Figure 2.

Statistical analyses of bacterial populations in low and high disease plaque. A) Principal coordinate analysis of bacterial communities from subgingival plaque of West Virginians with low oral disease (blue) as compared to plaque of West Virginians with varying degrees of oral disease (red). B) The Unifrac algorithm was used to compute the unique branch length for a given sub-sample. Cluster analysis split low (blue) and diseased samples (red) into different clades. Support for the clusters was evaluated by jackknife tests (1000 replicates). T was the least diseased sample in the Cognitive study, as ranked in Table 2. The 'D' has been removed from sample notations in these figures for clarity.