2.10. Amplicon Sequences Variants Abundance Enrichment Analysis

In order to determine how gut microbiome compositional variation among threespine stickleback in Oregon was influenced by host population genetic structure and environment, we first looked broadly at what bacterial groups were present in our gut samples. Successively, we quantified AVS abundance differentiation across the major geographic partitions. We did this by converting the phyloseq to a DESeq dataset using the phyloseq_to_deseq2 function in the DESeq2 v1.18.1 R package [50]. The function “estimateSizeFactors” does not handle ASV count values = 0. In order to handle zero values, we applied: gm_mean = function (x, na.rm = TRUE)^{Σ(log(x[x > 0])}, na.rm = na.rm/length(x), where x is ASV counts. The “estimateDispersions” function was used to estimate dispersions. We tested gut ASVs for significant differential abundance both between inland and coastal freshwater populations and between estuary and freshwater populations using the DESeq function with the default Benjamin–Hochberg multiple-inference correction and with the fitType parameter set to ‘local’. Only ASV differential abundances that were statistically significant (p ≤ 0.05) were considered in our biological interpretation of these results.