Spatiotemporal evolution of ecosystem health and its driving factors in the southwestern karst regions of China

Abstract

As one of the most typical karst landscapes globally, the karst regions in southwestern China are characterized by prominent rocky desertification and fragile ecological conditions. Consequently, exploring the spatiotemporal evolution and driving influences on ecosystem health (EH) in this region is of great significance for the improvement of ecosystems and green development. This study focuses on assessing EH in these regions from 2000 to 2020 using the “vitality-organization-recovery-service” (VORS) framework. Spatiotemporal changes in EH are analyzed through hotspot analysis, and the functional relationship between driving factors and EH is quantified using XGBoost and SHAP models. Key findings include: (1) Over the past two decades, the proportion of cities experiencing enhanced EH has generally improved in 73% of cities compared to 27% experiencing deterioration. (2) Spatial analysis reveals EH clustering in three regions. One cold spot cluster in the central north and two hot spot clusters in the southwest and southeast. (3) Urbanization level exhibits an inverse logarithmic relationship with EH. Precipitation affects EH in a cubic polynomial pattern—initial decrease, subsequent increase, and final decrease. Temperature impacts EH through a quartic polynomial function with fluctuating increases and decreases. PM2.5 shows a monotonically decreasing relationship with EH, while the normalized difference vegetation index demonstrates a monotonically increasing association. This research contributes to understanding EH dynamics in southwestern China’s karst landscapes, crucial for advancing ecosystem management and sustainable development efforts.

Keywords

• Ecosystem health
• Spatiotemporal evolution
• Driving factors
• VORS
• Machine learning