Results
Zooplankton species richness was similar in all four experiments, with the ten taxa that had the highest average density across experiments found in all experiments (Table 1). The mean and standard deviation of the density of taxa varied greatly among experiments (Fig. 1, Table 1). The density ranks also varied among experiments; e.g., the most common (rank 1) species in one experiment could be anything from the 2nd to the 14th most common taxa in another experiment. Nevertheless, there were some strong similarities among experiments. For example, the rank of some taxa (Chydorus , cyclopoids and calanoids) did not vary by more than 4. Further,Ceriodaphnia was the first or second most common species in Exps. 1-3 (but much less common in Exp. 4 in which Daphnia andBosmina were particularly common). We also report occupancy of each taxon as the percentage of tanks in which it was found. Only in one case was the occupancy of a common taxon (i.e. found in all four experiments) less than 70% (in Exp. 1, Bosmina was found in 54% of the tanks). Variation in community composition of zooplankton across experiments was much larger than the variation within an experiment, as shown by the significant effects of experiment (p = 0.0001) in the permutation multivariate analysis, and illustrated in Fig. 2, in which the NMDS values differ among experiments.
Caged fish had an effect on zooplankton community composition that varied across experiments, as evidenced by the significant experiment x caged fish interaction (p = 0.0036), but no effect of caged fish when averaged across the four experiments (p = 0.47). This varying effect of caged fish on community composition is evident in Fig. 2, as the NMDS values of the caged-fish treatment were observed to be higher, lower, or not different than the values of the no-fish treatment in different experiments (e.g. for NMDS dimension 3, the fish treatment is higher than the no-fish treatment in Exp. 1, but lower than the no-fish treatment in Exp. 4).
Caged fish had a significant effect on the abundance of some taxa, and in some cases the direction of that effect varied among experiments (bottom box, Table 1). For two taxa all three possible effects occurred: the presence of caged fish increased, decreased, and had no effect on the density of calanoids and Chydorus . For four taxa two possible effects occurred: fish increased or did not affect the density cyclopoids, Bosmina and Scapholeberis ; fish decreased or did not affect the density of ostracods. There were also several cases where there was a significant positive or negative effect of fish on a taxon in one experiment, and a nonsignificant trend in the opposite direction in another experiment. For example, there is a strong negative fish effect in Exp. 4 on ostracods, and the “no effect” result in Exp. 1 shows a trend in the positive direction with small enough within-treatment variation that the influence is unlikely in the negative direction. In all cases in which there was a positive or negative effect of fish, the taxa were observed in all tanks in both treatments. For four taxa fish did not have a significant effect in any of the experiments (Alona , Ceriodaphnia , D. pulex , and Diaphanosoma ).