While direct toxicant effects on populations can be estimated from single-species bioassay results, the complete range of population and community responses can best be investigated at the ecosystem level. The objectives of this research were (1) to determine the effects of a contaminant (a coal-derived light distillate oil) on the zooplankton communities of laboratory pond microcosms and outdoor experimental ponds, (2) to compare responses of microcosm and field zooplankton communities with each other and with laboratory bioassay data, and (3) to explore the use of community structure data for impact detection.

The responses of the microcosm and pond zooplankton communities to oil treatment were quite similar. Their close agreement lends support for using shallow-water pond microcosms as surrogates for field experiments in hazard evaluation. Changes in cladoceran densities were the most sensitive indicators of stress in the zooplankton communities. Copepods were slightly less sensitive, and rotifers were least sensitive to oil treatment. Indirect effects on populations were observed at intermediate treatment levels. The lowest treatment levels at which persistent and significant ecological changes occurred were the same in both systems. These levels were similar to the lowest observed-effect concentration (LOEC) for the Daphnia magna chronic bioassy, or approximately 3% of the D. magna 48-h LC₅₀.

Community structure responses were generally consistent between systems; however, sensitivity for detecting oil treatment effects varied greatly among the measures of community structure. Species richness and diversity measures were not as sensitive as individual species responses. Cluster and multivariate post-clustering analyses of community similarity values reflected indirect effects and had sufficient sensitivity to allow early impact detection. An expanded principal components technique, which used both zooplankton and water quality data, was also relatively sensitive and would be applicable for routine monitoring.