Page 754 - The Toxicology of Fishes

734 The Toxicology of Fishes

(APHA). Consensus protocols provide guidance on the application of toxicity tests for assessing the
toxicity of single chemicals, complex effluents, and ambient samples of water or sediment. Many fish
species are used in laboratory toxicity tests (see Chapter 15 in this text). Such tests typically use one
species, and have a duration of 96 hours or longer (APHA, 1999). Embryo–larval development tests and
a 7-day larval growth and survival test have been developed using Pimephales promelas to represent
freshwater fish; marine fish are represented by the sheepshead minnow (Cyprinodon variegatus), a species
indigenous to the Atlantic and Gulf coasts (USEPA, 1994).
Several research studies have used model aquatic ecosystems of varying design and complexity to
evaluate contaminant fate and effects in aquatic ecosystems. Such systems are designed to simulate
ecosystems or portions thereof. As research tools for fish toxicity testing, model ecosystems contribute
to our understanding of the manner in which contaminants affect fish populations in aquatic ecosystems
(Crossland et al., 1993). If designed well, these systems allow ecologists to address hypotheses on a
manageable scale and with control or reference systems. They also provide ecotoxicologists with models
of ecosystem functioning in the absence of perturbation so direct and indirect effects might be better
separated from natural events such as succession or inherent variation (Crow and Taub, 1979). Tradi-
tionally, model ecosystems are referred to as either microcosms or mesocosms. The distinction between
microcosm and mesocosm is subjective and is mainly a function of size or volume (Giesy and Allred,
1985). Degrees of organizational complexity and realism often vary in these systems, depending largely
on study goals and endpoints.
Giesy and Odum (1980) defined microcosms as “replicable, artificially bounded subsets of naturally

occurring environments.” Microcosms can contain several trophic levels and exhibit system-level prop-
erties. Mesocosms are defined as either physical enclosures of a portion of a natural ecosystem or
manmade structures such as constructed ponds or stream channels (Voshell, 1990). Mesocosm size and
complexity should be of sufficient size for the system to be self-sustaining and hence suitable for long-
term studies of fish growth and reproduction (Voshell, 1990). In this regard, they differ from microcosms,
where smaller size and fewer trophic levels do not allow for long study durations, particularly with fish.
Cairns (1988), however, did not distinguish between microcosms and mesocosms because “both encom-
pass higher levels of biological organization and have high degrees of environmental realism.” Lack of
a defined distinction between microcosms and mesocosms has caused some confusion among researchers
around the world. The European Workshop on Freshwater Field Tests operationally defines microcosms
3
on the basis of size, defining outdoor lentic microcosms as surrogate ecosystems of volume ≤15 m and
3
mesocosms as ponds of 15 m or larger. Experimental stream channels were also characterized on the
basis of size. Microcosms are defined as smaller and mesocosms larger than 15 m in length. In this
chapter, we define model systems based on the European Workshop on Freshwater Field Testing
(EWOFFT) definitions.

Mesocosms as Model Ecosystems
The use of model systems in aquatic research has grown considerably since the use of replicated ponds
in community structure analysis in the late 1960s by Hall et al. (1970) and the pesticide studies of
Hurlbert et al. (1970). Aspects such as community composition and spatial heterogeneity can be con-
trolled to a greater extent in model (constructed) systems relative to natural ones. Model ecosystems are
logistically more manageable and replicable for statistical analyses. In addition, model systems are
effective tools in aquatic research because they allow a focus on important cause-and-effect pathways
expected to occur in natural systems (Cairns, 1988; Odum, 1984). For fish, model ecosystems can retain
a high degree of environmental realism relative to laboratory single-species toxicity bioassays (Cairns,

1988). Single-species tests are inadequate when chemical fate is altered significantly under field condi-
tions, when organismal behavior can affect responses to a toxicant, or when indirect effects occur due
to alterations in competitive or predator-prey relationships (La Point et al., 1989). Mesocosm tests,
however, should be viewed as part of a tiered testing sequence and not as replacements of single-species
bioassays (Cairns, 1989). The ability to detect and accurately measure responses of fish populations in
mesocosms can be influenced by system characteristics and experimental designs that influence vari-
ability. If fish growth and reproduction are to be endpoints of interest in mesocosm testing, all key