The alpha-diversity indices and Bray-Curtis similarities between samples were computed using Primer 6 (Plymouth Marine Laboratory, UK). Jaccard similarity index, bacterial community compositions and relative abundance comparisons were analyzed by R software. One-way analysis of community similarity (ANOSIM) of different layers and non-metric multidimensional scaling (NMDS) were performed in Primer 6. A standard Mantel test at each sampling layer was run in R to evaluate the correlation between horizontal geographic distance and bacterial community compositions (based on the abundances of the top 50 OTUs). Redundancy analysis (RDA) was implemented with Monte Carlo permutation tests to analyze variations in the bacterial assemblages under the constraint of environmental factors by Canoco (Version 5, Microcomputer Power). Significant environmental parameters without multicolinearity effects (variance inflation factor < 20) were obtained to explain community variability31 (link). BIOENV and BVSTEP analyses were then performed in order to statistically determine correlations between community compositions and environmental variables.
Free full text: Click here