COMMUNITY SIMILARITY AND THE ANALYSIS OF MULTISPECIES ENVIRONMENTAL DATA - A UNIFIED STATISTICAL APPROACH

The number of species or variables in some designed environmental studies is too large for analysis using standard inferential statistics. To analyze the type of data, a two-stage procedure and follow-up methods are described. First, data are reduced using a measure of similarity or distance. Then, a permutation procedure is used to make inferences. Assuming the hypothesis is rejected, several follow-up analyses are presented as tools in understanding the causes of rejection. For example, to further understand treatment effects, a randomization based, multiple comparison procedure may be used; to better understand what differences the similarity or distance measure describes, a method based on removal of a species is given. The methods are illustrated on a study of the effects of zinc on the periphyton community in the New River, Va, U.S.A. © 1990.