Dynamic Changes in the Potential Suitable Habitat of Caragana korshinskii Under Climate Change Based on a Biomod2 Ensemble Model

Projecting the spatiotemporal dynamics of the potential distribution of dominant species under climate change is essential for desertification control and vegetation restoration in drylands. Here, we modeled the current (1970–2000) and future (2080–2100) suitable habitats of Caragana korshinskii Kom, an ecologically important shrub species in northwestern China, by constructing an ensemble of eight species distribution models on the Biomod2 platform using three CMIP6 Shared Socioeconomic Pathways (SSP126, SSP370, SSP585) and 40 environmental variables representing climate, soil, topography and drought conditions. Key environmental drivers were identified through variable importance ranking and response curves, while area changes, spatial patterns, and centroid shifts in suitable habitats were quantified. The ensemble model demonstrated good to excellent predictive performance (mean AUC > 0.9, mean TSS > 0.5). Soil base saturation (t-bs) and soil moisture contributed the most (>38%), highlighting the dominant role of edaphic factors. The current total suitable habitat of C. korshinskii is approximately 182.2 × 104 km2, with all future scenarios projecting a consistent decline. Under SSP585, habitat loss reached 9.8% with contraction (30.5 × 104 km2) far exceeding expansion (12.6 × 104 km2). The distribution centroid shifted markedly eastward with a minor southward fluctuation, establishing the Ordos–Bayannur region as a stable core habitat. Overall, our findings suggest that the distribution of C. korshinskii is strongly constrained by edaphic and moisture conditions, and future contraction of marginal habitats may compromise ecosystem services.