Fig. 3

Functional diversity and relatedness between industrialized and non-industrialized communities. a Heatmap showing the z-scores of read abundances of the ECs with high weights in the top principal components. Standard Ward-linkage hierarchical clustering of the EC profiles of industrialized and non-industrialized microbiomes was performed using Pearson correlation. The two top-level clusters found by hierarchical clustering perfectly capture the separation of non-industrialized and industrialized microbiomes. For display purposes, we show only individuals with read abundances falling outside one standard deviation of the mean in at least nine of the highly variable ECs. See Additional file 4: Figure S6 for the corresponding heatmap and clustering on all individuals. b Heatmap showing co-abundance association across core metabolic pathways. Co-abundance associations between pathways wee calculated as the pairwise Kendall rank correlations between the pathway abundance profiles (obtained using Carnelian-generated EC profiles) of microbiomes from both communities considered together. Ward-linkage hierarchical clustering was used to partition the pathways using Euclidean distance, generating either 2, 3, 4, or 5 clusters. Although hierarchical clustering can be used to identify clusters of co-abundance pathways between the non-industrialized vs industrialized communities, the clusters were not significantly different from each other with respect to the industrialized/non-industrialized label (PERMANOVA test p values > 0.05). Thus, in contrast to the top-level EC label clustering from a, the partitions are not simply recapitulating the industrialized/non-industrialized labels