Energy Research
and Development
Administration
United States
Environmental Protection
Agency
Division of Biomedical
and Environmental Research
Germantown, Maryland 20767
Office of Research and Development
Office of Energy. Minerals and Industry
Washington, D. C. 20460
EPA-600/7-77-096
August 1977
TROPHIC STRUCTURE
MODIFICATIONS BY
PLANKTIVOROUS
FISH IN AQUATIC
MICROCOSMS
Interagency
Energy-Environment
Research and Development
Program Report
-------
LBL-5978
RUNNING HEAD: Microcosm trophic structure
Trophic structure modifications by planktivorous fish
in aquatic microcosms.1
A. Jassby, J. Rees, M. Dudzik, D. Levy, E. Lapan and J. Harte
Energy and Environment Division
Lawrence Berkeley Laboratory
University of California
Berkeley, California 94720
Research supported by the U.S. Energy Research and Development Administration
and the Environmental Protection Agency D5-E681 through contract #77BCC
-------
-1-
ABSTRACT
Two of 4 replicate 700-liter aquatic microcosms each were stocked with
2 mosquito fish (Gambusia affinis). The dominant zooplankter shifted from
the large cladoceran Simooephalus vetulus to the smaller Alona guttata. The
subsequent release of grazing pressure resulted in a rise in both phytoplankton
and bacteria levels, which in turn were responsible for an increased rotifer
biomass. Particulate organic carbon was higher and dissolved inorganic nitrogen
was lower in the presence of Gambusia, reflecting a net shift of nutrients
from inorganic to organic form, presumably because of smaller zooplankton
respiratory losses. Ratios of particulate to dissolved organic carbon and of
phytoplankton carbon to chlorophyll a_ were unnaturally high in the microcosms
containing fish. An increase in total nitrogen was deduced for all 4 systems
during the experiment; the increase could be explained by the presence of hetero-
cystous Anabaena sp. and was independent of the presence of fish.
-------
-2-
I INTRODUCTION
Aquatic microcosms potentially are valuable tools for dissecting the
important mechanisms operating in natural ecosystesm, as a number of studies
have suggested (reviewed by Cooke, 1971). Microcosms also may have a central
role to play in predicting the effects of human-derived disturbances on these
mechanisms (Draggan, 1976). The applicability of microcosms for assessing
human impacts on aquatic ecosystems ultimately depends on our ability to demon-
strate basic analogies between microcosms and natural water bodies. Certain size-
related differences between microcosms and natural systems, such as the surface-
to-volume ratio of the containing structure, never will be bridged, but the
usefulness of microcosms for a given research problem usually does not require the
creation of a perfect analog to some naturally-occurring system. If microcosms
are to fulfill their potential, detailed investigation of both the similarities
and the insurmountable discrepancies between laboratory and natural water bodies
is required. Only then can we have a basis for accepting or rejecting with
confidence a specific prediction derived from research with microcosms.
Two kinds of experiment can be distinguished in the investigation of
relationships between synthetic and natural ecosystems. In the first type of
experiment the microcosms are allowed to develop without disturbance, the
primary purpose being to determine whether or not the undisturbed system is
capable of simulating production cycles or other basic features characteristic
of any naturally-occurring counterpart. The second type of experiment, consists
-------
™" ,J™
of an experimental manipulation of the physical, chemical, or biological
characteristics of a microcosm; the response of the microcosm then is com-
pared with the response of natural ecosystems to the same disturbance. It
is this latter kind of experiment that is a true test of the predictive value
of microcosms.
The research reported here summarizes an experiment of the second kind,
in particular, the addition of planktivorous mosquitofish (Gambusia affinis)
to 2 of 4 replicate 700-liter freshwater microcosms. A number of studies have
been published that document the effects of increasing or decreasing a plankti-
vorous fish stock in natural water bodies (e.g., Hrbacek et al., 1961; Brooks
and Dodson, 1965; Postolkova, 1967; Straskraba, 1967), so that adequate infor-
mation exists to compare with the results of a similar trophic manipulation
in aquatic microcosms. Detailed biological and chemical monitoring of the
microcosms in this study permits such a comparison to be made, as well as
providing new information on the results of manipulating a planktivorous
trophic level.
-------
METHODS
Each of the 4 700-liter systems (designated A,B,C,D) consisted of a
fiberglass cylinder, 60.9 cm in radius, filled to a depth of 60.1 cm with
denrineralized water. The water was enriched with a modification of a common
freshwater algal growth medium (Woods Hole MBL; Nichols, 1973) and inoculated
with a 3.5 liter water sample collected from Lake Anza, a small eutrophic lake
in the Til den Park area of Berkeley, Ca. The enrichment levels of C, N, and
P were 0.21 mmol liter"1 NaHC03, 1.7 ymol liter"1 NaN03, and 3.4 ymol liter"1
Na2HP04 • 7H20, respectively, a situation conducive to eventual N limitation
of algal growth. The systems were maintained in a temperature controlled room
at 19±1 °C, illuminated by a bank of 8 4-ft high-output fluorescent lights
on a 12:12 light:dark cycle, and aerated at a rate of 1.2 liter min" . On day
35 (enrichment was on Day 0), 2 mosquitofish (Gambusia affinis] of length 2.5
cm were added to each of A and C.
As demonstrated in other experiments (Jassby et^ a]_., 1977), the initiation
conditions described above give rise to a phytoplankton bloom that is terminated
by zooplankton grazing within 2 months, after which zooplankton grazing and
phytoplankton growth remain loosely balanced for some time. Regular monitoring
of each parameter for this experiment began sometime between Day 48 and Day 65,
depending on the parameter, after the initial bloom had terminated and the
Gambusia were given the opportunity to adjust to their new surroundings. The
following parameters were measured twice per week: temperature, Og, pH, inor-
ganic carbon (1C), organic carbon (OC), NH^, N03 + N02, chlorophyll a_ (Chi a_),
phytoplankton species and numbers, zooplankton species and numbers, and bacteria
-------
-5-
numbers. Except for phytoplankton and zoonlankton, all measurements were
duplicated. Bacteria numbers were obtained with the Standard Plate Count
method (APHA, 1971), and OC was fractionated into dissolved (DOC) and parti-
culate organic carbon (POC) by measurement before and after filtration through a
precombusted Whatman 6F/C filter. Remaining analyses and the nutrient composi-
tion of the medium were as described elsewhere (Jassby et^ al_., 1977). Monitoring
was terminated on Day 100, except for bacterial numbers, which were followed
until Day 117.
-------
-6-
RESULTS
A detailed presentation of the temporal patterns that can be observed
in these microcosms appears elsewhere in conjunction with a different set of
experiments; the results of this trophic manipulation experiment thus are
reported in summary form only.
The temperature of the microcosms remained at 18±1 °C or approxi-
mately 1 °C below room temperature, presumably because of evaporative cooling.
i
02 concentrations fluctuated between 0.28 and 0.31 mmol liter . Aeration
apparently was sufficient to prevent any significant biological modification
of Op levels and most of the fluctuations can be ascribed to temperature varia-
tions in the microcosms. The pH values persisted at approximately pH 7.
The presence of Gambusia led to a significant decrease in the average
occurence of the large cladoceran Simoaephalus vetulus (Table 1), the adult
female form of which reaches lengths of 2 to 3 mm. The copepod Cyclops
vernalis (adult females approximately 1.5 mm in length) also decreased in
concentration, but this decrease was much less significant. The mean levels
of the small cladoceran Alona gu.tta.ta (adult females approximately 0.5 mm in
length) were similar both in the presence and absence of Gambusia. Among the
5 rotifers dwelling in the microcosms, only Leeane sp. responded significantly
to the introduction of Gambusia, in the form of an order-of-magnitude increase
in individual numbers. Changes in Anuraeopsis sp., Keratella ooohleavis,
Keratella quadrata, and Triohotria sp. were not significant.
Although Chi a_ concentrations were unaffected by Gambusia, total phyto-
plankton volume exhibited a large increase in the systems containing Gambusia
and ratios of phytoplankton carbon to Chi a_also were higher (Table 1). A
-------
-7-
total of 31 phytoplankton species were recorded but no obvious relationship of
community composition to the presence of Gambus-ia could be detected. The
phytoplankton communities were dominated by 2 forms of Synedra radians, which
was the dominant or second most dominant alga (by volume) in 79% of the 84
samples. Synedva ulna and a heterocyst-containing Andbaena sp. were next in
importance, being most or second most dominant in 29% and 20% of the samples,
respectively. In the 84 samples, the dominant and second most dominant algae
accounted for an average of 86% of the community volume. A light periphytic
growth on the sides of each container appeared by the end of the experiment
and consisted primarily, of Mougeotia sp.
Bacteria plate counts increased markedly in the systems containing
Gambusia (Table 1). Tony tarsus larvae were observed first on the container
sides of system C and apparently were present in the initial inoculum for C.
The Tony-tarsus spread via the adult flying stage to the remaining 3 systems
(first to B and D, then to A) where larval populations also become established.
The Gambusia in A and C were observed to feed on the larvae when they detached
from the container sides and swam to the surface in preparation for their
emergence as adults. The presence of Gambusia, however, did not alter signi-
ficantly the occurrence of the midges within the water column (Table 1).
Mean levels of 1C and DOC (Table 2) were unaffected by the presence of
fish. On the other hand, stocking of systems A and C with Gambusia led to
significant increases in POC and the ratio of POC to DOC, and to significant
decreases in NH4 and N03 + N02- It should be noted that POC values do not
include the carbon content of Gambusia. Accordingly, the results of Table 1
-------
-8-
indicate a shift of nitrogen from dissolved inorganic to particulate organic
form upon the addition of a higher planktivorous trophic level. An increase
in total N over the initial level of 1.7 ymol liter" NO., also can be deduced
from the results for each treatment.
The Gambusia individuals in A and C, although starved for 48 h before
addition, remained active and with normal coloration throughout the experiment.
No quantitative observations on their size change were collected, but qualitative
observation definitely suggested a size increase.
-------
-9-
DISCUSSION
B-Lological effects. The average levels of the chemical parameters demonstrated
good replicability within treatments (Table 2); this was not the case for the
biological parameters, especially the zooplankton (Table 1). Part of the
variability within pairs may be ascribable to the variable Tanytarsus coloniza-
tion. However, other experiments, in which variability remained even when
Tanytapsus larvae were stocked simultaneously in all systems, suggested that
poor replication may result from random deviations of protozoa numbers in the
initial inoculum (Jassby ejt al_., 1977). Whatever the case, the poor replication
entails that many of the potential treatment effects on zooplankton levels
may not be extractable from the data.
Nonetheless, certain biological responses to the presence of a planktivore
are" evident (Table 1), and are consistent with the following explanation:
the Gambusia fed on Simocephalus and reduced the average concentration almost
10-fold; Cyclops vernalis, being of smaller size, was fed on to a lesser extent,
while the smallest crustacean, Alona guttata, appeared to have been ignored by
the mosquitofish. The dominant zooplankters thus shifted from large cladocerans
and cyclopoid copepods to smaller cladocerans. Similar transitions have been
observed with heavy stocking of cyprinids in natural ponds (e.g., Hrbacek et al.,
1961) and the introduction of the alewife Alosa aestivalis into Crystal Lake,
Connecticut (Brooks and Dodson, 1965); and a transition in the opposite direc-
tion was observed when fish were excluded from the littoral areas of small
natural lakes (e.g., PoStolkova, 1967).
Despite the fact that they are larger than Simocephalus individuals,
the midge largae did not show a significant response to the presence of
-------
-10-
Gambusia. Most of the midge larvae were on the container sides where they
remained free from from Gambusia predation. The larval population in the
water column constantly was losing individuals that emerged from the water
surface and gaining individuals that detached from the container sides. The
resulting high turnover rate for larvae in the water column probably was
sufficient to disguise the effects of Gambusia predation.
Because simocephalus is capable of such high filtering rates (reported
up to 90 ml ind" d" ; Sushtchenia, 1958), the decrease of approximately 6 ind
liter in mean simoeephalus concentrations without a corresponding increase in
other crustaceans implies that much of the grazing pressure on phytoplankton and
bacteria was eliminated with the introduction of Gambusia. Indeed, the bacteria
plate counts increased by a factor of 3, and the phytoplankton community volume
by a factor of 5. According to one trophic state classification based upon
phytoplankton volume (Vollenweider, 1968), it may be said that the presence of
Gambusia resulted in a shift from an oligotrophic to a mesotrophic system. A
change in mean phytoplankton cell size was not observed in this study, although
smaller sizes in the presence of planktivores have been reported in some of the
studies referred to earlier.
In view of the large increase in mean phytoplankton levels where Gambusia
were present, the lack of any difference in Chi a_ values may appear somewhat
puzzling. However, these values were not corrected for phaeopignment inter-
ference and thus represent the fluorescence of grazed phytoplankton and other
forms of detritus (Currie, 1962), as well as of living algae. Accordingly,
the proportion of the Chi a_ attributable to living phytoplankton is least in
B and D, where more grazing activity presumably is taking, place. For similar
/
reasons, the ratio of phytoplankton carbon to Chi a_ appears to be lower in the
\
absence of Gambusia.
-------
-11-
The increased levels of bacteria and phytoplankton imply a larger food
source for rotifers, competition with crustaceans being ameliorated somewhat
by the smaller Sirhocephalus populations. Only the numbers of Leoane sp. (Table
1) definitely reflect the increased food supp_ly, but the mean value«.=of the rotifer
community volume increased from 0.013 mm3 liter"1 in B,C to 0.062 mm3 liter"1
in A,C. These results concur with the general observation from the above
studies with natural systems that rotifers increase with increasing fish stock.
Chemical effects. Because of the aeration, biological activity in this experi-
ment was not sufficient to alter the 1C levels after the initial bloom, either in
the presence or absence of the mosquitofish (Table 2). DOC also remained
unchanged, suggesting that the dissolved compounds produced were of a labile
nature that were processed quickly by bacteria before differences could arise
between treatments. The increase in POC in A,C demonstrates that the loss
of crustaceans in the presence of Gambusia was more than compensated for by
increases in the rotifer, phytoplankton, and bacterial biomasses. This is to
be expected, because the lowered grazing rates imply that a smaller amount of
of primary production will be lost to zooplankton respiration and excretion.
The POC:DOC ratio of 0.1 in B,D is more characteristic of natural systems
that the value of 0.3 observed in A,C (Wetzel and Rich, 1973). This probably
is a reflection of the fact that the concentration of Gambusia in A and C was
•3 1
approximately 5 mm liter wet weight, higher than that for phytoplankton
and a level that would arise only rarely in a natural mesotrophic system.
Similarly, the C:Chl a_ ratio of 21 observed in B, D is much more likely to
be encountered in natural systems (Paerl et_ al_., 1976) than the value for A,C
of over 100. The abnormal POC:DOC and C:Chl a_ ratios in A,C underline the
-------
-12-
fact that the effects attributable to the presence of mosquitofish can be
accepted only in a qualitative manner; it is not possible to stock laboratory
microcosms with realistic concentrations of planktovorous fish and obtain
effects that are acceptable quantitatively.
Lower levels of NH4 and NO., + N02 in A, C simply reflect, the fact that
POC (and, presumably, particulate organic nitrogen) increased when Gcaribusia was
stocked in the microcosms. However, a marked increased over the initial N
levels of 1.7 ymol liter" took place in all 4 systems. If we assume a C:N
molar ratio of 12 for the POC (typical of autochthonous detritus; Wetzel, 1975),
then the increase in mean total N ( = NH. + NO., + N02 + [POC/12] - 1.7) was about
7 ymol liter" in both treatments. As mentioned previously a heterocystous
Andbaena was the third most dominant alga in the microcosms and the possibility
exists that the additional N was provided by algal fixation. An average fixation
rate of only 0.004 ymol liter" h~ is required to provide an additional 10 ymol
liter" N by Day 100. Such rates are well within the range observed in nature
(label 11-3; Wetzel, 1975). The fact that the N increase was identical in
both systems is consistent with this explanation; Andbaena filaments, by virtue
of their size and mucilaginous sheaths, are not very susceptible to zooplankton
grazing and so their population levels probably are unaffected by the presence
of a planktivorous fish. Indeed, no significant differences were found in the
Andbaena concentrations between treatments.
Conclusion. In sum, we conclude that the response of the microcosms to stocking
of a planktivorous trophic level closely resembled the response of natural
systems to the same manipulation. As has been observed in naturally-occurring
-------
-13-
water bodies, introduction of planktivorous fish led to (i) a shift in the
dominant zooplankters from large to small cladocerans; (ii) an increase in phyto-
plankton resulting from lowered grazing pressure; and (iii) a higher rotifer
biomass in response to the increased phytoplankton. Additional responses
observed in this study included (iv) an increase in bacteria, presumably for
the same reason as the phytoplankton increase; (v) higher POC levels reflecting
smaller losses to zooplankton respiration and excretion; (vi) lower inorganic
N levels as the fish induce a net transition from inorganic to organic form; and
(vii) increased ratios of POC.-DOC and phyto C:Chl a_. The agreement with
observations from natural systems and the ability to suggest additional features
of the response to trophic structure manipulation clearly establishes the use-
fulness of these microcosms for research on trophic structure.
-------
-14-
REFERENCES
APHA. 1971. Standard methods for the examination of water and wastewater.
American Public Health Association.
Brooks, J,L., and S.I. Dodson. 1965. Predation, body size, and composition
of plankton. Science 150;28-35.
Cooke, G.D. 1971. Aquatic laboratory microsystems and communities, p. 48-85.
IQ J. Cairns (ed.), The structure and function of freshwater microbial
communities. Virginia Polytechnic Institute and State University.
Currie, R.I. 1962. Pigments in zooplankton faeces. Nature 193:956-957.
Draggan, S. 1976, The role of microcosms in ecological research. Intern.
J. Environmental Studies 10:1-2.
Hrbac'ek, J., M. Dvotakova, V. Korinek, and L. Prochazkova. 1961. Demon-
stration of the effect of the fish stock on the species composition of
zooplankton and the intenstiy of metabolism of the whole plankton associa-
tion. Verh. Int. Ver. Limnol. 1_4:192-195.
Jassby, A., M. Dudzik, J. Rees, E. Lapan, D. Levy, and J. Harte, 1977.
Production cycles in aquatic microcosms. Lawrence Berkeley Laboratory
report LBL-5965. 52 p.
Nichols, H.W. 1973. Growth media-freshwater, p.7-24. I_n_J.R. Stein (ed.),
Phycological methods. Cambridge University.
Paerl, H.W., M.M. Tilzer, and C.R. Goldman. 1976. Chlorophyll a_ versus
adenosine triphosphate as algal biomass indicators in lakes. J. Phycol.
12:242-246.
-------
-15-
y
Postolkova, M. 1967. Comparison of the zooplankton amount and primary
production of the fenced and unfenced littoral regions of Smyslov Pond.
Rozpravy Ceskosl. Akad. Ved, Rada Matern. Pn'r. Ved, 77:63-79.
Straskraba, M. 1967. Quantitative study on the littoral zooplankton of the
Poltruba Backwater with an attempt to disclose the effect on fish. Roz-
pravy Ceskosl. Akad.Ved, Rada Matem. Prfr. Ved, 77:7-34.
Vollenwider, R.A. 1968. Scientific fundamentals of the eutrophication of
lakes and flowing waters, with particualr reference to nitrogen and phos-
phorous as factors in eutrophication. OECD, Rep. DAS/CSI/68.27. 192 p.
Wetzel, R.G. 1975. Limnology. Saunders.
Wetzel, R.6., and P.M. Rich. 1973. Carbon in freshwater systems, p.241-263.
Jn^ 6.M. Woodwell and E.V. Pecan (eds.), Carbon and the biosphere. Proc.
Brookhaven Symp. in Biol. 24.
-------
-16-
TABLE HEADINGS
Table 1. Biological parameters monitored in the microcosms. N is the number
of times measurements of the parameters were taken in each system. Chi a_
and bacteria counts were duplicated. The entry for each pair of
systems (A,C and B,D) is the mean ± standard error of the average
value of the parameter for each member of the pair. The value of
t is the t-statistic for the null hypothesis that the 2 pairs of
systems (A,C with Gcaribusia and B,D without Gambusia} have equal
means.
Table 2. Chemical parameters monitored in the microcosms. N is the number
of times duplicate measurements of the parameter were taken in each
system. The entry for each pair of systems (A,C and B,D) is the
mean ± standard error of the average value of the parameter for each
member of the pair. The value of t is the t-statistic for the null
hypothesis that the 2 pairs of systems (A,C with Gambusia and B,D
without Gambusia} have equal means.
-------
Parameter
(units)
bacteria
(104 ml"1)
Phytoplankton
3 1
(mm liter )
Chi a.
(yg liter" )
phyto CrChl a1
Rotifera
(ind liter'1)
Ama>aeopsis sp.
Keratella cochleayis
Keratella quadrata
Lecane sp.
Triehotria sp.
Cladocera
(ind liter"1)
Alona guttata
Simoaevhalus vetulus
N
17
14
11
11
14
14
14
14
14
14
14
Mean ± Standard
A,C
16±4
3,4±0.1
5.4 ±2.0
113±27
0.12 ±0.00
4.5±4.4
130 ±120
42 ±3
23 ±31
20 ±4
0.80±0.71
Error
B,D
5.0±1.4
0.69 ±0.37
5.1 ±0.7
21 ± 7
0.59 ±0.83
66 ±93
6.2±6.7
6.4±6.4
0.60 ±0.85
15±7
6.8±1.0
t
3.90*
10.05**
0.20
4.68*
1.00
0.94
1.48
7.15**
1.00
0.82
6.96**
Copepoda
(ind liter"1)
Cyclops ve-malis 14 1.7±0.1 7.2±3.3 2.39
Diptera
(ind liter"1)
Tanytarsus sp. 14 0.59±0.83 1.8±0.8 1-41
* Null hypothesis rejected at the 80% level of significance
** Null hypothesis rejected at the 90% level of significance
Assuming that phyto C is 10% of wet weight
Table 1
-------
-18-
Parameter N Mean ± Standard error
(units) A,C B,D
1C , 12 0.21 ±0.04 0.23 ±0.04
(mmol 1 i ter~ )
DOC _, 11 0.32 ±0.01 0.31+0.01
(mmol liter" )
POC n 11 0.76 ±0.011 0.022 ±0.001
t
0.57
1.34
6.70**
(mmol liter" )
POCrDOC 11 0.29 ±0.01 0.090 ±0.003 19.61***
NH. 14 2.2±0.2 5.6±1.5 3.21*
i -i
(ymol liter)
NO, + N00 12 0.50±0.05 0.90 + 0.14 3.82*
-1
(ymol liter" )
* Null hypothesis rejected at the 80% level of significance
** Null hypothesis rejected at the 90% level of significance
*** Null hypothesis rejected at the 95% level of significance
Table 2
------- |