Diversity indices, community comparison indices and canonical discriminant analysis: Interpreting the results of multispecies toxicity tests

Abstract

Canonical discriminant analysis, diversity indices and community comparison indices were evaluated to determine their utility in quantifying macroinvertebrate response to a complex effluent in laboratory microcosms. The Shannon and Simpson diversity indices and the following community comparison indices were analyzed: (1) Renkonen's percentage similarity; (2) Bray-Curtis index; (3) average χ 2 index; (4) Stander's SIMI index; (5) Pinkham-Pearson index; and (6) Canberra metric index. Statistical inferences were made with the diversity measures using an ANOVA in conjunction with Duncan's test. A permutation or randomization procedure was used to test the hypothesis of no treatment effect based on the community comparison indices. The canonical discriminant analysis was primarily influenced by the drop in mayfly densities in the higher dose microcosms and indicated that the high-dose microcosms were different from the other microcosms. The Shannon and Simpson diversity measures indicated that both the medium- and high-dose microcosms were significantly different ( P ⩽ 0.05) from the control microcosm, but this response was due solely to the increase in density of chironomids. When the diversity indices were calculated without chironomids, they failed to reflect adverse effects on mayflies in the medium-dose microcosms. The statistical inferences for all community comparison indices were similar and indicated that the high dose microcosms were significantly ( P ⩽ 0.06) different from the control. However, the Bray-Curtis index provided the most meaningful condensation of data.