FIGURE 1 Study areas and occurrence records of Francois’ langur in the present study
2.2 Environmental variables and data processing
A total of 26 environmental variables, including climate (19 bioclimatic variables), topography (elevation, slope, aspect), vegetation (Normalized Difference Vegetation Index (NDVI)), water sources (distance to river) and human activities (distance to road, distance to housing), were provisionally selected to build the initial models based on previous research on the habitat of Francois’ langur.
19 grid-based bioclimatic variable data were downloaded from the WorldClim (www.worldclim.com) dataset. The current climate data were obtained with a spatial resolution of 30 arc-second (approximately 1km2) (Li et al., 2022). Future climate data were obtained from the Beijing Climate Center Climate System Model (BCC-CSM) under the Coupled Model Intercomparison Project Phase 6 (CMIP6) global climate model in four periods of 2021-2040, 2041-2060, 2061-2080, 2081-2100 with the shared socio-economic pathways 245 (SSP 245 for moderate scenario). These climatic variables represent annual trends (mean annual temperature and precipitation), seasonality (annual range of temperature and precipitation), and limiting environmental factors (temperature and precipitation of a certain quarter) (Hijmans et al., 2005; Thapa et al., 2018). As the 19 bioclimatic variables were all extended from temperature and precipitation based on different calculations, to avoid the effects of model overfitting caused by multicollinearity of these variables (Zhao et al., 2020), the 19 predictor climatic variables used in this study were tested using Pearson correlation analysis with the ’Cor’ function in R 4.3.0 (Team, 2014) (Appendix 1), and variables with high correlation coefficients (| r |≥0.8) and relatively low contribution rates in the initial models were removed to improve model predictability (Ji et al., 2020; Liu et al., 2022).
Three topography variables including elevation, slope and aspect were derived from Digital Elevation Model (DEM) data, which was obtained from geospatial data cloud (http://www.gscloud.cn) with a resolution of 30m using ArcGIS 10.2. NDVI was downloaded from the Resources Environment Data Center, Chinese Academy of Sciences (http://www.resdc.cn) with a resolution of 500m. River, road and housing data were obtained from the National Basic Geographic Information System, and the Euclidian distance tool of the ArcGIS 10.2 was used to calculate the distance between each grid cell layer and the nearest river, road and housing.
Consequently, five predictor climate variables were retained, namely, the Mean Diurnal Range (Mean of monthly (max temp-min temp)) (Bio2), Min Temperature of Coldest Month (Bio6), Temperature Annual Range (Bio7), Precipitation Seasonality (Coefficient of Variation) (Bio15), Precipitation of Coldest Quarter (Bio19), additionally, elevation, slope, aspect, NDVI, distance to river, distance to road, distance to housing were also used in the model prediction (Table 1). To facilitate spatial analysis within China, all variables selected for modelling were projected to WGS_1984_UTM_Zone_48N and resampled to 30 arc-second (approximately 1 km2) using the spatial analysis tools in ArcGIS 10.2.
TABLE 1 Environmental variables included in the analysis and their percentage contribution and permutation importance in predicting the current distribution of Francois’ langur