2.5 Statistical analysis
Statistical analysis was conducted on the ASV level of theclr -transformed 16S rRNA dataset. Prokaryotic community profiles were constructed at the phylum level, and the relative abundance of major phyla (> 0.1 % of the total microbial community) in environmental samples was compared between sites by Analysis of Compositions of Microbiomes with Bias Correction (ANCOM-BC) (Lin & Peddada, 2020). Alpha diversity metrics including the abundance-based coverage estimator (ACE), diversity (Shannon diversity index), and evenness (Inverse Simpson’s diversity index) were computed. A comparison between each of the alpha diversity indices by sites and source was performed by two-way ANOVA followed by a post-hoc Tukey honestly significant difference (Tukey HSD) test after ensuring data normality and homogeneity with tests as above.
To address the community variation between samples collected from different sites and sources, we adopted a permutational multivariate analysis of variance (PERMANOVA) analysis and Analysis of Similarity (ANOSIM) with 999 permutations to compare their compositions. A distance matrix based on the ASVs was first constructed with the distancefunction in the “vegan” R package (Oksanen et al., 2022) with the Euclidean distance (Gloor et al., 2017). Then, the Adonis method via theadonis function with 999 permutations was used for the comparison of communities. Pairwise PERMANOVA was carried out withpairwise.adonis with the “pairwiseAdonis” package (Arbizu, 2023). Permutation tests for homogeneity of multivariate dispersions (PERMDISP) were conducted using betadisper and permutestto verify significant PERMANOVA outcomes. Principal Component Analysis (PCA) was used to visualize the beta diversity matrix in the “phyloseq” package (McMurdie & Holmes, 2013).