Multimodel ensemble projections of Shorea robusta habitat suitability in Nepal under current and future climates
Abstract
Shorea robusta Gaertn. (sal) is an ecologically and economically important tree species that forms the dominant forest type in the lowland and mid-hill regions of Nepal. Understanding the habitat suitability of sal under current and future climatic conditions is important for supporting conservation and forest management planning. This study assessed habitat suitability for sal in Nepal using ensemble species distribution modelling based on presence-only data and 31 environmental variables. Four algorithms-Generalized Linear Model (GLM), Random Forest (RF), Boosted Regression Trees (BRT), and Maximum Entropy (MaxEnt)-were implemented using a 70/30 training–testing split and 10 bootstrap replicates. Model performance was generally high, with RF, BRT, and MaxEnt contributing most strongly to the final ensemble projection. The results indicate that suitable habitat for sal is currently concentrated in the Terai and lower Siwalik regions, closely matching its known present distribution. Under the 2070 RCP8.5 scenario, the projection indicates lower suitability in parts of western Nepal and higher suitability in parts of northern and eastern Nepal compared with the baseline current-ensemble projection. Minimum temperature of the coldest month (Bio6) emerged as the most important predictor, followed by temperature seasonality (Bio4), precipitation of the coldest quarter (Bio19), and terrain roughness. These findings provide scenario-based information on sal habitat suitability under current and projected future climatic conditions and can support climate-informed conservation, restoration, and forest management planning in Nepal.