Use of Cluster and Discriminant Analyses to Compare Rhizosphere Bacterial Communities Following Biological Perturbation

We present an approach to comparing the diversity and composition of bacterial communities from different habitats and for identifying which members of a community are most affected by an introduced bacterium. We use this method to explore both previously published and new data from field and growth chamber experiments in which we isolated heterotrophic bacteria from samples of root-free soil, roots of nontreated soybean seedlings, and from the roots of soybean seedlings grown from Bacillus cereus UW85n1-treated seeds. We characterize bacterial isolates for 40 physiological attributes, and grouped the isolates hierarchically using two-stage density-linkage cluster analysis. Multivariate analysis of variance and discriminant analysis of the relative frequencies of the clusters in the soil and rhizosphere habitats were then used to determine whether there were differences among the bacterial communities from the various habitats, and which of the clusters were most useful in discriminating among the communities. We used rarefied estimates of richness as a measure of community diversity in the various habitats. Introduction of UW85n1 affected the composition and/or diversity of rhizosphere communities in three of four experiments.