Page 763 - The Toxicology of Fishes

Aquatic Ecosystems for Ecotoxicological Research 743

and fish community structure. Decreases in the number of insects in this study were correlated with
reductions in aquatic macrophytes and associated algae and increases in fish predators. Insect responses
were not a direct effect of the pesticide. The wide range of chemical, structural, and biotic interactions
dependent upon macrophyte type and density, as outlined above, emphasizes the central role of this part
of the community in lentic systems. It is apparent that the design of surrogate ecosystems needs to
consider plant density and diversity as a contributor to system variability and the inability to detect
ecosystem changes.

Fish

Whether to include fish, what species or complex of species to select, the loading rates, and their potential
for reproduction are critical factors to consider in experimental design. Fish populations are known to
have direct and indirect effects on ecosystem functioning. Fish predation is known to alter plankton
community composition (Brooks and Dodson, 1965; Drenner et al., 1986; Vinyard et al., 1988), and the
presence of fish in limnocorral or microcosm experiments may alter nutrient dynamics and cycling
(Mazumder et al., 1988, 1989). During an outdoor microcosm experiment, Vinyard et al. (1988) found
that filter-feeding cichlids altered the quality of nitrogen (shifting the dominant form) and decreased
limnetic phosphorus levels via sedimentation of fecal pellets. Additionally, unequal fish mortality among
replicate microcosms may influence nutrient levels independently of any other treatment manipulations
(Threlkeld, 1988). In separate limnocorral studies, Brabrand et al. (1987) and Langeland et al. (1987)
concluded that fish predation alters planktonic communities in eutrophic lakes and that the very presence
of certain fish species may contribute to the eutrophication process. These studies offered a number of
interesting hypotheses regarding fish effects in limnetic systems; unfortunately, the experimental designs
of these studies lacked treatment replication, limiting their inferential capability.
Many studies completed in 1986 through 1992 in the United States, under EPA guidelines(Touart and

Slimak, 1989) for pesticide studies, required that mesocosms include a reproducing population of bluegill
sunfish (Lepomis macrochirus Rafineque). Presumably, these fish and their offspring are integrators of
system-level processes. Variances in numbers, biomass, and size distribution among different pesticide
exposure levels provide requisite endpoints for risk management decisions. Chemical registration studies
by Hill et al. (1994b), Giddings et al. (1994), Johnson et al. (1994), Morris et al. (1994), and Mayasich
et al. (1994) have determined that the abundance of young bluegill in mesocosm experiments obscured
or complicated the evaluation of pesticide impacts on many invertebrate populations. This is consistent
with Giesy and Odum’s (1980) suggestion that higher trophic levels assert a controlling influence on

lower trophic levels in microcosms being used for effects studies. Ecological research with freshwater
plankton and pelagic fish communities indicates that both top-down and bottom-up influences affect
planktonic community structure and biomass (Carpenter et al., 1985; McQueen and Post, 1988; Threlkeld,
1987). These relationships have not been investigated to the same degree in littoral zone communities,
and the role of benthic macroinvertebrates in these trophic relationships requires further study. Along
these lines, Deutsch et al. (1992) stocked largemouth bass in pond mesocosms to control unchecked
bluegill population growth, thereby potentially limiting among-system variability and providing a more
natural surrogate system. The desirability of adding bass to mesocosms, however, must be balanced
against possible increases in experimental error variances that may result from differential predation on
bluegill if variable bass mortality occurs in the ponds (Stunkard and Springer, 1992). The only way to
control variability in predation of bluegill would be to maintain equal levels of predator mortality in all
ponds. Scaling is important, and criteria for fish stocking levels are highly dependent on system size.
Fish population density should not exceed the carrying capacity of the test system. Biomass densities
should generally not exceed 2 g/m (Fairchild et al., 1992). It may be useful to stock mesocosms with
3
low adult densities and remove adults and larvae after spawning; however, the life stage, number and
biomass of fish added will depend on the purpose of the test. If the emphasis is on an insecticide, for
example, larval fish might be added to monitor their growth in relation to the invertebrate food base.
The FIFRA requirement of using a single species (e.g., bluegill sunfish) in mesocosm experiments was
very likely not sufficiently protective of natural fish communities for a number of reasons. First, the
inherent sensitivity of other fishes compared to bluegill is not known with any degree of certainty. Second,