Statistical and Bioinformatics Analyses

KEGG Orthologous (KO) distributions were normalized as counts per million reads (CPMR) calculated as (R/N)^∗10⁶, where R is the number of mapped reads to a gene and N is the total number of reads of that sample. One-way analysis of similarity (ANOSIM) and principal coordinates analysis (PCoA), using Morisita index, were performed with PAST software version 3.15². The significance of the relative abundance difference in KO was performed using Kruskal–Wallis and multiple test correction via Benjamini and Hochberg false discovery rate, which are implemented in the STAMP software version 2.1.3³. Relative abundance of bacterial genera in each sample was obtained by dividing the number of reads mapped to all the contigs classified to a genus by the total number of reads in each sample. ANOVA was used to test whether there are significant differences between the relative abundance of bacterial genera from those consortia. Contigs were assigned to the last common ancestor where at least 30% of the genes had USEARCH hits (Huntemann et al., 2016). Only contigs with 10 or more genes were used for this analysis. A network of association between bacterial genera and samples was visualized using Cytoscape v3.0 software.