 1981
Project Summary
Microcosms as Test
Systems for the Ecological
Effects of Toxic Substances:
An Appraisal with Cadmium
Paul F. Hendrix, Christine L Langner, Eugene P. Odum, and Carolyn LThomas
A two-phase set of experiments was
conducted to address some of the
problems inherent in ecological screen-
ing of toxic substances in aquatic
microcosms, and to test two hypoth-
eses concerning the response of
ecosystems to perturbations. Phase I
was a 4 X 4 factorial experiment (four
levels of cadmium versus four levels of
nutrient enrichment) with static micro-
cosms designed to test the "subsidy-
stress" hypothesis, and focused on
the interactive effects of cadmium and
nutrients. Phase II was a 2 X 4 factorial
experiment (continuous and pulsed
cadmium inputs versus phosphorus
limited and non-limited inputs) with
flow-through microcosms designed to
test the "biomass increment" hypoth-
esis, and focused on temporal aspects
of system behavior (especially out/
input for several elements) in response
to nutrient limitation and chronic
versus acute cadmium perturbations.
Phase I results supported the sub-
sidy-stress hypothesis with respect to
cadmium inputs: Increasing cadmium
concentrations (0, 1, 10, 100 ppb)
caused a decrease in the P/R ratio, a
decrease in grazing herbivores, increase
in nighttime respiration and fungi, all
indicators of system stress. Since net
daytime production and nighttime
respiration increased with nutrient
enrichment, there was no nutrient
stress effect even at the highest level.
There was a significant interaction
effect of cadmium and nutrients with
high nutrient levels reducing, some-
what, the stress effect of cadmium.
Phase II results generally supported
the biomass increment hypothesis and
suggested a retention pattern for
continuous, low concentration cad-
mium inputs similar to that of essential
elements. Cadmium may have accu-
mulated to a toxic threshold in some of
the microcosms. Pulsed, high concen-
tration cadmium inputs had significant
effects on system behavior, depending
on timing of inputs.
Conclusions relevant to toxicity
screening in microcosms are:, 1) Of the
variables measured, community me-
tabolism, community composition by
trophic groups, and output/input
ratios for NO3-N, Mn and Fe, provided
the best indicators of system response
to cadmium. 2) Nutrient enrichment
and phosphorous limitation signifi-
cantly influenced cadmium effects on
most of the variables studied. 3)
Pulsed cadmium inputs early in suc-
cession significantly affected system
response to cadmium pulses later in
succession.
For screening a suspected toxic
substance, we recommend a hierarchy
of microcosm experiments including:
1) static microcosms (with and without
sediments), 2) flowthrough microcosms
(with and without sediments), and 3)
-------
microcosm subsamples from specific
natural ecosystems. Each step results
in increased information about effects
of a toxicant and each step more
closely approximates natural ecosys-
tems.
A bibliography of microcosm litera-
ture is presented at the end of the
Project Report.
This report was submitted in ful-
fillment of Grant No. R805860010 by
the University of Georgia under the
sponsorship of the U.S. Environmental
Protection Agency. The report covers
the period 22 May 1978 to 31 Sep-
tember 1980, and work was completed
as of 30 September 1980.
This Project Summary was devel-
oped by EPA's Environmental Research
Laboratory, Athens, GA, to announce
key findings of the research project
that is fully documented in a separate
report of the same title (see Project
Report ordering information at back).
Introduction
In accordance with the Toxic Sub-
stances Control Act of 1976, the U.S.
Environmental Protection Agency is
developing testing standards for evalu-
ating potential hazards of chemicals
before they are manufactured and
released into the environment. To this
end, interest is currently being shown in
the use of microcosms for toxicant
screening and predictive model vali-
dation.
The use of microcosms for these
purposes is somewhat controversial
because of the uncertainty involved in
extrapolating results to natural condi-
tions. When considered as generalized
models of ecological processes, how-
ever, small scale microcosms might
provide a means for evaluating gross
effects of toxic substances on ecosys-
tems because such microcosms do
mimic certain functional properties of
ecosystems. For example, a number of
studies have demonstrated similarities
between temporal processes in natural
systems and in microcosms, including
species succession, biomass accumula-
tion, net production and community
respiration, and radioisotope uptake
and distribution. In addition, similar
responses of natural and microcosm
systems have been reported for various
perturbations, including radiation, tem-
perature, heavy metals, arsenic, organic
toxicants, and nutrient enrichment.
Thus, although quantitative extrapola-
tion from microcosms to the real world
results is not currently feasible, the
qualitative response of microcosms to
inputs of toxic materials under controlled
laboratory conditions may provide a
basis for a "first approximation" of the
ecological effects of toxic substances.
The development of standardized
testing procedures requires answers to
several important questions, including:
1. Which ecosystem properties are
most sensitive or best reflect
ecosystem response to toxicant
perturbations?
2. What influence will other environ-
mental variables (e.g., pH, nutrient
enrichment, light intensity, etc.)
have on ecological effects of a
toxic substance?
3. Will ecosystem response be a
function of the timing or frequency
of toxicant inputs with respect to
stages of ecosystem development?
4. What degree of realism (biotic and
abiotic complexity) should be in-
corporated into microcosms for
use in toxicity screening?
To address these questions and to
further evaluate the potential utility of
microcosms as ecological screening
tools, we have conducted a series of
experiments in which aquatic laboratory
microcosms were exposed to a toxic
substance. Because most of these
questions are important, not only for
screening protocol development but for
ecosystem analysis, in general, we have
designed the experiments to test two
hypotheses which have been developed
to explain ecosystem behavior in re-
sponse to stress (toxic substances being
a specific form of stress). The experi-
ments were conducted in two phases,
each addressing a different hypothesis.
Phase I
It has been suggested that ecosystems
respond to environmental perturbations
in a "subsidy-stress" fashion. At low to
moderate levels of intensity, system
inputs often act to subsidize or increase
overall system function (e.g., the effects
of nutrient enrichment or increase in
temperature or productivity). Conversely,
high levels of the same input can de-
crease system function or result in
development of an entirely different
system (replacement). The overall pat-
tern is a unimodal, bell-shaped curve of
system response along a gradient of
increasing perturbation intensity. It also
is hypothesized that relative variance of
system response increases monoton-
ically along the perturbation gradient.
System response to a toxic or lethal
input is hypothesized to be a stress at;
levels of input. Complicating the:
general response patterns are the ii
fluences of environmental and develo)
mental gradients, such that systei
response to a given level of perturbatio
might vary with environmental cond
tions or successional stages. Thes
interactive effects are especially impoi
tant considerations for toxicity screenin
because test results will be unavoidabl
biased by standard testing condition:
An alternative to single factor expert
ments (i.e., varying levels only of,
toxicant) might be a multifactor c
factorial experimental design that wouli
allow for consideration of the interaction
of several factors simultaneously.
Phase I was designed to test tht
subsidy-stress hypothesis and to evalu
ate the influence of an environmenta
variable (nutrient enrichment) on aquatic
microcosm response to a toxic substance
(cadmium). The experiment was arrangec
in a 4 X 4 factorial design with increas-
ing levels of nutrient enrichment super-
imposed on increasing cadmium levels.
Of particular interest were the inter-
active effects of nutrients and cadmium
on several system level variables.
Phase II
A number of ecosystem studies sug-
gest that nutrient output/input ratios
are sensitive system level measures of
ecosystem behavior and stress response.
These studies indicate that the loss of
essential elements from ecosystems
often increases significantly after dis-
turbance. For example, loss of calcium
has been shown to be a sensitive indi-
cator of stress. The "biomass increment"
hypothesis suggests that nutrient out-
put is an inverse function of the rate of
biomass production within an ecosystem.
Briefly, the hypothesis for an essential
nutrient is: Prior to biotic colonization of
an area, nutrient outputs are equal to
inputs (barring abiotic uptake or loss).
As biota become established and eco-
system development proceeds,, nutrient
output becomes less than input as a
result of biotic uptake and storage in
growing tissues. At the time of peak net
ecosystem productivity, the ratio of
nutrient output/input is at a minimum,
thereafter gradually increasing to unity
as net productivity approaches zero at
ecosystem maturity (steady state). A
pulsed perturbation to the ecosystem
results in an increase in nutrient out-
put/input followed by secondary suc-
cession and an abbreviated repeat of the
-------
initial patterns of productivity and nutri-
ent uptake. For nonessential elements
output/input remains near unity
throughout the entire sequence of
events; for limiting quantities of es-
sential elements, deflection of the
output/input curve is related to the
degree of limitation. In addition, it .has
been proposed that ecosystems must
have a finite capacity to accumulate
toxic elements, unless they have a
capacity through microbial transforma-
tions to gasify the elements (as can
occur with mercury, for example). As
that capacity is approached, increasingly
greater proportions of toxicant input
should appear in system outputs, such
that the retention pattern should fall
somewhere between those of non-
essential, not accumulated, and limiting
elements. If this is true, the potential for
an ecosystem to become a source of
(rather than a sink for) toxic elements
increases as the system approaches
maturity.
Phase II was designed to test these
hypotheses and to evaluate the utility of
output/input ratios of several elements
as indicators of microcosm response to
toxic element (cadmium) perturbations.
Several other factors were incorporated
into the experiment to determine: 1) the
influence of pulsed versus continuous
toxicant inputs on system response, 2)
the effects of toxicant exposure early in
succession, and 3) the influence of
nutrient limitation on system response
to toxicant exposure. The experiment
was arranged in a 2 X 4 factorial design
with phosphorus-limited (N:P = 100) and
non-limited (N:P = 10) input regimes
superimposed on four modes of cadmium
input (zero input, continuous input,
cadmium pulses early and late in suc-
cession, and a cadmium pulse late in
succession).
Results and Discussion
Levels of nutrient enrichment used in
Phase I(fromN=0.1,P = 0.01 toN = 10,
P = 0.1 ppm) failed to produce a subsidy-
stress response as far as productivity/
respiration (P/R) relations were con-
cerned; both P and R increased equiva-
lently along the nutrient gradient.
Whether higher levels of enrichment
would disrupt P/R remains to be deter-
mined. In contrast, cadmium treatments
(0 to 100 ppb) caused a significant
decrease in P/R (deviation increasing
with cadmium concentration) as pre-
dicted by the hypothesis. This effect was
attributed to a decline in grazing her-
bivores (micro-crustacea were virtually
eliminated at higher cadmium levels)
and an increase in bacterial and fungal
populations with consequent increase
in community respiration. Accordingly,
cadmium had an overall impact of
switching energy flow from a grazing
food chain to a detritus food chain.
Phase II results generally supported
the "biomass increment" hypothesis.
Outputs of B, Ca, Cu, Mg, Na and Zn, all
essential but in excess of biotic demand,
remained equal to inputs throughout
the experiment. Two exceptions to the
predicted trends were noted. First,
maximum uptake of all essential ele-
ments did not coincide with maximum
metabolic activity. Luxury consumption
may have been responsible for the early
occurrence of maximum phosphorus
retention and might be expected to
occur for other essential, limiting ele-
ments as well. Second, disturbances
(i.e., cadmium pulses) that caused
significant changes in metabolic activity
were not reflected most strongly in the
retention patterns of phosphorus, the
element most limiting in system inputs.
Outputs of NOa-N, present in abundance
relative to phosphorus, showed the
strongest disturbance response, possibly
as a result of selective cadmium effects
on nitrogen metabolism. Results from
Phase II also indicated cadmium reten-
tion patterns similar to those for essential
elements and suggested that cadmium
accumulation was a function of produc-
tivity. In the less productive systems,
cadmium outputs approached input
levels by the end of the experiment (286
days), whereas the more productive
systems continued to accumulate cad-
mium. Because inorganic sediments
were not present in the microcosms,
cadmium must have been retained or
stored in the biomass but it is not
possible to tell from these data whether
the mechanism was active biochemical
uptake by living cells or sorption onto
detrital materials.
Conclusions and
Recommendations for
Toxicity Testing
Results of these experiments suggest
some tentative answers to the ques-
tions raised in the Introduction.
1. Which ecosystem properties are
most sensitive or best reflect eco-
system response to toxicant pertur-
bations?
Of the ecological variables measured
in this study, community metabolism
(net daytime production and especially
nighttime respiration) and densities of
various taxonomic groups provided the
most consistent indicators of cadmium
effects. The ratio of net production to
community respiration (P/R) has been
suggested as a useful measure of toxi-
cant stress in microcosms but proved
responsive to cadmium only in the static
systems in our study; in the f lowthrough
systems, PD and Rn both responded
similarly, resulting in no net change in
P/R. Reasons for this difference are not
clear, but it does seem apparent that PD
and Rn, expressed individually, are
important and easily measured variables
in microcosm studies. Biomass and
plant pigment concentrations were the
least sensitive to cadmium of the vari-
ables measured. Biomass accumulation
rates and pigment ratios might prove to
be more useful. In the flowthrough
systems, output/input ratios of N03-N,
Mn and Fe showed significant responses
to cadmium treatment. This illustrates
the potential utility of output/input
ratios (especially nitrogen) for toxicity
screening and suggests further that
some toxicants might selectively alter
specific metabolic pathways. Estimates
of rates of metabolism of certain essen-
tial elements (e.g., N, P, S) should be
considered for use in microcosm screen-
ing tests.
2. What influence will other environ-
mental variables (e.g., pH, nutrient
enrichment, light intensity) have on
ecological effects of a toxic substance?
Nutrient enrichment and phosphorus
limitation significantly influenced the
cadmium response of most of the vari-
ables measured. In general, the poorly
enriched microcosms were more sensi-
tive to cadmium than their highly en-
riched counterparts. The importance of
this finding for toxicity screening in
microcosms is that standard testing
conditions, such as levels of nutrients
and other factors, are likely to influence
test results. This unavoidable bias can
be minimized or at least accounted for
by conducting screening tests in matrix
or factorial experimental designs that
include potentially interacting factors.
In particular, if tests are run in micro-
cosms of site-specific derivation, then
environmental factors important in a
given geographic area (e.g., salinity, pH,
temperature extremes) could be incor-
porated into the test, along with toxicant
levels, for a more meaningful evaluation.
Any number of factors could be included
in such a scheme (including several
toxicants), but experimental costs would
increase with each factor. Judicious
-------
choice of potentially important factors
would be required.
3. Will ecosystem response be a func-
tion of the timing or frequency of
toxicant inputs with respect to stages
of ecosystem development?
The mode of toxicant introduction into
microcosms is an important considera-
tion for toxicity testing. Because toxic
substance inputs into natural ecosys-
tems occur over wide ranges of fre-
quency and magnitude, one-time
additions of a toxicant to microcosms
might not provide a meaningful evalua-
tion of ecological effects. In the present
study, cadmium was added to the static
microcosms only at the beginning of the
experiment, precluding any considera-
tion of toxicant input dynamics. We
addressed this problem in flowthrough
microcosms by applying cadmium in
pulses at several stages in succession.
Results showed that cadmium pulses
early in succession significantly affected
system response to later pulses, possibly
due to selection for tolerant organisms.
We also compared flowthrough micro-
cosm responses to continuous, chronic
versus acute, pulsed cadmium exposure.
Continuous 10-ppb Cd inputs may have
caused a toxic threshold response, but
results are inconclusive.
4. What degree of realism (biotic and
abiotic complexity) should be incor-
porated into microcosms for use in
toxicity screening?
Generally, the microcosms used in
this study (small, 6-liter volume with
naturally derived communities) were
sensitive to moderately low concentra-
tions of cadmium (100 ppb). The lowest
concentrations, however, caused no
response in the static system (1 and 10
ppb Cd) and a possible but inconclusive
response in the flowthrough systems
(10 ppb Cd). In contrast, others have
found significant ecological responses
to low levels of cadmium and copper in
relatively large, ecologically complex,
outdoor microcosms. This suggests a
possible direct relationship between
outdoor microcosms. This suggests a
possible direct relationship between
microcosm size (or complexity) and
toxicant sensitivity, but the relationship
is not clear. Conversely, it has been
suggested that the most sensitive sys-
tems (i.e., least resistant to perturbation)
are relatively low in "functional com-
plexity." Until some empirical means is
found to evaluate functional complexity,
however, this problem will be difficult to
resolve. It is also possible that physical
or chemical properties (e.g., pH or water
hardness) of the various microcosms
are related to their sensitivities. In any
event, our results suggest that small
laboratory microcosms are potentially
useful for estimating gross ecological
effects of toxic substances, perhaps as
an early phase in multiple-stage testing
followed by later but more selective
studies in more complex systems.
A Hierarchical Approach
to Toxicity Screening
Based on results from this and other
studies, we suggest that a potentially
useful screening protocol for aquatic
ecosystems might consist of a series of
factorial experiments in aquatic micro-
cosms of increasing complexity: 1) rela-
tively simple static microcosms (with
and without sediments), 2) flowthrough
microcosms (with and without sedi-
ments), and 3) detailed but selective
studies in microcosm subsamples from
specific ecosystems. The factors to be
included in each experiment will have to
be determined on a case-by-case basis
depending on available information
concerning properties and expected
distribution of each chemical. Measure-
ments in each experiment should in-
clude, but not necessarily be limited to,
community metabolism, community
composition (relative abundance of
major trophic groups), and dynamics of
essential elements (metabolism of N, P,
S, etc., and input-output relationships in
flowthrough systems). The inclusion of
sediments might require additional
measurements (e.g., redox potential,
microbial activity, sediment character-
istics).
The advantage of such a hierarchical
approach is that each step yields in-
creasingly more information and serves
as a guide for subsequent experiments.
In addition, each step more closely
approaches the real world.
The complete report, entitled "Microcosms as Test Systems for the Ecological
P. F. Hendrix, C. L Langner, f. P. Odum, and C. L Thomas are with the Institute
of Ecology, University of Georgia, Athens, GA 30602.
D. L. Brockway is the EPA Project Officer (see below}.
The complete report, entitled "Microcosms as Test systems for the Ecological
Effects of Toxic Substances: An Appraisal with Cadmium," (Order-No.
PB 81-209 595; Cost: $15.50, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield. VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Research Laboratory
U.S. Environmental Protection Agency
Athens, GA 30613
-757-OU/7Z17
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
Official Business
Penalty for Private Use $300
U5S ENVIR2PROTECTION AGENCY
230XS DEARBORN STREET
CHICAGO IL 60604
------- |