SLM multivariate models for the residuals of species richness of all mammals and its habitat groups.

Six variables that explained most of the variance of the residuals of species richness were selected based on univariate regression models and hierarchical partitioning. We established the best multivariate model using multivariable GLM regression. To avoid inflation of type I errors and invalid parameter estimate owning to spatial autocorrelation, we then performed SLM multivariate regression (see Methods). All continuous variables were log10-transformed (n = 2376; *: Pr(>|z|)|z|)|z|)SLM multivariate models for the residuals of species richness of all mammals and its habitat groups.