Standardized slopes, standard error, and 95% CI for effects of the four environmental predictors (dots and thick/thin bars) from multiple linear regressions and generalized additive models (GAMs), which accounted for spatial autocorrelation with a spline-based trend surface term of geographical coordinates. Note that predictors were z-scaled for comparison. Detailed model estimates shown in (D) are provided inTables 1andS2. For the effects of interactions between environmental drivers and the proportion of burrowing species on species richness seeTable S3.