2.6 | Statistical analysis
All statistical analyses were conducted in R 4.0.2 (R Core Team, 2020). We used microbial OTU richness as an index of microbial diversity. A t-test was used to evaluate the differences in microbial richness between rhizosphere and non-rhizosphere. Differences in microbial community composition between the two regions were evaluated using permutational multivariate analysis of variance (PERMANOVA) with the function “adonis” in the package “vegan” with 999 permutations (Oksanen et al., 2020), and visualized by the principal coordinate analysis (PCoA). One-way ANOVA followed by Tukey-Test was used to assess the effects of grazing intensity on microbial richness, composition (i.e., the relative abundance of main microbial taxa), biotic variables (shoot biomass, root biomass, root total carbon, and root total nitrogen of C. songorica ), and abiotic variables (concentrations of total carbon, nitrogen, and phosphorus, concentrations of available nitrogen and phosphorus, soil pH and soil moisture). Prior to analyses, Shapiro-Wilks was used for normal distribution evaluation, while Bartlett’s test for homogeneity of the data. The dissimilarity in microbial community composition was determined by the Bray-Curtis distance among treatments, performed with the “vegdist” function in package “vegan” (Oksanen et al., 2020).
A Mantel test was used to evaluate how changes in microbial community diversity or composition are related to biotic and abiotic variables by the “mantel test” function in the package “vegan”, and visualized by the package “linkET” (Huang et al., 2021). The multicollinearity of variables was assessed based on variance inflation factors (VIF), and those with a VIF value less than 10 were regarded as variables with low collinearity. The relative importance of variables in affecting microbial richness and composition was evaluated with the package of “rdacca.hp” (Lai et al., 2021).