United States
Environmental Protection
Agency
Environmental Research
Laboratory
Gulf Breeze FL 32561
Research and Development
EPA/600/S3-87/016 July I987
Project Summary
Field Validation of Multi-
Species Laboratory Test
Systems for Estuarine Benthic
Communities
Robert J. Diaz, Mark Luckenbach, Sandra Thornton, Morris H. Roberts, Jr.
Robert J. Livingston, Christopher C. Koenig, Gary L. Ray, and Loretta E.
Wolfe
The major objective of this project
was to determine the validity of using
multi-species laboratory systems to
evaluate the response of estuarine
benthic communities to an introduced
stress. Over a 5-year period, experi-
ments in Apalachicola Bay, Florida, and
the York River, Virginia, sought to (1)
develop criteria for microcosm tests to
evaluate the capacity of microcosms to
model natural communities in the
presence and absence of pollution-
induced stress, and (2) assess the
validity of extrapolating test results
from one location to another. Individual
species response patterns in the micro-
cosms were highly variable and seldom
showed good agreement with patterns
in the field. Species richness in the
microcosms and field sites showed
good temporal agreement and provided
a conservative indicator of community
response to a toxic stress. An ecolog-
ically based guild approach to grouping
species proved to be a powerful and
reliable method of extrapolating from
microcosm test results to responses of
field communities.
This Project Summary was deve-
loped by EPA's Environmental
Research Laboratory, Gulf Breeze, FL,
to announce key findings of the
research project that is fully docu-
mented in a separate report of the same
title (see Project Report ordering
information at back).
Introduction
Objectives
The principal goal of environmental
toxicology is to understand pollution-
induced changes in ecological systems,
primarily in an effort to predict the
environmental consequences of a toxi-
cant. Recently, emphasis has been
placed on using multi-species laboratory
systems to evaluate the response of
aquatic ecosystems to an introduced
stress. The assumption has generally
been made that multi-species systems
provide more realistic estimates of the
effects of toxicants on complex natural
ecosystems than do single species tests.
A requisite part in the development of
a laboratory multi-species aquatic test
system is field verification. The authors
define field verification as the testing of
the capacity of specific laboratory test
systems to predict the responses of
ecosystems to toxicants. The process of
field verification raises methodological
considerations concerning criteria for
conducting tests and interpreting data.
An essential part of this process is
simultaneous investigation of commun-
ity dynamics in the laboratory test system
and in the natural communityfrom which
it was derived.
In a 5-year project designed to field
verify multi-species laboratory systems
(microcosms) for use with estuarine
benthic communities, four major issues
were addressed: (1) development of
criteria for conducting microcosm tests
-------
and for interpreting the data; (2) evalua-
tion of the capacity of a microcosm to
track natural field communities in the
absence of a toxicant; (3) comparison of
community response patterns of labor-
atory and field communities to a
pollution-induced stress; and (4) deter-
mination of the validity of extrapolating
from microcosm tests conducted in one
location to natural communities in
another. Focus has been on the infaunal
macroinvertebrate communities from
unvegetated, soft-sediment sites in
Florida (Apalachicola estuary) and Virgi-
nia (York River estuary). The approach
was to synoptically conduct field and
laboratory experiments at both locations
over different seasons.
Experimental Approach
The study sites in the Apalachicola Bay
system (East Bay and St. George Sound)
were located in polyhaline and oligoh-
aline areas, and those in the York River
were located in the meso-polyhaline
portion of the estuary. All sites were
shallow (1-2 m), unvegetated areas.
Sediments in the oligohaline sites were
silty sand, and in the polyhaline and
meso-polyhaline sites sediments were
predominantly fine sands.
Microcosms ranging from approxi-
mately 0.1 to 1.0 m were used to
evaluate the influence of microcosm size
on the system response. Microcosms
were constructed of a series of cores
collected with hand-operated box cores
(10 cm x 20 cm x 10 cm deep). Core
samples were placed in trays on a
seawater table in the same arrange-
ments as the original field orientations.
For the duration of the experiments,
microcosms were maintained in flow-
through seawater tables where condi-
tions of light, temperature and salinity
were similar to the field. Macroinverte-
brate samples in both the field and
laboratory were collected in random
sampling designs with coring devices (5
cm diameter, VIMS; 7.5 cm, FSU).
Samples were preserved, rinsed onto
500- and 250-//m-mesh sieves, and the
organisms identified to species and
enumerated.
Over a 5-year period seven field-
laboratory experiments were conducted
during spring and fall (Table 1). These
seasons represented periods of peak
biological activity. While the basic
protocol for the tests remained similar,
several different field and laboratory
treatments were employed in the various
tests. In all tests, replicated laboratory
microcosms maintained in flow-through
systems were sampled simultaneously
with field treatments located in the same
sites from which the microcosm com-
munities were derived. The various
treatments included predator exclusion
and inclusion cages, and both field and
laboratory dosing with unpolluted, hydro-
carbon polluted (from the Elizabeth River,
VA) and pentachlorophenol (PCP) pol-
luted sediments. Toxicant dosing proce-
dures in the field and laboratory were
investigated by testing both the total
volume and application method of
toxicant-laden sediments. Sampling
interval and duration were similar in a
field-laboratory tests, but an effort wa
made to determine the appropriat
sampling schedule to observe response
in the dosed experiments.
A major focus of the work was t
identify response variables that ade
quately reflect important community
level responses in both the field am
laboratory systems. Community-widi
descriptors (total number of individuals
number of species, species diversity an<
evenness), numerical abundances o
dominant species, and biomass ani
productivity estimates were all used a:
Table 1. Sampling Schedules for the Combined (FSU-VIMS) Experimental Program (1981
1985)
Weekly samples
A. FSU
1. oligohaline stations (11 /24/81-11/17/83)
2. polyhaline station (11 /2S/81-3/15/84)
B. VIMS
1. polyhaline marine lab station (10/13/79-12/18/83)
Microbiological data
A. FSU
1. oligohaline stations (fall 1982; spring 1983)
2. polyhaline stations (spring 1982)
B. VIMS
1. marine lab station (spring 1982)
Combined (field-laboratory) experiments
A. Spring 1982
1. Florida
2. Virginia
B. Fall 1982
1. Florida
2. Virginia
C. Spring 1983
1. Florida
2. Virginia
D. Fall 1983
1. Florida
2. Virginia
3. Treatments included:
a. Field controls
b. Field predator exclusion cages
c. Field predator inclusion cages
d. Microcosm controls
e. Field and lab treatments dosed with PCP
E. Spring 1984
1. Virginia only
F. Spring 1985
1. Florida (station ML)
2. Virginia
3. Treatments included:
a. field controls
b. microcosm controls
c. replicate lab and field treatments dosed with PCP
d. azoic sediments
G. Fall 1985
1. Florida (station ML)
2. Virginia
3. Treatments as in F. 3.
-------
measures of community response. A
more fruitful approach was to categorize
individual species into guilds based upon
ecological similarities and to treat these
groups as ecological units in assessing
response to toxicant stress.
Project Results
The laboratory-field experiments
coupled with the weekly monitoring data
permitted us to address each of the
objects stated above. Criteria for con-
ducting microcosm toxicity tests are
outlined below and detailed in the final
report. Definition of ecologically relevant
functional groupings and the classifica-
tion of species into guilds provided the
basis for successful comparisons
between laboratory and field commun-
ities and between the Florida and Virginia
test systems.
Comparisons of different microcosm
sizes suggested that small benthic
microcosms (approximately 0.1 m2)
provide good laboratory systems. These
small microcosms had similar commun-
ity patterns to larger (0.8 - 1.0 m2)
microcosms and are easier to construct
and handle. Sampling of small micro-
cosms is a destructive process and each
replicate microcosm may be sampled at
only one time, therefore requiring a large
number of replicate microcosms.
Experimental durations of up to 6
weeks and sampling intervals of 1-2
weeks were used in the various field-
laboratory experiments. In one experi-
ment, samples were taken within 24
hours after dosing. Experience suggests
that early sampling after dosing (within
24-48 hours) followed by increasing
intervals up to 5 weeks provided a good
sampling regime. Beyond 5 weeks micro-
cosms may begin to experience changes
in sediment geochemistry that cause the
laboratory to diverge greatly from the
field.
The addition and monitoring of toxic
substances are critical steps in any
laboratory microcosm test syrtem. The
experiments suggest that a good dosing
procedure was to apply a 1 -cm thick layer
of toxicant-laden sediments to the
surface. The authors found that this
same amount of uncontaminated sedi-
ments did not have noticable effects upon
the community. Dosing with less sedi-
ment (1 mm thick) and the same toxicant
load proved to be less effective. In tests
using benthic microcosms derived from
other habitats, it is important to include
a treatment that adds uncontaminated
sediments.
The microbiotic component in a labor-
atory microcosm is highly variable, and
its capacity to predict natural trends
depends on a combination of habitat
characteristics in the area of origin.
Microbial communities in microcosms
deviate progressively with time from field
associations and extended equilibration
periods are ill advised. In microcosms of
sediments from polyhaline areas,
microbes did not follow field conditions
as closely as those in microcosms from
oligohaline portions of the estuary.
Without detailed knowledge of microbial
ecology in the source area an interpre-
tation of results from laboratory micro-
cosms could be misleading.
Comparisons of field and laboratory
community dynamics of infaunal
macroinvertebrates revealed much vari-
ation between experiments and loca-
tions, but some generalization emerged.
The population dynamics of many dom-
inant species in both the Apalachicola
Bay and the York River estuary were
similar in the microcosms and the field.
This generalization was qualified by
finding that some species occasionally
underwent rapid population blooms in
the laboratory microcosms. For instance,
in Florida experiments the polychaete
Mediomastus ambiseta sometimes exhi-
bited large population increases in the
laboratory relative to the field. A similar
response was observed in Virginia for the
oligochaete Paranais littoralis. These
population increases are related to the
ability of these organisms to reproduce
in the microcosm where new individuals
survive better than in the natural field
site. This pattern has been experimen-
tally observed in field cage treatments
for M. ambiseta. The major population
fluctuations in both the field and the
laboratory (for undosed treatments) were
associated with recruitment events.
Since recruitment intensities for most
species differ between the laboratory and
the field, recruitment events may result
in substantial differences between field
and laboratory populations. Moreover,
the year-to-year variability in the timing
and intensity of recruitment for any given
species introduces a stochastic element
into microcosm testing when single-
species fluctuations are emphasized.
Using community-level parameters to
describe field and laboratory system
responses avoids some of the variability
associated with individual species fluc-
tuations. Species richness provided a
good descriptor of the macroinvertebrate
communities (Figure 1). In undosed
treatments, species numbers in the field
and the laboratory were often similar and
very conservative. In dosing experiments
with both hydrocarbon contaminated
sediment and PCP-laden sediment, the
species richness component showed
similar negative responses in both the
field and laboratory communities (Figure
1). By contrast total numbers of Individ
uals in the system fluctuated widely,
largely as a result of recruitment events,
and were not particularly responsive to
toxicant treatments. Species diversity
and evenness measures reflect combi-
nations of these two components and
were variable in their correspondence
between the laboratory and field sites.
Species richness is an important com-
ponent of natural systems which is
modeled well in aquatic microcosms and
proved to be a sensitive community-level
response to stress by a toxicant.
Grouping species into guilds based
upon functional groups according to
trophic, mobility, and dispersal modes
proved to be a powerful approach for
interpreting community responses. This
approach served two purposes. First, it
permitted identifying those guilds of
organisms for which laboratory micro-
cosm populations do not serve as good
analogs of natural populations in the
absence of any toxicant (e.g., those
brooding or asexually reproducing spe-
cies which have capability of blooming
within the microcosm). These types of
organisms may be excluded a priori horn
analyses to assess toxic effects. The
second advantage is that identifying
types of organisms that act as similar
ecological units facilities comparisons
between microcosms and field sites from
different locations. For instance, while
the species composition varies between
the Virginia and Florida sites, function-
ally similar ecological groups are found
in both sites and provide a basis for
comparison. Figure 2 shows summary
examples of this approach for one guild
which was modeled well in the laboratory
and one which was not.
Comparisons between field-laboratory
experiments in Florida and Virginia
identified both similarities and differen-
ces. The most notable difference is that
major recruitment periods at each loca-
tion occur in different seasons. The major
recruitment period in the York River
estuary is in the spring with only a minor
fall recruitment; the pattern in the
Apalachicola Bay is temporally reversed.
Since, as noted above, recruitment
events play a major role in the population
-------
A. Florida
B. Virginia
30
20
1
if
<0
.5>
10
14
12
10
c
•c
•$
u
Field Control
0135
Time Since Initiation (Weeks)
01359
Time Since Initiation (Weeks)
Figure 1.
Field High Dose
Lab Control
Lab High Dose
Species richness in field/laboratory PCP-dosed treatments from fall 1985 in Apalachicola Bay, Florida and the York River. Virginia.
fluctuations observed, this becomes an
important issue when attempting to infer
the responses of natural communities to
a toxic stress in one location based upon
microcosm experiments located in
another.
Conclusions
The authors conclude that properly
constructed and replicated multi-species
laboratory test systems with estuarine
macrobenthic invertebrates can serve as
effective tools for predicting responses
to toxicant stress. Several caveats apply.
Variability in natural estuarine systems
is high, necessitating large numbers of
experimental replicates and samples to
observe even major responses. Species
richness measures provide a conserva-
tive indicator of community response to
pollution-induced stress which is not
subject to much of the variation observed
for abundance measures. However, this
measure may also gloss over much of
the ecologically relevant response to the
stress. The population dynamics of
individual species within laboratory
microcosms are too variable to provide
adequate models of field populations, but
grouping species into ecologically similar
guilds alleviates much of this problem.
The detailed ecological data required to
construct these groups may be difficult
to obtain for many species.
The authors emphasize the importance
of good ecological characterization of the
habitats from which the microcosms are
derived and the habitats about which
inferences are to be made. When con-
siderations of different recruitment
seasons are taken into account, similar
community responses to a toxicant are
found in both the Florida and Virginia
experiments. Experiments with PCP
dosing conducted in Florida during the
spring of 1985 had a similar response
(in species richness) to experiments in
the fall of 1985 in Virginia, and the fall
experiments in Florida resembled those
from the spring in Virginia. At each
location experiments conducted during
the peak reproductive seasons resulted
in blooms of single species in the
microcosm. A functional guild approach
to analyzing community response patt-
erns enhances the ability to make
predictions concerning toxic responses
in one site based upon laboratory exper-
iments conducted at another site.
It was concluded that laboratory micro-
cosms can provide a valuable tool for
assessing natural benthic community
responses to introduced toxicants, pro-
vided that the caveats and conditions
described above are heeded. The authors
recommend using microcosms to provide
realistic estimations of field effects.
Recommendations
The results of the field-laboratory
comparison experiments indicate that
multi-species laboratory aquatic micro-
cosms may yield valuable information
regarding the responses of natural
communities to pollution-induced stress.
However, several very important cau-
tions are offered for conducting and
interpreting microcosm toxicity tests and
in extending the findings to natural
systems:
4
-------
Guild: Deposit-Feeder, DETRIV/OMNIV, Mobile Borrower,
Limited Dispersal.
Apalachicola Bay, FL
Guild. Deposit-Feeder, DETRIV/OMNIV, Mobile Burrower,
Wide Dispersal
York River, VA
300
200
§
I
I
1
roo
\
i
n>
50
40
30
20
10
123456789
Time Since Initiation (Weeks)
0
01234569
Time Since Initiation (Weeks)
Field
Lab
Figure 2. Comparison of temporal patterns for 2 guilds in the laboratory and the field. Data are composites of the control treatments in
all tests. Data for the guild with limited dispersal (for which only Florida data are shown) reveal that although lab and field
abundances track one another well initially, individuals in this group may undergo population blooms in the lab The guild with
wide dispersal (Virginia data shown) shows a consistent pattern through the first 5 weeks with some divergence between lab
and field thereafter.
1. Close attention must be paid to
physio-chemical characteristics of
microcosms and it is important that
these lie within realistic ranges for
field values at the time the exper-
iment is conducted.
2. Monitoring of toxicant levels and
distribution within the microcosms
throughout the experiment is
necessary to evaluate dissipation
and breakdown of toxicants.
3. The high spatial variability inherent
in benthic communities necessi-
tates that sufficient replicates be
employed.
4. The temporal variation in recruit-
ment adds a nearly random com-
ponent to the community response
in microcosm tests from year to
year and site to site. To overcome
this problem it is mandatory that
microcosm tests, while being prop-
erly timed to correspond with
biologically important reproductive
seasons, identify and exclude spe-
cies with aberrant recruitment
patterns in the laboratory test
system.
5. Successful extension of toxicity
test results from laboratory micro-
cosms to the field sites from which
they were derived and beyond to
other sites requires detailed knowl-
edge of the systems. Just as impor-
tant as an understanding of the
physical conditions of the habitats
is a good knowledge of the repro-
ductive seasons and modes, tro-
phic types, and life history charac-
teristics of the component fauna.
Since species composition will vary
between sites it is necessary to
characterize the response of differ-
ent "ecological types" (guilds) m
microcosms if the results are to be
useful
------- |