We gathered climatic, topographic and edaphic data from various sources, encompassing 26 variables (Table S3). The 19 bioclimatic variables and altitude data were acquired from Worldclim 2.1, which offers both current and future bioclimatic variables, with a spatial resolution at 2.5 arc–minutes, covering the periods from 1970–2000 and 2081–2100 (Fick & Hijmans, 2017). We extracted slope and aspect data from the Digital Elevation Model (DEM) Using spatial analysis tools in ArcGIS 10.8 (Esri, Redlands, CA, USA). Organic carbon stocks and total nitrogen were sourced from the World Soil Information (WoSIS) database (Batjes et al., 2020). Forest variables were downloaded from the Global Climate Analysis and Modeling (GCAM–Demeter) database (Chen et al., 2020). Additionally, the anthropogenic variable (human footprint index) was obtained from the Human–Footprint database (Venter et al., 2016). All the environmental variables were resampled to a resolution of 2.5 arc minutes. To avoid multicollinearity among the variables, we used the ‘vifstep’ function in the usdm R package (Naimi et al., 2014) to remove strongly correlated climate factors with VIF (variable inflation factor) > 5, and all variables were given in Table S3. Then, the environmental variables were extracted based on the study area generated by alphahull R package (Pateiro-López & Rodríguez-Casal, 2010).