FIELD VALIDATION OF LABORATORY-DERIVED
MULTISPECIES AQUATIC TEST SYSTEMS
by
Robert J. Livingston*
Robert J. Diaz**
David C. White*
*Department of Biological Science
The Florida State University
Tallahassee, Florida 32306
**Virginia Institute of Marine Science
College of William and Mary
Gloucester Point, Virginia 23062
CR810292-01-0
Project Officer
Dr. Thomas W. Dukej
Office of the Director
Environmental Research Laboratory
Gulf Breeze, Florida 32561
ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
GULF BREEZE, FLORIDA 32561
-------
DISCLAIMER
The information in this document has been funded wholly or in part by
the United States Environmental Protection Agency under assistance
agreement number CR810292-01-0 to The Florida State University. It has
been subject to the agency's peer and administrative review, and it has
been approved for publication as an EPA document.
OCT I 51991
-------
ABSTRACT
A three-year study was carried out to determine the feasibility of
using multispecies microcosms of benthic microorganisms and infaunal
macroinvertebrates to predict the responses of estuarine systems to toxic
substances. Criteria were developed to evaluate the field validation of
laboratory microcosms. Simultaneous laboratory/field experiments were
carried out in the Apalachicola Bay system, Fla., and York River estuary,
Va., to test the potential for extrapolation of validation results from one
ecological system to another. The study demonstrated that microcosms of
microorganisms and infaunal macroinvertebrates can be established for short
periods (5-6 weeks) and that the microcosms can be used to simulate speci-
fic features of field assemblages within the range of uncertainty that is
characteristic of natural systems. Moreover, validation results can be
extrapolated from one system to another as long as the systems share common
habitat features and dominance relationships of important populations.
Water quality in the microcosms essentially paralleled that in the
field, although variation of certain water features and sediment charac-
teristics was noted. These laboratory artifacts were apparently caused by
the isolation of the microcosms from natural phenomena of the estuarine
environment that were not replicable in the laboratory. Physical habitat
features and biological responses in the respective study areas were extre-
mely complex and highly variable in space and time. Factors such as water
and sediment quality, predator-prey relationships, recruitment, and domi-
nance relationships among infaunal populations influenced the community
structure of benthic organisms in the laboratory and the field. However,
the relative influence of physical and biological factors varied con-
siderably between habitats and through time. Consequently, the extent to
which the microcosms paralleled field conditions depended to a considerable
degree on the time of testing and dominance/recruitment features of the
system in the source area.
Specific populations of microorganisms and infaunal macroinvertebrates
in the microcosms sometimes differed from field populations. These dif-
ferences were often associated with known artifacts in the laboratory and
could be qualified as such. The establishment of the microcosm was a cru-
cial factor in the continuity and success of the microcosms.
Microorganisms were particularly sensitive to disturbance during the
establishment of the microcosms. Microbes in microcosms taken from polyha-
line areas showed considerable divergence from field associations within
two weeks after the intiation of the experiment. However, in microcosms
taken from highly variable (physically dominated) oligohaline portions of
the estuary, the microbial associations paralleled field conditions rather
well over a six-week period. Microcosms of infaunal macroinvertebrates
differed from field conditions depending on the timing of the experiment
iii
-------
and recruitment features of the subject populations. Thus, the predictive
capacity of the laboratory microcosm depended to a considerable degree on
habitat characteristics and recruitment trends in the source area at the
initiation of the microcosm.
Microcosms of benthic organisms, although sometimes characterized by
changes in individual populations, provided a broad spectrum of biological
information for use in evaluating the impact of toxic wastes. Community
indices such as species richness approximated field conditions quite
closely. Criteria developed during this study can be used to determine the
applicability of laboratory microcosms to toxicity tests.
iv
-------
CONTENTS
Abstract . . ±±±
Figures vi
Tables viii
Foreword. ^x
Actcnowledgment x
1. Introduction 1
2. Conclusions 3
3. Recommendations 5
4. Materials and Methods 6
5. Results and Discussion 11
Experimental Program: Florida State University .... 11
Infaunal Macroinvertebrates: Background 11
Field Experiments 12
Laboratory Microcosms 15
Microbiology 22
Summary of Findings 28
Experimental Program: Virginia Institute of Marine
Science 29
Infaunal Macorinvertebrates: Background ..... 29
Laboratory-field Experiments 31
Test Comparisons 37
Summary of Findings 39
Preliminary Toxicology 41
6. Criteria for Verification Procedures 43
References 49
Appendices
A. Field operations for the verification project including a
list of physical/chemical and biological sampling parameters
taken weekly at the respective studies sites (FSU,
oligohaline, polyhaline; VIMS, polyhaline) 51
B. Generalized protocol for laboratory microcosm/field-semi-
field validation studies 52
C. Sampling schedules for the combined (FSU-VIMS) experimental
program (1981-1984) 54
Glossary 55
-------
FIGURES
Number
1 Chart showing study sites in the Apalachicola Bay system
(Florida)
2 Lower Chesapeake Bay showing the location of the polyhaline site
in the lower York River (Virginia) 8
3 Diagram of a typical field array showing placement of cages
(inclusion/exclusion cages), cage controls, and full-field
sampling areas 10
4 Numerical abundance and species richness of infaunal macroinver-
tebrates taken weekly at station 3 and ML.in the Apalachicola
Bay system over the study period. All numbers represent the
totals for 10-12-cores samples 13
5 Experimental results in polyhaline area (station ML) showing
long-term field changes and short-term (during Spring 1982,
experiment) treatment effects involving changes in numbers of
individuals, numbers of Mediomastus, species richness, and
diversity of infaunal macroinvertebrate assemblages. All
numbers represent totals for 12-core samples 14
6 Experimental results in the oligohaline area (station 3) showing
changes in numerical abundance, diversity, and top dominant
(500- and 250-^m sieve fractions) in various treatments during
Fall 1982. All numbers represent totals for 10-core samples. . 16
7 Experimental results in the oligohaline area (station 3) showing
changes in numerical abundance, species richness, and a top
dominant (500- and 250-^jm sieve fractions) in various treatments
during Spring 1983. All numbers represent totals for 10-core
samples 17
8 Comparison of weekly field collections (field weeklies) with
laboratory microcosms (controls, predator inclusions, sediment
disturbance) with organisms taken from station ML during Fall
1983 (500-ym sieve). All numbers represent totals for 12-core
samples 19
VI
-------
Number Page
9 Comparison of weekly field collections (field weeklies) with
laboratory microcosms (controls, predator inclusions, sediment
disturbance) with organisms taken from station ML during Fall
1983 (250-ym sieve). All numbers represent totals for 12-core
samples ............................ 20
10 Cluster analysis of changes in species composition by treatment
(summed cage results) for field and laboratory data over a
six-week period in tests with polyhaline macroinvertebrates
during spring, 1982. Also shown is the cluster analysis by
treatment for all replicates summed over the experimental
periods (TI-TS) ........................ 21
11 Numerical abundance (numbers per core sample) of macroinver-
tebrates on 250-ym sieves taken weekly at the study sites
in the Apalachicola system from November 1981 through
December 1983. All numbers represent totals for 10-12-core
samples ............................ 23
12 Distribution of field and microcosm derived samples of polyhaline
(station ML) sediments along a single discriminant function
after two weeks. Each centroid represents at least 144 data
points (12 variables by 12 replicates) ............ 25
13 Distribution of field- and microcosm-derived samples of
oligohaline (station 3) sediments along a single discriminant
function after three weeks .................. 26
14 Discrimination of oligohaline (station 3) sediment samples from
the field, regularly sampled microcosms, and undisturbed
microcosms after six weeks .................. 27
15 Long-term population variation of dominant annelids at the York
River, Virginia, site ..................... 30
16 Patterns of Paranais littoralis abundance for Test 1 ...... 33
17 Patterns of Streblospio benedicti abundance for Test 2 ..... 36
18 Cluster analysis of Spring 1983, Test 3 York River, Virginia,
site ............................. 38
VII
-------
TABLES
Number Page
1 Dominant species (individuals/m^) from the field validation
experiments, York River, VA 32
2 Community structure indices from the field validation experiments
(York River, VA) 35
3 Criteria for review of the validation of infaunal macroinvertebrate
microcosms with semi-field mesocosms and full-field conditions
in estuarine systems 44
viii
-------
FOREWORD
The primary aim of this project was to carry out a series of experi-
ments with multispecies microcosms of estuarine microbes and macroinver-
tebrates to develop criteria for the verification of laboratory test
results relative to the field. The project was conducted in two completely
different areas (i.e., the Apalachicola estuary in the northeast Gulf of
Mexico and the York River estuary in the Chesapeake Bay system) to develop
the data necessary for the extrapolation of results from one estuary to
another. This work was undertaken by three separate research groups under
the direction of Robert J. Livingston (Florida State University), Robert J.
Diaz (Virginia Institute of Marine Science), and David C. White (Florida
State University). The first year of the project (1982-1983) was devoted
to the development of experimental protocols. Experiments were carried out
during the subsequent two years to develop the criteria for the verifica-
tion process. These criteria will be employed to evaluate the usefulness
of multispecies laboratory tests in predicting the response of estuarine
organisms to toxic substances under field conditions.
Henry^T. Enos
Director
Environmental Research Laboratory
Gulf Breeze, Florida
IX
-------
ACKNOWLEDGMENT
A number of people were associated with the Florida State University
effort. Various undergraduate and graduate students aided in the collec-
tion and analysis of the data. D. A. Meeter, G. C. Woodsum, and L. E.
Wolfe were responsible for computer operations and statistical analyses.
C. C. Koenig and D. Cairns ran the marine laboratory operations. M.
Kuperberg was in charge of overall coordination of the project, while
R. Howell and W. Greening were responsible for field collections.
Taxonomic experts on the project included K. Smith (oligochaetes), G. Ray
(polychaetes, benthic macroinvertebrates), W. Clements (benthic macroinver-
tebrates), and Bruce Mahoney (polychaetes). K. Burton and M. Hollingsworth
were involved in the handling of the field samples. T. W. Federle set up
the microbiological analyses. R. F. Marz aided in the photomicrography of
the microbial associations. Other support for the microbial studies was
provided by J. Nickels.
For the Virginia Institute of Marine Sciences, we thank Sandra
Thornton for conducting the laboratory and field operations and contri-
buting to the preparation of the manuscript. We also thank Eric Koepfler,
Dave Stilwell, Charlie Strobel, Ruth Williams, Betty Bieri, Erik Zobrist,
and Chip Neikirk for their field and laboratory assistance.
We are grateful for a series of excellent reviews of this manuscript
by different colleagues.
-------
SECTION 1
INTRODUCTION
The basic question that underlies the considerable effort to
understand pollution-induced changes in aquatic systems is well
established: what is required to predict the environmental effects of a
toxicant or stimulatory substance on a given ecological system? With the
recent development of sophisticated toxicological methods to evaluate acute
and chronic effects of toxicants on laboratory populations, the question
then becomes: what is required to establish a reliable measure of the
capability of specific laboratory test systems to predict actual environ-
mental effects of a given toxic agent?
We define the process of field validation as the testing of the capa-
city of specific laboratory test systems to predict the environmental
responses of natural ecosystems, or portions thereof, to toxicants. Once a
test system is validated, it provides a means of generating toxicological
data that can be realistically correlated with expected field impacts. The
process of validation necessitates two pursuits: selection of a particular
test system and acquisition of knowledge about the natural variation and
dynamics of field populations from which the test system is derived.
Without knowledge of ecosystem structure and function, it is practically
impossible to evaluate toxic effects.
There is a need to develop and use multispecies laboratory tests to
evaluate the environmental implications of toxic waste disposal (Hammons et
al., 1981). However, because of basic differences between laboratory and
field conditions, the prediction of effects, in the field, based on labora-
tory results, is extremely complex and not well understood. Cairns et al.
(1981) provided a thorough background for the need to verify laboratory
test systems in the field; these authors have already provided procedures
for determining the response of multispecies test systems to perturbation.
A detailed review of the problems related to field verification of labora-
tory results is given by Livingston and Meeter (in press). Realistic simu-
lation of field conditions and generality of test results from ecosystem to
ecosystem require an identification of those aspects of the laboratory
microcosm that "most accurately represent the natural prototype" (Hammons
et al., 1981). In addition to the laboratory-to-field relationship, the
question of generality of findings (i.e., location-to-location extrapola-
tion of results) is a significant part of the verification process.
The focus of our three-year project has been microbial and infaunal
macroinvertebrate communities of unvegetated soft sediments of shallow
estuaries in Florida (Apalachicola Bay system) and Virginia (York River
1
-------
estuary). Natural soft-bottom macroinvertebrate communities were selected
for the test systems because of their trophic importance and the close
association of such organisms with sediments, which mediate the fate of
most toxicants. Also, there is a considerable body of literature con-
cerning the response of the benthos to disturbances, and this information
will facilitate the validation process. In terms of the ability to
replicate, realism, generality, and potential application to protocol deve-
lopment, sediment cores as multispecies microcosms represent one of the
most potentially useful methods in intermediate and advanced stages of
hazardous waste assessment (Hammons et al., 1981).
Based on extensive background work in the respective estuaries by
Florida State University and the Virginia Institute of Marine Science, a
project was designed to answer the following questions:
1. What are the criteria for laboratory-to-field verification of
multispecies test systems based on communities of estuarine micro-
bes and macroinvertebrates?
2. What are the predictive capabilities of laboratory multispecies
microcosms relative to natural processes under semi-field and
full-field conditions?
3. Can results be extrapolated from one estuary to another?
Our principal objectives (a) were to evaluate the capacity of the
laboratory test as a realistic analog or simulation of the natural com-
munity from which it was derived and (b) to develop criteria for field
verification of laboratory results. The evaluation considered validation
at three levels: physico-chemical differences, differences in population
and community structure, and functional differences between full-field and
semi-field treatments and laboratory microcosms. Results of such tests are
being applied to current experiments that concern the predictive capability
of microcosms exposed to toxic substances. Concurrently, a complete review
is being undertaken to determine the potential for extrapolation of valida-
tion results from one location to another.
-------
SECTION 2
CONCLUSIONS
A three-year study was conducted to determine the capability of speci-
fic laboratory test systems to predict environmental (field) effects of
toxic substances. Multispecies microcosms of infaunal macroinvertebrates
and microorganisms were used with simultaneous experiments in the
Apalachicola Bay, Florida, system and the York River, Virginia, estuary to
test for the feasibility of extrapolation of validation results from one
ecological system to another.
Field conditions in the study areas were characterized by short-term
disturbances (i.e., wind and tidal currents) and seasonal changes in the
physical environment. The microcosms followed various physical aspects of
the field habitat rather closely. However, storm-induced disturbances were
not replicated and current regimes in the field were not simulated in the
laboratory. Despite slightly increased accumulation of silt under labora-
tory conditions relative to the field, no significant changes were noted in
various sediment properties among laboratory and field treatments.
Biological interactions in the field were complex and highly variable
in space and time. Physico-chemical habitat changes, predation, and
recruitment influenced the macroinvertebrate assemblages with differential
effects exerted along habitat gradients and during different seasons of
the year. Changes in the macroinvertebrate assemblages in the microcosms
were due, in part, to alterations during transfer from field to laboratory,
lack of motile predators in the laboratory, and altered recruitment. Such
changes appeared to depend on the timing of the test and the natural
assemblages of macroinvertebrates in the source areas at the initiation of
the microcosm.
Experiments carried out in two different estuaries showed that the
basic controlling features and microcosm response relative to the field
were quite similar. The initial establishment of the microcosm and time-
based alteration of recruitment in the laboratory microcosms were the most
important elements contributing to changes in the microcosms relative to
field conditions. The timing of the test, relative to seasonal changes in
recruitment, was also an important aspect of the validation process. Thus,
correct interpretation of microcosm results relative to field processes
depends on an understanding of natural community processes. No single spe-
cies in the laboratory was consistently representative of field conditions
either because of laboratory artifacts or because of specific responses of
individual populations to laboratory conditions.
-------
The laboratory microcosm approach has considerable potential for eva-
luating microbial or macrobiological responses to natural disturbances or
toxic effects in the field. Multispecies microcosms have the advantage of
incorporating various forms of community level information into the experi-
mental design; such information is not available in single-species tests.
However, because of the extremely complex relationships of such asso-
ciations, a thorough knowledge of the ecology of a given site is necessary
for a reasonable application of laboratory-to-field or field-to-field
extrapolations. For example, the microbial community in estuarine areas
was particularly sensitive to both physical and biological sources of
disturbance. In polyhaline areas, a relatively low variance was noted in
field assemblages of microbes. Microcosms from polyhaline areas showed
significant differences in the microbial associations after two weeks, and
such differences between the laboratory and field increased substantially
between week 2 and week 6 of the experiment. However, in physically driven
oligohaline portions of the estuary, differences between laboratory and
field assemblages were small and did not increase with time. Thus,
although the microbial component of a laboratory microcosm was highly
variable, the capacity of such a microcosm to predict natural trends
depended on the habitat characteristics in the area of origin. It appeared
that microcosms taken from areas subjected to continuous natural physical
disturbances were better indicators of natural phenomena than were those
taken from more physically stable systems.
Our experimental results demonstrated that microcosms of soft-sediment
macroinvertebrates can be established for short periods (5-6 weeks) and
that changes in the field populations can be either reflected in the
overall response of the microcosms or accounted for in terms of specific
laboratory artifacts. Moreover, extrapolation of such results from one
system to another is possible within the range of uncertainty that is
characteristic of natural systems. Just as extrapolation of results from
the microcosm to the field cannot, by definition, be a direct process, so
too is extrapolation from one ecosystem to another seriously qualified by
funtional differences in community processes of such systems. With ade-
quate qualification based on ecological knowledge of the areas in question,
both verification and extrapolation are feasible within the limits of
natural variation.
-------
SECTION 3
RECOMMENDATIONS
The strength of the validation of a given microcosm depends on an
assessment of the laboratory reaction of populations of individual species
within the uncertainty that is natural to ecological systems. It is recom-
mended that validation processes be evaluated according to criteria deve-
loped by our studies. Further analysis is needed to relate how well
microcosms reflect the response of natural ecosystems to toxicants. The
validation approach proposed by our research reflects the need to calibrate
laboratory microcosms with established processes in the field. More work
is needed to develop validation procedures for processes in natural com-
munities in addition to structural aspects of the estuarine communities
that have been emphasized in this research.
-------
SECTION 4
MATERIALS AND METHODS
All field and laboratory operations in the respective study areas
followed standardized methods. Aside from certain differences inherent in
the two study sites, experimental procedures were carried out in a com-
parable manner. Detailed accounts of methods and material used in this
project are given by Livingston et al. (1893) and Diaz et al. (1984).
Prior to the initiation of the project, all background field data from
the study areas were updated and evaluated to establish a preliminary pro-
tocol for the full-laboratory, semi-field, and full-field treatments.
Based on preliminary analyses of background data, the spatial limits and
frequency and location of sampling were determined. A brief review of
habitat parameters, biological characteristics, and methods of sampling
appears in Appendix A.
The study sites in the Apalachicola Bay system (East Bay and St.
George Sound) were shallow (1-2 meters (m)), unvegetated soft-bottom areas
located in oligohaline (stations 3, 5A) and polyhaline (station ML) areas
(Figure 1). Sediments in the oligohaline areas were silty sand, whereas
sediments in the polyhaline zone were largely fine sands (1-2% silt-clay).
The York River study site (Figure 2) was a shallow (1.5 m) , unvegetated
soft bottom located in the meso-polyhaline portion of the estuary.
Microcosms were constructed of a series of cores collected with hand-
operated box corers (10 x 20 centimeters (cm); 10 cm deep). Core samples
were placed in trays on sea-water tables in the same arrangement as the
original field orientation of the cores. The size of each microcosm was
0.8 to 1.0 square meters (m^). Light, temperature, and salinity regimes
followed field conditions. Synoptic biological sampling of microcosms and
field was done randomly with coring devices (5 cm, VIMS; 7.5 cm, FSU).
Sieves of mesh sizes 250 and 500 micrometers (urn) were used for the
infaunal macroinvertebrates. Microbial samples were taken from field areas
and laboratory microcosms with a 3.2-cm-diameter corer and analyzed for
lipids and fatty acids.
Four field-laboratory experiments were carried out over a 2-year
period. The tests were conducted during spring and fall periods of peak
biological activity and change in the respective study sites. Although
some changes were made to the sampling program over the study period, a
basic protocol was developed and followed for experiments at both sites
(Appendix B). The approach was to sample replicated flow-through
laboratory microcosms (0.8-1.0 m^) derived from natural soft-sediment
-------
09
c
n
n>
CO
rr
a.
ft)
co
Pi
n
o
M
W
W
Cu
9
xx
Tl
M
O
H
(-
O-
APALACHICOLA
RIVER
ST
VINCENT
ISLAND
KM.
-------
00
OQ
1-1
N)
"XI
o
fu
3
tt>
rt
(D
0
rt
H*
O
(0
to
1-1
H-
i-l
TO
H-
3
H-
10
-------
areas, simultaneously with field treatments (exclusion cages, inclusion
cages, cage controls) (Figure 3). A sampling schedule is given in
Appendix C.
Variables analyzed during the experimental series included numerical
abundance (total number of individuals and dominant populations), numbers
of species, and species diversity. All analyses were carried out with and
without logio(x+l) transformations. A nested ANOVA analysis to test for
differences between laboratory microcosms was carried out with 250- and
500-ym sieve fractions (macroinvertebrates) and microbial parameters. To
test the null hypothesis that there was no significant difference among
field and laboratory treatments with respect to the variables listed above,
selected ANOVA models were employed. A one-way ANOVA was run on all treat-
ments by sampling period. A randomized block repeated-measures ANOVA was
used with the field data with location as the blocking factor and time as
the repeated measure. Tukey's method of multiple comparisons was used to
test the differences between all possible pairs of means. Analyses of
qualitative changes in infaunal assemblages were carried out using "rho"
(Matusita, 1955; Van Belle and Ahmad, 1974) and Czekanowski (Bray-Curtis;
Bloom, 1981) similarity coefficients and the flexible grouping strategy
with beta = -0.25 (Lance and Williams, 1967).
-------
18 meters ( 59 ft)
FA = Screened Exclusion
FB= Screened Inclusion
FC = Control
FA
FB
FC
FC
FB
4m»-|-2m-|
FA
FA
FB
FC
T
6m
10 m
38m
(125ft)
Figure 3. Diagram showing placement of cages (inclusion/exclusion cages),
cage controls, and full-field sampling areas.
10
-------
SECTION 5
RESULTS AND DISCUSSION
EXPERIMENTAL PROGRAM: FLORIDA STATE UNIVERSITY
Infaunal Macroinvertebrates; Background
Preliminary experiments were carried out to determine scaling
parameters associated with the spatial distribution of infaunal macroinver-
tebrates. Data from these experiments were used to develop sampling
methods. Tests were carried out in oligohaline and polyhaline areas of the
Apalachicola Bay system. During October 1981, three sets of 100 core
samples were taken from the sampling platforms at each of the study areas.
Species accumulation (with and without rare species) and rarefaction
analyses indicated that 10 replicate cores per sample at stations 3 and 5A
and 12 replicate cores per sample at station ML would take at least 85% of
the available species. Oligohaline areas were characterized by moderately
high numerical abundance, high dominance of a few polychaetes (primarily
Mediomastus ambiseta and Streblospio benedicti), and low species richness
and diversity. The polyhaline station was characterized by relatively low
numerical abundance, low relative dominance (Paraprionospio pinnata and
Mediomastus ambiseta), and high species richness and diversity. Analysis
of variance of species richness and diversity and numbers of the two top
dominants indicated relatively low within-station differences among repli-
cates. Analyses of subsequent field experiments indicated statistically
significant differences (p < 0.05) of one or more biological indices in at
least one replicate in a given treatment. The results indicated that at
least three samples per treatment were necessary for a given sample.
Highly significant differences (p < 0.05) were noted between stations.
The oligohaline study area was characterized by low salinities that
were highly variable (mean at station 3, 4.6°/oo; standard deviation
(s.d.), 6.2) with peaks during summer-fall periods of low Apalachicola
River flow. The polyhaline area was characterized by uniformly higher and
less variable salinities (mean, 28.5; s.d., 3.5). Slight reductions of
salinity were noted in this area during late winter-spring periods.
Maximum water temperature occurred during July and August; winter lows were
noted in December and January. The oligohaline experimental site (station
3) was an unvegetated, soft-sediment area that was quite turbid and domi-
nated physically by tidal currents and Apalachicola River flow. Sediments
were composed of silty sands. The polyhaline experimental area (station
ML) was an unvegetated area dominated by sandy sediments and low to
moderate tidal currents. Thus these stations represented significantly
11
-------
different habitats in terms of major features such as salinity and sediment
composition.
Overall, seasonal changes in macroinvertebrate assemblages at the
oligohaline and polyhaline sites were somewhat similar although dominance
hierarchies remained distinctive.
During the study period, oligohaline dominants included Paranais
litoralis, M_. ambiseta, S_. benedicti, Cerapus sp., Hobsonia florida, and
Grandidierella bonnieroides. Polyhaline dominants were II. Ambiseta, tubi-
ficids, and 7_. pinnata. Community indices were seasonally and annually
variable at the oligohaline site, with peak numerical abundance during the
winter of 1981-82 (Figure 4). Species richness was high at the polyhaline
station, with seasonal peaks during winter-spring periods. Most of the
dominant species taken at the oligohaline site reached the highest numeri-
cal abundance during winter-spring periods. At the polyhaline site, there
were complex dominance cycles with winter-spring peaks for II. ambiseta and
spring-fall peaks for Brania wellfleetensis. Others, such as Aricidea
fauvelli, showed no distinct seasonal patterns. Winter recruitment pat-
terns were most common at both stations for species such as II. ambiseta,
Paranais littoralis, and immature tubificids. Peak abundance of the
meroplankton, dominated by bivalve and gastropod mollusks, occurred from
May through November.
Field Experiments
Sediments from the oligohaline and polyhaline sites tend to differ in
terms of the relative proportions of sand, silt, and clay. The polyhaline
site was characterized by sandy sediments with relatively small components
(1-2%) of silt and clay. Mean grain size ranged from 2 to 3 phi units. No
seasonal trends in sediment characteristics were noted at the polyhaline
site. Sediments in the oligohaline study area tended to have higher silt
and clay fractions (usually ranging from 40% to 80% sand, 13% to 48% silt,
and 3% to 10% clay). Average grain size was usually smaller (mean grain
size, 3-4 phi units). Seasonal trends were apparent in the oligohaline
system, with generally higher percentage of sand noted during spring
periods of high river flooding.
An analysis of sediment characteristics (mean grain size, standard
deviation, skewness, kurtosis; percent sand, silt, clay, and percent
organics) as a function of treatment-related effects for each of the four
experiments indicated no qualitative treatment effect on sediment com-
position. Analysis of variance for mean grain size and percent sand indi-
cated no significant (p < 0.05) treatment-related effects on sediment
composition of field samples of laboratory microcosms. No time-based dif-
ferences were noted in sediments taken from field treatments or laboratory
microcosms over the course of an individual experiment.
Field interactions are complex and highly variable in spatial as well
as temporal terms. Benthic macroinvertebrate assemblages are influenced by
complex series of physical, chemical, and biological stimuli and processes.
In polyhaline areas during the spring of 1982, exclusion of predators led
12
-------
OQ
C
i-t
ID
t W rt 53
o ft> fD c
i ^ cr a
H- I-t ft)
NJ CO ft) I
I
own)
o
rt p.
.. . o
rt co (u
«g «H
ft) ft)
CO O !T O"
ft) < (D C
0 (D 3 3
(D rt fD 3
CO 3* fD O
(D ?r CD
M
CO ^ P
rt 3
C cu d.
D- rt
^ co
co id
id rt CD
fD ft) O
1 ft H-
H- H- (D
O O CO
O- 3
O
cu p-
3 3
O. CD
co
S eo
cr-
(D
fl
CO
i-t
(D
"O
H
ro
CO
n>
3
rt
rt
O
rt
CO
O
H- l-h
3
H-
rt 3
fD ft)
0 O
O H-
(- 3
ft) <
fD
1
TOO
00-
UJ
O
CD
SOO-
400-
300-
200-
100-
STATION ML
( 900 xM St«v«t)
I I
NOV.
I I I I I I
FEB. MAY
III T I I
AUG. NOV
I I I I I I
FEB. MAY
AUG. NOV.
I I I 1
FEB. MAY
I I I I I I I I I
NOV. FEB MAY AUG.
NOV. FEd
I I
MAY
AUG. NOV.
I I I I
FEB. MAY
to
Ul
O
8) 30-
i
z
2O-
10-1
90-
40-
30-
20-
10-
0 '| | I | I I I I I I I I I I I I I I I I I I I I I I I I I I I
NOV. FEB. MAY AUG. NOV. FEB MAY AUG. NOV. FEB. MAY
TIME (Months)
I I I l l I I I I I l l l I l l l l l l l l l i l l I I l l i
NOV. FEB MAY AUG. NOV. FEB. MAY AUG. NOV. FEB MAY
TIME (Months)
-------
UJ
O 500
MARCH
Experimental
Period
SPRING 1982
Experimental
Period
APRIL MAY
TIME ( Months)
JUNE
FIELD WEEKLIES
EXCLUSION TREATMENT
INCLUSION TREATMENT
CAGE CONTROL
LAB MICROCOSMS
APRIL
TIME ( Monfhs)
Figure 5.
Experimental results in polyhaline area (station ML) showing long-term field changes and short*
term (Spring 1982) treatment effects. Totals are for 12-core samples.
-------
to increased numbers of the dominant polychaete Mediomastus ambiaeta
(Figure 5). Although species richness was not affected by predator
exclusion, species diversity and evenness were significantly reduced, in
large part because of the increased dominance of Mediomastus. Such effects
were corroborated by the predator-inclusion treatments, which were not
statistically different from the cage controls and uncaged areas. This
predation effect occurred during the usual spring decline of the macroin-
vertebrate community (Figure 5). Randomized block ANOVA tests showed
significant (p < 0.05) exclusion treatment effects for species diversity
indices and log numbers of Mediomastus and total numbers of individuals.
Time-based (week-to-week) effects were significant (p < 0.05) for all
variables except species richness and numbers of individuals. The T x W
interaction was significant, which indicated that different treatments had
diverging temporal trends. Less significant effects (p > 0.05) were noted
because of the disturbance of sampling. The main treatment-induced effects
were shown for organisms taken in the top two centimeters of sediment,
where the highest numbers of invertebrates are located. There were also
indications of treatment-related effects on recruitment of young organisms
in the field, with lower numbers of smaller organisms in the uncaged field
controls. Recruitment of the cosmopolitan Mediomastus was an important
factor in the relative abundance patterns of the macroinvertebrate
assemblages. This species was a major colonizer of azoic sediments during
the spring experimental period.
Subsequent experiments in oligohaline and polyhaline areas during dif-
ferent seasons of the year revealed no similar treatment-related effects in
the field. The oligohaline tests indicated no treatment-related effects
during fall or late winter-spring periods (Figures 6 and 7). These results
confirmed previous findings by Mahoney and Livingston (1982). Tests during
the spring indicated significant storm-induced effects on the macroinver-
tebrate assemblages although the effects were noted across all treatments.
The lack of a predation effect in the field was tentatively attributed to
the high dominance values, recruitment trends, and habitat variability in
oligohaline portions of the study area. Analysis of 6-week experiments in
the polyhaline area of study during the fall showed no treatment effects
concerning either the quantitative or the qualitative distribution of orga-
nisms over the study period. There was no clear evidence of significant
(p < 0.05) predation or sedimentation effects on the major population and
community indices of infaunal macroinvertebrates. Time-based differences
were noted for important species with high recruitment values (i.e.,
Mediomastus and immature tubificids).
Overall, results from the field tests indicated spatial and temporal
gradients of habitat (e.g., salinity) on predation effects, dominance
characteristics, and community indices. Predation was an important
controlling variable in high-salinity areas during spring periods of low
general recruitment and high predation pressure on infaunal species.
Laboratory Microcosms
The establishment of microcosms sometimes led to changes in the
laboratory assemblages relative to the field assemblages from which they
were derived. Such changes included a general reduction of numerical
15
-------
UJ
O 1600
I2OO-
(D IOOO
800J
< 600-
£ 4OO
2 200-
(O
tM SIEVES
£
§
*
£
£
Ib
Ul
i
i
1800-
1400-
I2OO-
1000-
800-
600-
4OO-
200-
0-
/
/
/
/
/
/
/
/
/
/
J
/
/
'
i i r i ii ^J
1234567
250-
200-
150
100-
50'
1
3
4
6
WEEK F'ELD WEEKLIES
EXCLUSION TREATMENT
INCLUSION TREATMENT
CAGE CONTROLS
LAB MICROCOSM
WEEK
Figure 6.
Experimental results in the oligohaline area (station 3) showing
changes in numerical abundance, diversity, and top dominant
(500- and 250-ym sieve fractions) in various treatments during
Fall 1982. All numbers represent totals for 10-core samples.
16
-------
O 400-
<. 350-
O
Z 300-
5 230-
O
CC
UJ
200-
150-
100'
50-
0-
SIEVES
- 17.5 -
t/»
L^
- 15
CL 1.23
J ,0-
«, 7.3-
5
2.5
0
(4 160-
§ i^o
$ '20
u 100
5 80-
cc
LJ 60-
CO
2 40-
i 20-
I T
I I T I I I I ' I
1 23456789
100
80-
60-
40-
20-
10
8-
6-
4-
2-
0-
30
25-
20 -
15-
10-
5 -
0
250 MM SIEVES
T=r
I
i I i I I i I
3456789
WEEK
Figure
yuppie FIELD WEEKLIES
EXCLUSION TREATMENT
INCLUSION TREATMENT
CAGE CONTROLS
LAB MICROCOSM
7. Experimental results in the oligohaline area (station 3) showing
changes in numerical abundance, species richness, and a top
dominant (500- and 250-yn sieve fractions) in various treatments
during Spring 1983. Numbers represent totals for 12-core samples.
17
-------
abundance and species richness due to the loss of certain species popula-
tions (Figures 8-9). However, the pattern of reduced abundance and loss of
populations varied from experiment to experiment depending on the seaso-
nally variable dominance relationships. Throughout each 6-9 week experi-
mental period, the macroinvertebrate microcosms changed relative to the
field. This change was caused by sampling activities and was more apparent
in the microcosms than in the field treatments. Significant (p < 0.05)
differences were noted between sampled and unsampled microcosms after
weekly sampling over a six-week period. Such differences were also noted
between the unsampled microcosms and the field samples. Sampling effects
were noticeably reduced in the fall of 1982 when weekly coring was reduced
to biweekly coring.
During the second week (T].) of the Spring 1982 experiment at the poly-
haline site, inclusion and exclusion cages were grouped together in a
cluster analysis of treatments by week, as were cage controls and labora-
tory microcosms (Figure 10). By week 3 (T£), the laboratory microcosms
were somewhat different from the other treatments. However, for the
balance of the experiment (weeks 4-6), field exclusions were grouped with
laboratory microcosms. The inclusion treatment was intermediate between
the above treatments and the two field controls. Although the laboratory
microcosms differed from the field controls, the changes (i.e., numbers of
Mediomastus with time) resembled the exclusion treatment, which indicates
that, under the circumstances of the experiment, the lack of predation may
have affected the laboratory microcosms (Figure 5). Subsequent experiments
with predators in the laboratory microcosm indicated predation-related
effects on the laboratory test systems. The effects varied according to
the system (no effects were noted under oligohaline conditions) and the
time of experimentation (no release of dominant species was noted during
the fall experiment in the polyhaline system). Thus, predation was a
complex but potentially important factor in the establishment and main-
tenance of laboratory microcosms.
Experiments in polyhaline areas during the fall showed significant
differences in the total numerial abundance and species richness in the
laboratory microcosms relative to the field (Figures 8-9). Differences
between the lab and field were also noted in the oligohaline tests (Figures
6-7). The differences were caused by occasional increases in a given popu-
lation or reduced numbers of organisms in the microcosms at the start of
the experiment (Figure 7). Natural disturbances in the field led to
responses in the infaunal macroinvertebrate assemblages that did not occur
in the laboratory microcosms. The general increase in recruitment in the
field following the storm in the spring of 1983 was not evident in the
laboratory microcosms (see 250-pm sieve data, Figure 7). Specific changes
in physical and biological features of the natural environment had complex
effects on the laboratory test systems, which were largely dependent on the
timing of the test and the nature of the associated macroinvertebrate
assemblages. However, indices such as species richness often showed simi-
lar, if not parallel, trends in the lab microcosms and field treatments.
Such indices were not appreciably affected by changes in a single
population.
18
-------
Z 43O-
§ 400-
g 330-
3OO-
OC 200-
\li
3 ISO-
Z iOO-
30-
CO
8 -
O^ 40-
30-
23-
20-
13-
5OO J1M.SIEVES
2
X3-
20-
10-
0
80
SOO
EVES.
70-
SO-
5O-
40-
30-
20-
10-
0
1
tr
3
70-
60-
30-
40-
30-
20-
10-
WEEK
FIELD »eCKLIS
L»8 MICROCOSM-CONTKOL
LAB MICROCOSM-PREOATOK INCLUSION
LAB MICROCOSM-SEDIMENT DISTURBANCE
Figure 8. Comparison of weekly field collections (field weeklies) with
laboratory microcosms (controls, predator inclusions, sediment
disturbance) with organisms taken from station ML during Fall
1983 (500-um sieve). Numbers represent totals for 12-core samples.
19
-------
FIELD WCCKLICS
LM IMCaOCMM-CONTROL
LM MICROCOSM-MCMTOft INCLUSION
LM MKftOCOSM- MOIMCNT DtSTUftMNCE
Figure 9. Comparison of weekly field collections (field weeklies) with
laboratory microcosms (controls, predator inclusions, sediment
disturbance) with organisms taken from station ML during Fall
1983 (250-pm sieve). All numbers represent totals for 12-core
samples.
20
-------
r, ;
.9
.8
.7
.5
.4
I
1
1
1
T. I .9 .8 .7 .6 .5 .4 .3
'4 | i i i | i I J
.9
OFA
OFB
OML
OFC
OLW
OFA
OLW
OFB
OFC
OML
.7
I
£
I
.9
I
.8
.7
.6
1
.5
.4
.3
OFA
OLW
OFB
OFC
OML
.9
.6
I
.5
? T,-T8
.9
I
.7
I
.5
.4
OFA = EXCLUSIONS
OFB - INCLUSIONS
OFC = CAGED CONTROLS
OML UNCAGED CONTROLS
OLW > LABORATORY MICROCOSMS
OFA1
OFB3
OFB1
OFB2
OFC2
OFC3
OMLD
OFC1
OFA2
OFA3
OLW1
Figure 10. Cluster analysis of changes in species composition by treatment (summed cage results) for
field and laboratory data over a six-week period in tests with polyhaline macroinvertebrates
during spring 1982. Also shown is the cluster analysis by treatment for all replicates
summed over the experimental periods
-------
Recruitment is one of the important processes of benthic communities.
Recruitment of juveniles in the Apalachicola Bay system peaked from
November through February (Figure 11). Recruitment was apparent f or ^_.
littoralis, M. ambiseta, j>_. benedicti, and G_. bonnieroides at station 3; at
station ML, such dominants included Jl. ambiseta, Brania wellfleetensis,
Exogone dispar, Carrazziella hobsonae, immature tubuficids, and Tellina
texana. Recruitment in the laboratory microcosms did not follow natural
field conditions. Often, significant differences occurred in the recruit-
ment of dominant populations in the microcosms. These differences may have
led to increased recruitment in the laboratory for species such- as
Mediomastus (as a response to the disturbance of establishing the microcosm
and sampling) or reduced recruitment of other species such as tubificid
worms. Despite precautions taken with the seawater input from control
areas, recruitment was particularly low in the laboratory microcosms during
the fall experiments under polyhaline conditions (Figure 9). This result
contrasted directly with results of the spring experiments, which were
dominated by enhanced Mediomastus recruitment in the laboratory treatments
relative to field controls (Figure 5). Other species such as Wapsa grandis
showed higher recruitment in the laboratory during the fall experiments
with oligohaline systems (Figure 6). Significant (p < 0.05) differences
existed between laboratory microcosms and field conditions for various
population and community features (numerical abundance, species richness,
dominant populations) of the organisms taken in the 250-pm sieves. Such
results indicated that recruitment of juveniles was affected by laboratory
conditions, but that the exact effect was influenced by the nature of the
source area and the time of sampling.
Microbiology
The microbial community structure in sediments of the polyhaline area
was influenced by the activity of the epibenthic predators at the top of
the food web. Analysis of the lipid components of the sedimentary micro-
biota showed reproducible shifts in the community composition in areas from
which predators were excluded. After six weeks, the microeucaryotes and
cis-vaccenic acid-containing microbes decreased in exclusion areas. Such
disturbance effects could have been associated with observed increases in
the deposit-feeding polychaete Mediomastus ambiseta. Bioturbation by
invertebrate activity could have altered physical and chemical sediment
characteristics. As a result, the growth and activity of certain microbes
were stimulated. Predator exclusion experiments have demonstrated that
intermediate benthic predators can control microbial biomass, community
compositon, and trophic diversity through the immediate prey (infaunal
macroinvertebrates) to the lowest trophic level of the benthic food web
(Federle et al., 1983). Such findings could have implications involving
the use of microcosms in analyzing microbial response to toxic substances
in that removal of microbiota from natural predator-prey effects may
influence the microbiological characteristics of microcosms relative to
the field.
An attempt was made to compare the community structure of the field
with laboratory microcosms meant to mimic the field. Oligohaline and
polyhaline areas of the Apalachicola Bay system were used in this
22
-------
700
600-
LJ
O
< 500-
O
m 400-
< 300-
O
200-
100-
STATION 3
( 250 /tM Sieves)
NOV.' 'Fie! 'MAY
'NOV.' 'FEB.' 'MAY' 'AUG.'
UJ
O
400-
STATION ML
( 500 /
-------
comparison. The polyhaline system (i.e. that with low temporal variability
in salinity) was characterized by a highly organized trophic structure with
little variance in the microbiota. In microcosms prepared from this area,
significant differences existed between the laboratory and field microcosm
after two weeks (Figure 12). Such differences were defined by the molar
ratios of the fatty acids 15:0, 16:lu7, and 18:26 (Z = 2.14), 15:0 (Z =
0.83), and 18:2u6 (Z = -0.82). Samples could be reclassified with 94.4%
accuracy using the two discriminant functions. No field samples were
classified incorrectly as being derived from microcosms or vice versa. It
can be concluded that the classical stabilization period used in many
microcosm studies would be highly misleading in a polyhaline system such as
this one.
Further analysis of differences between the microbiota of micrcosms
and field samples was carried out in oligohaline areas. The distribution
of microcosm and field samples from station 3 along a single discriminant
function after 3 weeks is given in Figure 13. This function was defined by
the ratios of the fatty acids 15:0, 16:0, and 20:5w3. The standardized
discriminant function coefficients for these variables were 0.78, 0.52, and
-0.51, respectively. Microcosm samples tended to be enriched in 15:0 and
16:0 and low in 20:5w3 compared to those from the field. Using this single
discriminant function, 81.5% of the samples could be correctly classified.
Although only one field sample was misidentified, 33% of the microcosm
samples were incorrectly classified. Thus, after 3 weeks, changes in fatty
acid 20:5u3 (common in marine algae) indicated shifts in algal populations.
Also, higher levels of aerobic bacteria in the laboratory microcosms indi-
cated a disturbance effect from the intial establishment of the microcosms.
A graphical representation of the distribution of microcosm and field
samples along a single discriminant function after 6 weeks (Figure 14)
indicated a high degree of variability within each treatment as shown by
the spread of points around the centroid. The regularly sampled microcosms
differed most from the field, and the undisturbed microcosms resembled more
closely the field than did the regularly sampled microcosms. After six
weeks, the differences between the sampled microcosms and the field were
not much greater than they were after three weeks.
The oligohaline site was characterized by high and variable input of
detritus from the Apalachicola River, low salinity, and high secondary pro-
ductivity. There was also high temporal variability in the salinity and
nutrient regimes. Epibenthic predators had little or no influence on
24
-------
5 -
0
O
UJ
o: I -
u_
D FIELD SAMPLES
MICROCOSM SAMPLES
0 GROUP CENTROID
2
-3-2 -10 I
[I5:°J DISCRIMINANT FUNCTION
|6: aj7
I8:2a)6
12. Distribution of field and microcosm derived samples of polyhaline (station ML) sediments along a
single discriminant function after two weeks.
-------
o
2
-
3
ID
O
UJ
a:
u. 2
0
0
-3
-2
-I
D FIELD SAMPLES
MICROCOSM SAMPLES
0 GROUP CENTROIDS
[20:5*3] DISCRIMINANT FUNCTION
15:0"
16:0
Fig. 13. Distribution of field- and microcosm-derived samples of oligohaline (station e) seidments along
a single discriminant function after three weeks.
-------
NJ
>-
o
O
LU
Si*
I -
0
o o
D FIELD SAMPLES
REGULARLY SAMPLED
MICROCOSMS
U UNDISTURBED MICROCOSM
0 GROUP CENTROIDS
-3-2-1 0 I 2 3
[iso!5:0] DISCRIMINANT FUNCTION [anteisol5:0]
Figure 14. Discrimination of oligohaline (station 3) sediment samples from the field, regularly sampled
microcosms, and undisturbed microcosms after six weeks.
-------
deposit-feeding invertebrates or on the microbiota at station 3. Thus,
this system was largely driven by physical and chemical inputs. Although
there were detectable differences between the field microbiota and the
laboratory microcosms, these differences were small and did not increase
with time. In this oligohaline system, the classical "stabilization"
period would have had little effect on the microbial community structure.
The microbiotic component in a laboratory microcosm is highly
variable, and its capacity to predict natural trends depends on a com-
bination of habitat characteristics in the area of origin. Overall, there
is good evidence that, as time progresses, microbial elements in microcosms
tend to deviate progressively from field associations; extended equilibra-
tion periods may be ill advised under certain environmental conditions. In
microcosms of sediments from polyhaline areas, microbes did not follow
field conditions as closely as those in microcosms from oligohaline por-
tions of the estuary. As a result, without detailed knowledge of the
ecology of the source area, an interpretation of results from laboratory
microcosms could be misleading.
Summary of Findings
The relationship of laboratory microcosms to field conditions depended
on a number of variables that changed depending on time and the location of
the test. During the spring experiments in an oligohaline area, signifi-
cant differences were noted for total numerical abundance and species rich-
ness of macroinvertebrates because of laboratory artifacts in recruitment.
Similar experiments in the spring in polyhaline areas led to increases of
the dominant polychaete,. >f. ambiseta, in the laboratory microcosms,
paralleling changes in the field predator-exclusion treatments. Such
changes in recruitment and possible predation effects could have led to
significant differences of various community features between the labora-
tory and field assemblages of microorganisms and infaunal macroinver-
tebrates. The fall tests in oligohaline areas showed significant
differences between laboratory and field treatments as a result of blooms
of the oligochaete Wapsa grandis in the laboratory microcosms. These dif-
ferences became significant after the fifth week of testing. Fall experi-
ments in the polyhaline areas also resulted in significant differences
because of low numbers of individuals and reduced recruitment in the
laboratory treatments relative to the field.
Factors such as spatial habitat gradients, temporal changes in popula-
tion processes, and changes in the influence of predation pressure all
contributed to the complexity of the validation process. Also, the initial
establishment of the microcosms and continued sampling led to observed dif-
ferences between the laboratory microcosms and natural field conditions.
However, the broad spectrum of information provided by microcosms produced
indices such as species richness that were relatively conservative indica-
tors of field conditions. Thus, field validation of macroinvertebrates can
be qualified within known limits of spatial and temporal variability based
on specific ecological conditions in a given area.
28
-------
The results with microorganisms illustrated several points: (1) fatty
acid analysis, combined with multivariate statistical techniques, was a
powerful means of comparing the structure of different microbial com-
munities; (2) microcosms may or may not mimic natural microbial com-
munities; and (3) microbial communities from similar environments but
different ecological conditions may show a wide range of response when iso-
lated in the laboratory. This technique should have great potential in
evaluating a microbial community's response to toxic substances. The major
shortcoming of the microcosm approach is our current inability to interpret
the signifiance of changes in particular fatty acids. Based on this study,
it can be concluded that not all sediments will mirror the field to the
same degree when placed in relatively complex microcosms. Our findings
showed the importance of biological control of microbial communities in the
estuarine environment and the need to include biological as well as physi-
cal factors in the design of model laboratory systems. A priori, without
knowing the specific ecology of a particular site, one cannot conclude that
a reasonably designed microcosm will always simulate the field.
EXPERIMENTAL PROGRAM: VIRGINIA INSTITUTE OF MARINE SCIENCE
Infaunal Macroinvertebrates; Background
Patterns of natural population variations from 1979-83 among the
numerically dominant species indicated seasonal and non-seasonal changes
that related in part to the life histories of the species and the estuarine
environment (Figure 15). Two species, Streblospio benedicti and Paranais
littoralis, exhibited a classic opportunistic life-style of extensive
recruitment over a short period of time followed by mass mortality. The
species S_. benedicti recruited at the same times, spring and fall, each
year. However, the magnitude of recruitment declined each year. For
instance, P_. littoralis was virtually nonexistent in the community from
1979 until early spring 1982, when it had a large and brief recruitment
pulse. An equivalent pulse occurred at the same time in 1983.
The capitellids Mediomastus ambiseta and Heteromastus filiformis have
also been cited as opportunists. Heteromastus filiformis had a regular
recruitment period during the late spring each year. Mediomastus ambiseta
had a large peak only in the spring of 1982. Polydora ligni had less dra-
matic changes in population densities due to recruitment. With some
variation, P_. jligni recruited essentially at the same time of the year.
During 1981, the recruitment was weak, and populations were variable
throughout the year. In contrast, 1980, 1982, and 1983 were years when P_.
ligni exhibited strong recruitments in the early part of each year.
Tubificoides spp. had prolonged recruitment punctuated by periods of high
mortality. This oligochaete had both summer and winter recruitments that
were similar in magnitude from 1979 to 1984. Non-opportunistic, prolonged
recruitment with low mortality was exhibited by Phoronis sp. Population
densities decreased during the late winter but were otherwise maintained at
fairly constant levels.
29
-------
Paranais littoralis
CVJ
CO
<
3
0
>
Q
Z
30,000
20,000
10,000-
0-
1
(l
n
J\ A
iiii
79 8O 81 82 83
Tubificoides
79 80 81 82 83
Streblospio benedicti
79
80
81
82
83
8000
CVJ
*^ 6000 -
CO
_»
^ 4000 -
g
> 2000 -
O
z
0
20,000
w 40,000-
^ 30,000-
^ 20,000-
g
^ 10,000 -
Q
z
A A
^ K ^ /k _
1 1 1 1
CM
Heteromastus ambiseta
79 80 81 82
Polydora ligni sp.
80
81
82
79 8O 81 82 83
Mediomastus ambiseta
83
83
Figure 15. Long-term population variation of dominant annelids at the York River, Virginia, site.
-------
Laboratory-field Experiments
Eleven dominant species were identified by cluster analysis for the
first experiment (Spring 1982; Table 1). Approximately half of all the
individuals involved in the test were recruited during the spring test
period. The top numerical dominant, P_. littoralis, was recruiting heavily
before the test. By the end of the test, this species had almost vanished
in all treatments (Figure 16). Both S_. benedicti and £. ligni recruited
heavily during the test, and to a greater degree in the semi-field treat-
ments. Preferential recruitment into the cages was less dramatic, but
still recognizable for E_. heteropoda and Scoloplos sp. Taken together, the
immature capitellids and M_. ambiseta were more abundant in the full-field
treatments. The greatest recruitment was also into the full-field treat-
ments for Tharyx sp. The numbers of newly set bivalves were lowest in the
microcosms. Tubificoides sp. recruited uniformly into all treatments.
There were no apparent population changes due to recruitement for Phoronis
sp. However, its abundance was considerably lower in the microcosms
because of mortality caused by handling of the sediments in establishment
of the laboratory microcosms.
The variance-to-mean ratios for all dominants were > 1, indicating
aggregated spatial distributions of the individuals. Polydora ligni,
Tubificoides sp., and Paranais littoralis had the highest ratios.
Calculation of Morisita's indices showed that the greatest degree of
clumping was for P_. ligni and Tubificoides sp. in the full-field and semi-
field treatments. The amount of aggregation was also lowest in the
microcosm treatments for Streblospio benedicti and immature capitellids and
bivalves. Similar spatial patterns were found in all treatments for P_.
littoralis, Tharyx sp., Mediomastus ambiseta, and Phoronis sp. The
polychaetes Scoloplos sp. and Eteone heteropoda had less consistent
patterns. Eteone heteropoda was highly clumped in only one semi-field
treatment.
A two-way analysis of variance tested the hypothesis that there was no
difference in the abundance of a particular species among treatments or
through time. None of the two-way interactions between treatments and
weeks was significant. For six of the eleven dominant species there were
significant differences (a < 0.05) in log abundance among treatments.
There were significant differences with respect to time for five species.
Only P. littoralis and newly set bivalves exhibited effects of both treat-
ment and time. When a priori contrasts were used to evaluate treatment
differences, the abundances were not consistently higher or lower in any
one treatment. The lack of trend among the significant differences between
treatments gave positive support to the field validity to this test.
The cluster analysis grouped species mainly on the basis of abundance
and frequency of occurrence. A second analysis essentially separated sta-
tions (treatment replicates for each week) into three successive two-week
periods. The onset of recruitment caused time to have the greatest
influence in determining similarity. Community structure, as measured by
diversity and evenness, did not change greatly through time or between
replicates during Test 1 (Table 2). Even though community structure
31
-------
TABLE 1. Dominant species (individuals/m2) from the field validation experiments, York River, Va.
u>
NJ
FF*
Paranais littoralis 7769
Streblospio benedicti 6272
Mediomastus ambiseta 3570
Immature Capitellidae 5472
Tubificoides sp. 2762
Phoronis sp. 1240
Immature bivalves 772
Scoloplos sp. 262
Tharyx sp. 6984
Eteone heteropoda 1429
Polydora ligni 4198
Glycinde solitaria
Immature gastropods
Acteocina canaliculat
Paraprionospio pinnata
Heteromastus filiformis
Ampelisca abdita
Clymenella torquata
Scolelepis squamata
Spiochaetopterus oculatus
* FF Full-field
SF Semi-field
MC Microcosm
** ND Not a dominant spei
Test 1
SF
10398
11679
2547
4657
2650
1632
874
336
5267
2366
16028
ND**
ND
ND
ND
ND
ND
ND
ND
ND
cies dux
Test 2
MC
17624
5663
3776
2131
2162
697
315
228
3968
869
4175
ing this
FF
714
937
750
2385
1766
933
417
357
286
183
159
test.
SF
ND
1770
992
1143
2666
2103
1726
1237
ND
ND
ND
500
191
151
187
ND
ND
ND
ND
ND
MC
44
698
512
742
627
364
64
385
155
306
242
FF
1015
3211
1104
1526
2481
896
230
487
189
615
878
126
252
48
185
Test 3
SF
2400
9318
1582
2285
2207
522
278
452
285
815
2396
215
ND
ND
ND
663
315
219
ND
ND
MC
2318
3729
1422
582
1248
374
37
1081
156
82
896
337
248
111
255
FF
1012
2420
1444
2944
2056
772
413
358
302
173
543
265
Test 4
SF
ND
1234
1877
1389
2634
1895
802
407
ND
ND
ND
247
241
ND
173
ND
ND
ND
352
136
MC
43
2025
1203
1414
512
475
185
401
216
222
142
123
-------
Paranais littoralis
40,000
TREATMENT
Full-Field
Microcosm
Semi - Field
DATE
Figure 16. Patterns of Paranais littoralis abundance for Test 1.
33
-------
measures did not change much during the test, there were considerable
changes in individual species.
Of the eleven dominant species from Test 2, conducted in the fall,
seven were annelids (Table 1). Approximately half of the individuals were
recruited during the test period. Lower recruitment was observed in the
microcosm treatments, exemplified by _£. benedicti (Figure 17). As a
result, the species populations were generally lower in the microcosms than
in the field. All treatments, including the field, exhibited declining
abundance as expected during the fall season..
The analysis of variance showed two-way interactions for two of the
dominant species. Of the remaining species, two had significant among-week
differences, and five had significant differences in log abundance among
treatments. Evaluation of a priori contrasts showed that four of these
five species had the lowest~~abundances in the microcosm treatments, empha-
sizing the difference between treatments.
Cluster analysis of weekly treatment replicates produced groups based
on treatment. One cluster group contained the full- and semi-field treat-
ments, and another group the microcosm treatments. The grouping of the
microcosms together indicated a difference between the laboratory and
field, which turned out to be primarily caused by differential recruitment.
The measures of diversity and evenness did not change over time or between
treatments (Table 2).
Of the fifteen dominant species from Test 3 in the spring (1983),
three were not annelids (Table 1). In both full and semi-field treatments,
approximately half of the total community was newly recruited. In the
microcosm treatments, only one-fifth of the community was newly recruited.
During Test 3, j>_. benedicti was recruiting heavily into the semi-field
treatments. The populations were constant and smaller in the full field
and microcosm treatments. The same pattern was seen for populations of II.
filiformis. Both £. littoralis and P_. ligni were declining in abundance in
all treatments during the test periods, having reached their peaks before
the test. For P_. ligni, this decline was less rapid in the semifield
treatments. Four species, E_. heteropoda, Phoronis sp., immature capi-
tellids, and Tubificoides sp., had the lowest average abundances in the
microcosms. Of these four, only Tubificoides sp. was not recruiting. In
contrast, Scoloplos sp. was twice as abundant in the microcosms.
Populations of Tubificoides sp., P_. littoralis, £. ligni, and H_. fili-
formis were consistently aggregated, yet less so in the microcosms. In
contrast, Phoronis sp. was more aggregated in the microcosms. There were
no pronounced differences between treatments for immature capitellids, M.
ambiseta, or S_. benedicti. Random distributions were indicated for E.
heteropoda and immature bivalves in the microcosms, C. torquata in the
semi-field, Ampelisca abdidta in the full-field treatments and microcosms,
and £. solitaria and Tharyx sp. in all treatments.
Analysis of variance showed two-way interactions for two species.
the remaining thirteen species, only two exhibited differences based o
Of
on
34
-------
TABLE 2. Community structure indices from the field validation experiments (York River, Va.).
U)
Ul
Test I
H* - Shannon
JPR - Evenness
H - Brillouin
J Evenness
Test 2
H1 - Shannon
JPR - Evenness
H - Brillouin
J - Evenness
Test 3
H' - Shannon
JPR - Evenness
H - Brillouin
J - Evenness
Test 4
H' - Shannon
JPR - Evenness
H - Brillouin
Grand
Mean
2.87
.59
2.80
.59
3.47
.77
3.18
.77
3.11
.74
2.84
.75
3.56
.77
3.39
Full-field Semi-field Microcosm
Mean Range Mean Range Mean Range
3.03 2.58-3.31 2.77 2.16-3.10 2.80 1.00-3.37
2.96 2.53-3.25 2.72 2.14-3.07 2.73 1.95-3.25
3.54 2.97-3.87 3.32 2.95-3.60 3.57 3.08-3.98
3.26 2.76-3.59 3.10 2.75-3.46 3.17 2.72-3.46
3.22 2.85-3.58 3..13 2.48-3.66 2.99 2.71-3.52
2.91 2.57-3.29 2.82 3.21-3.30 2.79 2.48-3.13
3.58 3.21-3.90 3.48 2.97-3.93 3.61 3.08-3.94
3.63 2.99-3.63 3.25 2.77-3.68 3.30 2.81-3.66
J - Evenness
-------
Streblospio benedicti
3,000
cvi
2,000 -
ID
o
^ 1,000 -
o
Full-Fleld
- Microcosm
* Semi- Field
-------
time. Differences among treatments were significant for nine species. Of
these nine, a priori contrasts showed that five had the lowest abundances
in the full-field treatments, three in the microcosms, and one in the
semi-field.
Cluster analysis indicated that the effect of treatment rather than
time was more important in forming groups (Figure 18). The microcosm
treatments were separate from the group of full- and semi-field treatments.
Averaged over time, the highest diversity was calculated for the full-field
treatments, and the lowest in the semi-field .treatments. The -low diversity
in the semi-field treatments was probably due to a decline in diversity
during the last two weeks of the test.
Twelve species were identified as dominants from Test 4. Of these,
nine were annelids. About half of the individuals were recruited during
the test period. Recruitment was lower in the microcosms.
Analysis of spatial distribution patterns showed that six species were
consistently aggregated. Mediomastus ambiseta exhibited no differences in
the degree of aggregation between treatments. Immature bivalves,
Tubificoides sp., and Phoronis sp. were all the least aggregated in the
microcosm treatments. Species that were not highly aggregated and often
approached a random distribution were immature gastropods, j^. squamata,
Scoloplos sp., and _S_. oculatus. Of these, the gastropods were most aggre-
gated in the microcosms, and Scoloplos sp. and j^. squamata were the most
aggregated in the full-field treatments.
The trend evident from the cluster analysis was grouping by treatment,
the microcosms being separate from the full- and semi-field treatments.
With regard to community structure all treatments showed the same patterns
as the other tests. On average, the diversity was slightly higher in the
full-field treatments than in the microcosms. Diversity was often lowest
in the semifield treatments.
Test Comparisons
Summing the average abundances of the top eleven numerically dominant
species from each test illustrated a major difference in community abun-
dance. The approximate numbers for each test were: 50,000 individuals
m~2, Test 1; 9,000, Test 2; 16,000, Test 3; and 11,000, Test 4. From the
monitoring of the long-term dynamics it was expected that the fall tests (2
and 4) would have fewer individuals than the spring tests (1 and 3). The
larger discrepancy between the two spring tests was due to two factors.
First, the dominant species (M. ambiseta), the associated immature capi-
tellids, and £. ligni had above-average recruitment during the spring of
1982 (Test 1). Below-average recruitment was seen for several species
during the spring of 1983 (Test 3), especially for Tharyx sp. and E_.
heteropoda. Second, Test 1 was conducted during the peak in recruitment
for many species, including the end of this period for P_. littoralis, but,
during Test 3, the short but considerable peak in the abundance of P_.
littoralis was missed. Test 3 included the end of the recruitment period
for some of the other dominant species.
37
-------
2 3 4 5 6 7 8 9 10 II 12 13 14 15 16
i I i i i i i i i i i I i I I
? 18 19 20 21 22 23 24 25
I i I i I i i i i
00
Fll
F3I
Mil
M2I
M3I
M33
M35
MI3
MI5
M23
M25
F2I
S3I
Sll
S2I
SI3
FI3
F25
FI5
F35
S25
F23
F33
S23
S33
S95
S35
FI3 Full-Field I, week 3
M.. Microcosm
S.. Semi-field
Figure 18. Cluster analysis of Spring 1983, Test 3 York River, Virginia, site.
-------
The dissimilarity between Test 1 and the other three tests was also
reflected in the lower diversity indices (Table 3). The large numbers of
P_. littoralis decreased the measurement of diversity. In comparisons of
differences between treatments, the spring tests showed the highest diver-
sity in the full-field treatments. Of the fall tests, Test 2 had lower
diversity in the semi-field treatments, yet in Test 4 all treatments were
similar.
A cluster analysis of all four tests identified nine dominant species
based on abundance. Fifteen other species were identified as subdominants.
The analysis of stations divided the treatment replicates from each week
into two groups. The first group contained stations from the spring tests.
This group was further divided into two subgroups, one of which contained
all of Test 1, and the other, all of Test 3. The second group contained
the two fall tests, which were not clearly divided within the group based
on test, treatment, or time.
For comparison of species trends among the four tests, slopes of log
abundance over time were calculated. The abundances were standardized to
number per m^, and the first 5 weeks of each test were used. The slopes
were used to indicate only trends in abundance and not causal rela-
tionships. Several species exhibited changes caused by timing of the tests
and to treatment. When recruitment was evident., such as for _S_. benedicti
and Scoloplos sp. during both spring and fall tests and for immature
bivalves in the spring, the rate of recruitment based on slope was either
less pronounced or nonexistent in the microcosms. For some species, there
were declines in abundances in the microcosms. This was consistently true
for Phoronis sp. During the fall tests when populations of Phoronis sp.
were decreasing in all treatments, it was more pronounced in the
microcosms.
In some instances, a population trend was equivalent in all treat-
ments. Such was the case for recruitment of immature capitellids during
Test 1, and for the declining abundances of immature bivalves and P_.
littoralis during Test 1 and Test 4, respectively. The downward slope of
the recruitment of P_. littoralis was also included in the second spring
test (Test 3). However, the populations in the microcosms did not decline.
Two species, Tubificoides sp. and 11. ambiseta, had fairly constant popula-
tions. There were no sizable increases or decreases in abundances, and no
differences between treatments. Slopes of the dominant species indicated
an effect of exposure to hydrocarbon-contaminated Elizabeth River sediment
only in the microcosms. Scoloplos sp., immature bivalves, II. ambiseta, and
immature capitellids all declined in the contaminated microcosms more
rapidly than in the unexposed microcosms.
Summary of Findings
In the context of microcosm research, it is not necessary that we know
the causes of population fluctuation but only that fluctuations occur. It
is the interactive nature of the community and the environment that
generates the fluctuations we observe. So, in evaluation of a microcosm
39
-------
toxicity test, it is necessary to consider the broad, total-community
approach. We should avoid singling out one species for assessing toxicity.
Long-term population dymamics will result in periods when any given
species may be present in low abundance. This would make repeated testing
difficult if those species in low abundance were needed. Also, at any
laboratory conducting community microcosm tests, it is essential to know
the natural population fluctuations. Otherwise, major changes in the com-
munity associated with natural cycles would be missed, making interpreta-
tion of microcosm results difficult or misleading. The total community
represents a single energetic entity. From year to year, about the same
amount of energy flows through the community. Although individual species
patterns are different from year to year (and consequently the amount of
energy flowing through each species is different), the total energy budget
is relatively constant.
The following findings exemplify the need to consider the total
system:
1. Tharyx sp. declined by a factor of 10 from 2,000 m~2 in 1983 to 200
m~z in 1982. The Mediomastus ambiseta population increased from low
abundances in early 1980 to peak in mid-1982 and declined through
1983. Paranais littoralis did not have successful recruitment in
1980 or 1981, and it was not even a community dominant until 1982.
Most of the dominant species exhibited some year-to-year variation
that might make repeated testing difficult if it were based on a
single species.
2. The major natural fluctuations in the community were associated with
recruitment. Should the initiation of a microcosm test unknowingly
coincide with recruitment, populations could increase or decrease by
orders of magnitude in test treatments. The onset of recruitment can
generally be easily identified from the size of individuals. It is
the subsequent decline of the recruitment peaks that could cause
problems of interpretation. Without knowledge of the natural timing
of these declines, it might be difficult to identify toxic effects.
The species that consistently exhibited highest mortality after
recruitment from 1980 to 1983 were Paranais littoralis, Streblospio
benedicti, and Heteromastus filiformis.
3. Overall, there was about the same density of individuals in 1980,
1981, and 1982 (defining the year from October to September to better
coincide with recruitment). In 1983, the density dropped by a third.
In 1980 and 1981, populations of the dominant annelids were about the
same size both years. Based on this fact and the assumption that
total yearly production can be partitioned between species and still
remain constant from year to year, it seems likely that the total pro-
duction for the York River site was the same in both 1980 and 1981.
We have not looked at the size of individuals in 1982 and 1983 to see
whether this is the general trend. The importance of this produc-
tivity to microcosm testing is in understanding the interactive nature
of the community. If one species is in low abundance for a given
40
-------
year, then another may be more abundant and offset the loss in produc-
tivity. Although the community structure changes, the functioning of
the community remain unchanged. Microcosms need to capture this func-
tional response to represent field response truly.
It took a broad view of all parts of the community to see the rela-
tionship between the laboratory microcosm, which is the target of interest
as a tool to judge environmental consequences of toxicants, and the field.
Cluster analysis indicated that during Test 1 (spring of 1982) the micro-
cosms behaved very much like the field, but in Test 3 (spring of 1983) they
did not. Apparently, in the spring of 1983, recruitment into the micro-
cosms was reduced relative to the field, possibly because of some labora-
tory artifact or timing of the test relative to recruitment peaks. Results
of the fall tests were consistent, with recruitment being less in the
microcosms. With this understanding that recruitment into the microcosms
will likely be lower than in the field, because of the nature of the test
system, we can more accurately interpret toxic effects in the microcosms.
No one species was able to carry consistently sufficient information
about the validity of the microcosm test system. Analysis of the variation
in individual species abundance within and between tests showed that most
species did not have a consistent response to the full-field, semi-field,
or microcosm treatments. The exception was Phoronis sp., whose populations
were always lowest in the microcosms because of an artifact of the test
system (larger individuals live deeper than 10 cm in the sediment and were
damaged when the microcosms were established). It seems that the behavior
of the natural system, and any portion of that system brought into the
laboratory, has a stochastic component that precludes taking a few of the
species and putting the whole back together again.
Preliminary Toxicology
Preliminary toxicity tests were conducted to evalute further the vali-
dation criteria developed in the previous tests. These experiments were
carried out with contaminated sediments taken from the Elizabeth River (VA)
to develop techniques for application of a toxicant to laboratory micro-
cosms and field treatments. This sediment had (parts per thousand)
concentrations of polycyclic aromatic hydrocarbons. Unpolluted sediments
from the York River and Apalachicola estuary were used as treatment
controls. Contaminated sediments were applied to enclosures over a twenty-
four hour period to allow settling of this sediment. Even with nominal
toxicant concentrations, certain problems were noted concerning the
response of the laboratory microcosms and field treatments to the
toxicants:
1. Overall, simultaneous laboratory-field experiments require close
attention to the mode of application with comprehensive chemical ana-
lysis to evalute equivalence of exposure while specific objectives of
the validation process are fulfilled.
2. Close chemical surveillance is necessary concerning the distribution
of the toxicant.
41
-------
3. Field treatments should be carried out in such a way that control
areas are not contaminated.
4. Protocols for treatment should be developed so that recognizable but
transient effects are noted without causing persistent adverse impact
on the infaunal biota.
In summary, the nominal toxicant test indicated that the establishment
of the microcosm treatment was the most sensitive part of the experiment.
Most of the variation in abundance and changes in species could be attri-
buted to the microcosm treatment. Through the course of the experiment,
microcosms exposed to hydrocarbon-contaminated sediment did show the
greatest degree of change. This sensitivity of the laboratory microcosms
to toxic stress was documented even though there was a considerable contri-
bution to the variance from the treatments. The exposure was possibly not
as effective in the field treatments because of differences between
laboratory and field conditions in terms of water volume and the even
distribution of contaminated sediments. This problem may have reduced the
component of variance caused by exposure to contaminated sediments.
42
-------
SECTION 5
CRITERIA FOR VERIFICATION PROCEDURES
^ simultaneous use of replicated multispecies microcosms (as defined
in Giesy, 1980) and field mesocosms (Grice and Reeve, 1982) to test the
validation hypothesis has led to specific observations concerning the rela-
tionships of full-field, semi-field, and controlled conditions. Criteria
that relate the laboratory and field approaches to research of benthic
estuarine associations are given in Table 3. Physical and chemical changes
in the laboratory sea water quality relative to field conditions are una-
voidable. Laboratory artifacts include changes in hydrostatic pressure,
and current structure, which may lead to different sedimentation patterns.
Procurement, transfer, and placement of sediments in the microcosms also
sometimes leads to severe alterations of sediment conditions. Specific
changes in the microcosm habitat arise from its isolation from the field
and are enhanced by surface features of the laboratory enclosure. Although
the effects of laboratory conditions can be avoided in varying degrees,
duplication of field conditions is usually precluded by the conditions
imposed on the microcosms. The real problem is to define those aspects of
microcosm function that can be used to explain field conditions.
Some features of laboratory microcosms are especially difficult to
control. Sudden changes of temperature or sedimentation in the field can-
not be replicated in the laboratory microcosm. At the same time, the
microcosm often acts as a silt trap through time, thus altering sediment
and water column relationships relative to the field. Whereas natural phy-
sical disturbances such as storm effects are lacking in the laboratory
microcosm, other features of the sediment and water column within the
microcosm undergo a departure from natural conditions because of the limi-
tations imposed by the size of the microcosm as compared to a virtually
limitless natural environment.
Physical disturbance of the sediments in a given microcosm can be
divided into two primary sources of impact: transference of sediments in
the establishment of the microcosm and sampling during the course of an
experiment. Our experiments indicated that establishment of the microcosm
and separation from surrounding sediments can have an immediate impact on
the macroinvertebrate assemblages in the microcosm. Often, certain sen-
sitive species were lost in the transfer; the exact impact of this effect
on numerical abundance and species richness varied according to seasonal
patterns of relative abundance in the estuarine associations. Results of
some experiments indicated a deviation from field conditions within time
periods of 4-6 weeks. Too-frequent (i.e., weekly) sampling of the
43
-------
TABLE 3: Criteria for review of the validation of infaunal macroinvertebrate microcosms with semi-
field mesocosms and full-field conditions in estuarine systems.
Factor/Condition Full-field
Semi-field cage
Microcosm
Physico-chemical
Water source
Water supply
Currents
Sampling effects
Light
(continued)
No effect.
No effect.
Unaffected; variable in
magnitude and direction
because of tides and wind
effects.
Damage to fauna during
sampling; slumping of
sediment to fill core
holes.
Unshaded, but low light
intensity due to water
depth/turbidity.
No effect.
Flow impeded by screen;
some recruits set on
screen.
Slowed by screen mesh;
variable in magnitude and
direction because of tide.
Effects minimized by
choice of mesh and cage
design.
Same as for full-field.
Slumped holes may have
trapped organics, making
attractive site for larval
setting and immigration.
Same as full-field.
Drawn from near bottom,
50-100 m from field and
semi-field site.
In-pipe setting only in last
15 m; minimal reduction in
62; alterations of larval
setting.
Established by position of
input and output and by
sediment boxes; invariate
once stablished. No true
simulation of tidal and
wind-driven currents.
Replacement of cores with
azoic sediment; migration
into azoic sediment led to
dilution of populations.
Because of scaling effect,
sampling had more of an
impact on the microcosm than
under fi£ld conditions.
Microcosm tanks varied from
being partially shaded to a
general replication of
light intensity in the
field.
-------
TABLE 3 (continued).
Factor/Condition Full-field
Semi-field cage
Microcosm
Physical/Chemical (continued)
Sedimentation Frequent resuspension.
Physical
Disturbance
Sediment
Compaction/
pore water
Sediment
temperature
(continued)
Bioturbation by epifauna
(e.g., crabs and fishes).
Enhanced microbial
activity.
No effect.
No effect.
Possibly enhanced by
reduced water flow. May
be reduced near edges
by scouring. No major
effects noted in sediment
characteristics over 4-9
week periods.
Large sediment distur-
bances excluded, activity
of smaller species became
more important. Enhanced
microbial activity.
No effect.
No effect.
Enhanced by slow water flow
in microcosm tank. May be
changed by water intake
system. Accumulation of
silt may have been enhanced
beyond the rate in the
field.
Certain forms of bioturba-
tion were enhanced because
of limited area compared to
field. However, large-
scale disturbance due to
storms and tidal currents,
not reproduced in micro-
cosms .
Compaction reduced by remo-
val from field and sub-
sequent slumping; probably
gradual compaction as
experiment progresses.
Flow of pore water
restricted, possible
changes j.n granulometric
properties.
Temperature changes relati-
vely rapid, no insulation;
minor difference from
field.
-------
TABLE 3 (continued).
Factor/Condition Full-field
Semi-field cage
Microcosm
Physical/Chemical (continued)
Sediment pH
No effect.
No effect.
Substrate depth
No effect.
No effect.
Hydrostatic
pressure
Variable because of tides,
waves.
Same as full field.
Biological
Larval recruitment No effect,
Possibly affected by mesh
of cage. Solid substrate
may have attracted some
species. Affected by mesh
size and type. No effects
noted in exclusion cages
in this series of
experiments.
Sedimentary processes
affected pH by shallowness
of microcosm, sedimentation
enhancement, and changes in
compaction.
Limited escape routes,
deep-dwelling organisms
eliminated. Vertical orga-
nization of macroinverte-
brates altered by depth
restrictions.
Usually lower than field;
less variable.
Possibly affected by dif-
ference in water source;
potential change in avai-
lable recruits due to
passage through pipe; solid
substrate attractive for
setting pf some species,
selective (species-speci-
fic) mortality in lab.
(continued)
-------
TABLE 3 (continued).
Factor/Condition Full-field
Semi-field cage
Microcosm
Biological (continued)
Predation (large,
mobile epibenthic
organisms)
Immigration
Competition
Food source
Major impact under
specific conditions of
salinity and at certain
seasons of the year.
No effect.
Interference competition
may have been important,
although complexity
precluded generalization.
No effect.
Impact reduced by exclu-
sion of large mobile
predators.
Possible effect of screen
inserted into substrate
and in water column.
Same as full field.
Possibly enhanced
enrichment from cage
mesh.
Same as semi-field.
Probably eliminated; most
pelagic immigrants were
probably destroyed by
pumps.
Same as full-field and
semi-field.
Possibly altered by sea-
water system, microbial
effects.
-------
microcosm during such experiments also led to alterations in the microorga-
nism and macroinvertebrate assemblages.
If the laboratory microcosm was isolated from natural benthic
assemblages, recruitment processes were altered, as shown in results from
both research groups. Such changes can be enhanced by handling of water
prior to entry into the microcosm. Another possible recruitment problem is
isolation from surrounding populations that propagate through benthic
transfer of larvae rather than through the plankton. Immigration is
severely restricted. Patterns of recruitment and immigration are habitat-
and time-dependent. Each species recruitment pattern should be evaluated
to determine the potential for extrapolation of laboratory results to field
conditions.
Biological processes other than recruitment may be altered under
laboratory conditions. Isolation from natural field processes disconnects
microcosm assemblages from interactions with various types of predators.
Our experiments indicated that the impact of predation on field assemblages
of macroinvertebrates was extremely complex. In addition to gradient
effects of salinity on such impact, there were also seasonal differences in
the predator influence in the field. During spring periods of maximal
influence of predation impact in polyhaline areas of the Apalachicola Bay
system, isolation of the microcosm in the laboratory led to increases of
dominant populations that follow observed changes in exclusion cages in the
field relative to inclusion cages and cage controls. Such changes were
associated with altered microbial community structure. At other times of
the year and under oligohaline conditions, no such effects were observed.
Direct and indirect effects on natural energy relationships also occurred
in the microcosms. Such effects may have given selective advantages to
certain macroinvertebrate populations. Altered predation pressure,
together with unavoidable restrictions in the depth (and vertical popula-
tion distribution) of the microcosms, may alter competitive interactions
that occur naturally in the field. The difficulty of demonstrating such
complex competitive interactions under field conditions disallows strict
generalization. Overall, simultaneous experiments with laboratory micro-
cosms, semi-field conditions, and full-field conditions indicated that
biological interactions comprised an important element in the verification
of the predictive capability of microcosms to natural conditions.
The laboratory microcosms followed field conditions when viewed as
groups of interacting populations rather than as sets of individual popula-
tions. Specific community parameters, such as species richness and diver-
sity, and other indices of multispecies associations, when qualified by
known changes caused by laboratory artifacts, were representative of field
situations. Verification of both microbial and macrobiological assemblages
was possible only within the bounds of our knowledge of the systems in
question. Moreover, the critical factors that determined qualifications
(i.e., recruitment, predator-prey interactions, relative species dominance)
were relatively similar in two entirely different experimental areas.
Thus, field to field extrapolation of results is also possible when it is
based on a thorough knowledge of the subject systems.
48
-------
REFERENCES
Bloom, S. A. 1981. Similarly indices in community studies: potential
pitfalls. Mar. Ecol. Prog. Ser. 5:125-12.
Cairns, J., Jr., M. Alexander, K. W. Cummings, W. T. Edmondson, C. R.
Goldman, J. Harte, R. Hartung, A. R. Isensee, R. Levins, J. F.
McCormick, T. J. Peterle, and J. H. Zar- 1981. Testing for effects
of chemicals on ecosystems. National Academy Press, Washington, D. C.
98 pp.
Diaz, R. J., S. C. Thornton, and M. J. Roberts, Jr. 1984. Field
Validation of a Laboratory Derived Aquatic Test System. Unpublished
Renort. U.S. EPA, ERL, Gulf Breeze.
Federle, T. W., R. J. Livingston, D. A. Meeter, and D. C. White. 1983.
Modification fo estuarine sedimentary micrpbiota by exclusion of top
predators. J. Exp. Mar. Biol.Ecol. 73:81-94.
Giesy, J. P., Jr. 1980. Microcosms in ecological research. Technical
Information Center, U. S. Department of Energy. Washington, D. C.
1110 pp.
Grice, G. D., and M. R. Reeve. 1982. Marine Microcosms. Springer-Verlag,
New York. 430 pp.
Hammons, A. S., J. M. Giddings, G. W. Suter, II, and L. W. Barnhouse.
1981. Methods for Ecological Toxicology: A Critical Review of
Laboratory Multispecies Tests. U. S. Environmental Protection Agency.
Environmental Sciences Division Publication No. 1710. Washington,
D. C. 307 pp.
Lance, G. N., and W. T. Williams. 1967. A general theory of classifica-
tory sorting strategies. I. Hierarchical systems. Comput. J. 9_i
373-380.
Livingston, R. J., and D. A. Meeter. In press. Correspondence of labora-
tory and field results: what are the criteria for validation? Proc.
Symp. for Multispecies Testing. Ed., J. Cairns. Blacksburg, VA.
Livingston, R. J., R. J. Diaz, and D. C. White. 1983. Final report:
Field and semi-field validation of laboratory-derived aquatic test
systems (October, 1981-September, 1982). Unpublished report.
49
-------
Mahoney, B. M. S., and R. J. Livingston. 1982. Seasonal fluctuations of
benthic macrofauna in the Apalachicola estuary, Florida, USA: The
role of predation. Mar. Biol. 69:207-213.
Matusita, K. 1955. Decision rules based on the distance for problems of
fit, two samples and estimation. Ann. math. Statist. 26:631-640.
van Belle, G., and I. Ahmad. 1974. Measuring affinity of distribution.
Pp. 651-668 in Reliability and Biometry: Statistical Analysis of
Lifelength. Ed. by F. Proschan and R. .Serfling. Society for
Industrial and Applied Mathematics, pp. 651-668.
50
-------
APPENDIX A
FIELD OPERATIONS FOR THE VERIFICATION PROJECT INCLUDING A LIST OF
PHYSICAL/CHEMICAL AND BIOLOGICAL SAMPLING PARAMETERS TAKEN WEEKLY AT THE
RESPECTIVE STUDIES SITES (FSU, OLIGOHALINE, POLYHALINE; VIMS, POLYHALINE)
I. Water quality (surface and bottom)
depth (m)
salinity (°/oo)
color (Pt-Co units)
turbidity (JTU)
dissolved oxygen (ppm)
PH
II. Meteorological data (FSU only)
Apalachicola River flow (daily,
Apalachicola rainfall (daily-monthly, cm)
III. Sediment quality
temperature
water content
Eh profile
organic content
grain size
IV. Biological factors
microbiota
meiofauna (VIMS, FSU)
benthic macroinvertebrates (500- and 250-pm sieves)
meroplankton (FSU only) 82-pm plankton net)
epibenthic fishes and invertebrates (FSU only) (repetitive 5-m
otter trawl tows)
51
-------
APPENDIX B
GENERALIZED PROTOCOL FOR LABORATORY MICROCOSM/FIELD-SEMI-FIELD VALIDATION
STUDIES.
I. Laboratory microcosms (0.7-1.0 m2)
A. Physical/chemical data
1. temperature (°C)
2. salinity (°/oo)
3. color (Pt-Co units)
4. turbidity (JTU)
5. dissolved oxygen (ppm)
6. pH
7. sediment % organics
3. sediment grain size
9. sediment temperature, salinity, Eh
B. Infaunal macroinvertebrates (500- and 250-|am sieves)
1. repetitive cores (3 replicates, 1-3 treatments; FSU only)
2. vertical distribution (2-cm intervals)
3. azoic sediment samples (500- and 250-pm sieves
C. Microbes
1. repetitive cores (3 replicates, 1-3 treatments)
II. Field
A. Treatments (3 replicates)
1. unscreened platforms
2. screened platforms (exclusion cages)
3. screened platforms (predator-inclusion cages)
4. weekly core samples (no platform)
5. additional treatments (sepcific for individual experiments)
B. Physical/chemical data (same as I.A.)
C. Infaunal macroinvertebrates (same as I.B.)
D. Microbes (same as I.C.)
E. Epibenthic fishes and invertebrates (repetitive otter traw tows)
F- Meroplankton (repetitive plankton net samples)
III. Variables analyzed
A. Infaunal macroinvertebrates, epibenthic organisms
1. numerical abundance (total and dominant species)
2. ash-free dry weight biomass (total and dominant species)
3. species richness
4. species diversity and evenness indices
5. productivity (growth per unit time of dominant species)
B. Microbes
1. total biomass (phospholipid, 16:0, 16:lu7, 18:2oi6, 18:lo>9)
2. bacteria (iso and anteiso 15:0, called Br 15:0, 15:0)
3. photosynthetic microbes (16:lo>13t)
4. microeukaryotes (20:4(o6, 20:5o>3, 20:3to6, 22:4o)6, 22:6o>3)
5. bacterial ecotype (18:lu)7)
52
-------
APPENDIX B (continued)
IV. Experimental Methods
A. Weekly analyses
Weekly field samples (see Appendix A) provided background
information for the combined laboratory-field tests carried out
by personnel from FSU and VIMS. Samples of macrofauna were taken
weekly from the study sites from October, 1982 (FSU), and
September, 1979 (VIMS).
B. Laboratory-field Tests
The combined experiments were conducted in the spring and
fall, 1982, and spring and fall, 1983. Each test was from five
to nine weeks in duration. Test 1 was sampled weekly; all sub-
sequent tests were sampled at two-week intervals after the initial
two samplings (TQ, TI) , which were one week apart (Tg = initial
sampling).
For most of these tests, three basic field treatments were
used:
Full-field either a 2 m x 2 m platform or an area deli-
neated by a 30-m circle. No manipulation of
the environment involved.
Semi-field a 2 m x 2 m caged platform, screened with 6-mm
galvanized steel mesh, which extended 30 cm
into the sediment. The cage excluded large
predators that would also be excluded from the
laboratory microcosms.
Laboratory microcosms 0.8-1 m x 1 m x 0.1 m fiberglass
boxes placed in larger tanks with unfiltered
water flowing through at the rate of 12 to 19
turnovers/day.
In addition, the FSU group used a predator-inclusion treatment
using indigenous species in numbers resembling actual densities
at the time of testing.
Sediments for the microcosms were obtained by SCUBA divers
in the following manner. Rectangular metal cans (10 x 20 x 10
cm) with their bottoms removed and screw caps open were inserted
into the sediment to a depth of 10 cm. The lids were tightened
and intact sections of sediment were transferred in the can to
the laboratory. Multiple cans containing sediment were placed
adjacent to one another in the microcosms; the lids were opened
and the cans removed leaving the sediment in the microcosms. In
this way, relatively large areas of sediment were transferred to
the laboratory with minimal disturbance of their vertical
structure.
At weekly intervals, random samples of sediment were removed
from three of the microcosms using plastic core tube. At each
sampling four cores were removed from each microcosm. At the
same time, similar samples were removed from the field. In the
field, four or five random samples were taken from each of three
2 x 2 m control plot.
53
-------
APPENDIX C
SAMPLING SCHEDULES FOR THE COMBINED (FSU-VIMS) EXPERIMENTAL PROGRAM
(1981-1984).
I. Weekly samples
A. FSU
T7~ oligohaline stations 3, 5A (11/24/81-11/17/83)
2. polyhaline station ML (11/25/81-3/15/84)
B. VIMS
1. polyhaline marine lab station (10-13-79-12/18/83)
II. Microbiological data
A. FSU
1. station 3 (fall, 1982; spring, 1983)
2. station ML (spring, 1982)
B. VIMS
1. marine lab station (spring 1982)
III. Combined (field-laboratory) experiments
A. FSU
1. station 3 (fall, 1982; spring, 1983)
2. station ML (spring, 1982; fall, 1983)
B. VIMS
1. marine lab station (spring, 1982; fall, 1982; spring, 1983;
fall, 1983; spring, 1984)
Test 4 was carried out in a somewhat different way by the
two research groups. The VIMS group used a nominal toxicant test
using "naturally" hydrocarbon contaminated sediment from the
Elizabeth River, Virginia. This sediment had parts per thousands
concentrations of polycyclic aromatic hydrocarbons. Unpolluted
York River sediment was also used as a control, being similar in
compositoin (80-90% silt-clay) to the polluted Elizabeth River
sediment. Expsure was accomplished by creating of an enclosed
calm environment for 24 hours to allow settling of the sediments.
The FSU group conducted a five-week test with a protocol
similar to that used during previous tests. Contamination of the
field treatments and lab microcosms with the hydrocarbon-
contaminated sediment from the Elizabeth River, Virginia,
together with sediment controls from the VIMS and FSU study
areas, was then undertaken subsequent to the 15 sampling period.
Exposure methods were the same as those described above for the
VIMS program. These treatments were then sampled over a 4-week
period (Tj, Tg).
54
-------
GLOSSARY
Benthic of or pertaining to the bottom of any aquatic, estuarine, or
marine habitat.
Bioturbation process of mixing, or otherwise disturbing, the physical
structure of sediments by organisms.
Community a group of species living together in the same place.
Diversity a measure of the relationship between the number of species
present and their relative abundances, based on Shannon's and
Brillouin's formulas.
Dominant a species that is numerically most abundant in a sample or
habitat. Usually there are several dominant species and many more
less abundant species present in a habitat.
Evenness (equitability) A measure of the relative numbers of individuals
of sepcies prsent in a given area.
Full-field a term used to describe the natural undisturbed field habi-
tat; represents one of the three treatments used in the validation
process.
Infauna species that live in the sediment.
Long-term time spans of years.
Macroinvertebrates individuals of species that are retained on 250-pm
sieves.
Mesohaline salinities in the range of 5 to 18 °/oo.
Microcosm a portion of the natural field habitat that has been brought
into the laboratory, with as little disturbance as possible.
Microorganisms individuals of species that pass through a 62-vim sieve;
mainly bacteria and other unicellular forms.
Oligohaline salinities in the range of 0.5 to 5 °/oo.
Opportunistic species One whose life history includes rapid response to
disturbance and underutilized resources.
Polyhaline salinities in the range of 18 to 32 °/oo.
55
-------
Recruitment process whereby communities renew their numbers; usually
involves reproduction and settlement of new individuals, but can also
be accomplished by migration of adults.
Semi-field a term used to describe the areas in the field that are
enclosed by 6-mm mesh cages. Each of these cages was 2 m on a side.
Scaling parameters number and size of replicates needed to give sta-
tistically reliable .results.
Short-term time spans of weeks to months.
Species diversity index derived from species richness and evenness
components
Species richness a component of diversity that measures the number of
species relative to the number of individuals or simply the total
number of species in a sample.
Validation determination of the capacity of a specific laboratory micro-
cosm to predict the environmental response to toxic materials. A
field validated microcosm would then provide a realistic means of
correlating toxicological data with expected field results.
LIBRARY
U.S. ENVIRONMENTAL PROTECTION AGENCY
GULF BREEZE ENVIRONMENTAL RESEARCH LABORATORY
SABINE ISLAND
GULF BREEZE, FL 32661-6299
56
------- |