Fig. 2. Metagenomic diversity of fecal and oral samples from uncontacted Yanomami in relation to other human groups.

(A) Functional diversity (average ± SD) measured by the observed number of KEGG orthologs in fecal samples of Yanomami and Guahibo Amerindians, Malawians, and U.S. subjects. Metagenomic tables rarefied at 1 million sequences per sample. Interpopulation differences were significant for all comparisons (P < 0.001, ANOVA and Tukey’s HSD). (B) PCoA plot based on Bray-Curtis distances calculated on the metagenomic table of fecal samples rarefied at 1 million sequences per sample. (C) Top discriminative metabolic pathways in Yanomami fecal samples as determined by STAMP; pathways ranked by effect size (η²). (D) Normalized prevalence/abundance curves for all KOs found at 1% abundance or more in fecal samples. (E) Functional diversity (average ± SD) measured by the observed number of KEGG orthologs in oral samples of Yanomami and Guahibo Amerindians, Malawians, and U.S. subjects. Metagenomic tables rarefied at 900,000 sequences per sample. Interpopulation differences were not significant (P = 0.07, t test). (F) PCoA plot based on Bray-Curtis distances calculated on the metagenomic table of oral samples rarefied at 900,000 sequences per sample. (G) Top discriminative metabolic pathways in Yanomami oral samples as determined by STAMP; pathways ranked by effect size (η²). (H) Normalized prevalence/abundance curves for all KOs found at 1% abundance or more in oral samples. (I) Functional diversity (average ± SD) measured by the observed number of KEGG orthologs in skin samples of Yanomami and U.S. subjects. Metagenomic tables rarefied at 1 million sequences per sample. Interpopulation differences were significant for all comparisons (P < 0.001, ANOVA and Tukey’s HSD). (J) PCoA plot based on Bray-Curtis distances calculated on the metagenomic table of skin samples rarefied at 1 million sequences per sample. (K) Top discriminative metabolic pathways in Yanomami skin samples as determined by STAMP; pathways ranked by effect size (η²). (L) Normalized prevalence/abundance curves for all KOs found at 1% abundance or more in skin samples.