Influence of taxonomic resolution, biological attributes and data transformations on multivariate comparisons of rocky macrofaunal assemblages

Abstract

Multivariate comparisons of impacted and non-impacted biotic assemblages require decisions to be made about the taxonomic resolution to be used, biological attributes to be measured and whether the contributions of common, or combinations of common, intermediate and rare taxa should be emphasised. This study shows how these decisions affect comparisons of exploited and non-exploited rocky infratidal macrofaunal assemblages distributed across a biogeographic transi- tion zone on the south-east coast of South Africa. Two-way crossed ANOSIM tests were used to test for differences in the structure of assemblages among regions of the coast and between exploited and non-exploited localities. In tests based on presence/absences, significant differences in assemblages among regions and between treatments were evident only at the species level. In tests based on den- sities and biomasses, regional differences were evident at the species, order, class and phylum level, and were irrespective of whether dominant or combinations of dominant, intermediate and rare taxa were emphasised. Significant differences in exploited and non-exploited assemblages were evident in all of the comparisons except those in which numerically dominant higher taxa were emphasised. These results do not suggest that the natural environmental gradient influences the fauna more by species replacements than by changes to the proportions of higher taxa, nor do they indicate that natural environmental variations and anthropogenic disturbances modify macrofaunal assemblages at different taxonomic levels. Further studies are needed to establish whether the taxonomic level, at which modification of assemblages occurs, varies with the type of disturbance (e.g. pollution vs exploitation, strong vs weak gradients) or biotic component (e.g. rocky vs soft-bottom assemblages) under consideration.