Fig. 2.

COG analysis of predicted proteomes. a Hierarchical clustering of the study isolates based on the proportional abundance of COG categories in their predicted proteomes. Abundance is indicated by the blue colour intensity in the heat map. b PCA of Bray-Curtis dissimilarity matrices calculated from the proportional abundance data. Dissimilarity between the four subgroups is significant (pairwise ADONIS, p < 0.05, Bonferroni adjusted). Subgroup affiliations of isolates are indicated by colour as shown in the legend between the panels