Species mean elevational range size pattern
Consistent elevational hump-shaped patterns of species mean range size were found across different measurements (guided by the lowest AICc value, elevational range size pattern better fitted by a cubic equation than a linear or quadratic equation) (Table 1). Species mean range size increased with elevation, reached a peak at mid-elevation, and then decreased (Figure 3). The Steven’s method was the best-fitted method with the largest R ² (R ²=0.80, P <0.01 Figure 3a) value compared to the midpoint method (R ²=0.78, P <0.01 Figure 3b) and the specimen method (R ²=0.45, P <0.01 Figure 3c). The species mean elevational range size calculated by the Steven’s method was used in the multiple linear regression models explaining species mean elevational range size.
Drivers of species mean elevational range size
The elevational pattern of ATR increased monotonically; NDVI increased rapidly at low elevation, with a stable plateau at middle elevation, then decreased slowly with the increasing elevation; HH increased with fluctuations and peaked at middle elevation (Figure S2 in Appendix 2).
The best models (delta AICc<2) showed that ATR and NDVI were the most significant explanatory variable. ATR was negatively correlated with species mean elevational range size, and NDVI was positively correlated with species mean elevational range size (Table 2).
HH was a relatively weak explanatory variable of species mean elevational range size, and the area percentage of suitable habitat types was irrelevant with HH along elevation (R ²=0.17, P =0.04; Figure S3a in Appendix 2). The area percentage of suitable habitat types peaked at 2,350 m to 2,950 m a.s.l. and 3,650 m to 4,050 m a.s.l. (Figure S3b in Appendix 2), where the values of HH were the lowest.