3.1 Species richness patterns of all species of rodents
Rodents are widely distributed in China and are recorded in almost every
grid; however, their spatial distribution is uneven. All species (n =
237) of rodents occurred in 1665 grids, and the species richness in each
grid was between 0 and 84 (mean: 31.39 ± 16.51 SD) species (Figure 1a;
Table S5). The highest abundance of rodent species in China was found in
the subtropical and tropical regions of the oriental realm, with the
Hengduan Mountains being the most abundant region, followed by the
Qilian Mountains and Tianshan Mountains regions. In addition, Taiwan and
Hainan Island also have high species richness. The species richness was
the lowest in the Qinghai-Tibet Plateau and Tarim Basin regions, with
only a few species in most grids.
The simple regression results showed that, when considering the effect
of individual factors, all species richness was highly significantly
correlated with each environmental factor (p < 0.001).
The two most relevant variables were VEG and AET (Figure 2a, b; Table
S6).
Regarding the relationship between the predictor set and all species
richness, the best model was explained by a set of five variables
(EW1+CS1+HH1+HE1+HE2). The multiple regression model (OLS) explained
71% of the total variation in rodent species richness. Due to the
removal of the effect of spatial autocorrelation by the spatial
autoregressive model (SAR), the degree of explanation was reduced to
64% (Table 1). As shown in Table 1, the importance of each variable in
explaining the species richness pattern varied slightly across the
regression models, but the difference was not significant. Both the
regression models showed that HH1 and HE1 were the most important
predictors. The variance partitioning results showed that the four
predictor sets explained 71.7% of the variance in the total species
richness. The habitat heterogeneity predictor set explained 48.94% of
the variation in all species richness patterns, followed by human
factors (42.30%), energy-water (28.79%), and climatic seasonality
(18.30%) (Figure 3a, b; Table S7)