Statistical Analyses

All analyses were conducted with R (versions 3.4.3, R Development Core Team, 2015) using the packages “vegan” (Oksanen et al., 2017), “stats” (R Development Core Team, 2018) and “MASS” (Venables and Ripley, 2002). Figures were generated mainly by “ggplot2” package (Wickham, 2009) and “plotrix” (Lemon, 2006).

Three-way ANOVA using non-parametric tests was used to assess the effects of the site, habitat, cushion treatments and their interactions on Shannon diversity index and relative abundance of different taxonomic groups. Differences in microbial community composition among treatments were investigated with principal coordinate analysis (PCoA, “vegan” R package) based on Bray-Curtis dissimilarities using three-way non-parametric multivariate analysis of variance (NPMANOVA, “vegan” R package, Anderson, 2001) with 999 permutations. Variance partitioning analyses were conducted on the three-way ANOVA of microbial Shannon diversity and the three-way NPMANOVA of the microbial composition obtained from OTUs to quantify the relative contribution of each explanatory factor to soil microbial community variances (“stats” R package, “MASS” R package). Based on the tested abiotic factors (Jiang et al., 2018, Supplementary Figure S3), the site effect was used to quantify the large scale abiotic effects, which were mainly related to soil pH and abiotic hydrothermal conditions. The habitat and the habitat × site interaction effect were used to quantify the small scale abiotic effects (microhabitat effects) because different phenotypes and their paired-open microsites occurred in different local environmental conditions. Additionally, this microhabitat effects were mainly related to soil nutrition contents. The cushion and the site × cushion interaction effect was used to quantify the cushion presence effects (large-scale biotic effects), whereas the habitat × cushion and site × habitat × cushion interaction effects were used to quantify the cushion phenotypic effects (small-scale biotic effects) (Table 1).

Abiotic and biotic factors at different scales.