A nonlinear threshold-based study on ecological zoning for water conservation in Guangdong Province, China

Understanding the spatial heterogeneity and nonlinear drivers of water conservation is essential for sustainable ecosystem management. This study integrates the InVEST model, XGBoost–SHAP, and GIS analysis to quantify water conservation, identify threshold responses of key drivers, and delineate ecological zoning in Guangdong Province, China. Results show that water conservation increased slightly from 2000 to 2010 but declined by 2020, with high values persisting in the northern mountains and decreasing continuously in the Pearl River Delta. Slope and precipitation were the dominant drivers, with clear thresholds (3–11°, >1600 mm), whereas population and GDP showed negative nonlinear effects beyond critical limits. Moderate landscape heterogeneity was associated with higher conservation capacity. The threshold-based zoning framework identified stable protection, refining, and enhancement areas, revealing intensified spatial polarization and providing an adaptive basis for water conservation monitoring and spatial management in rapidly urbanizing regions.