Long-term landscape risk dynamics and human–environment interactions in the Qiantang River Basin (1980–2020)

Understanding the long-term dynamics of landscape ecological risk is crucial for sustainable territorial management in rapidly urbanizing basins. This study investigates the spatiotemporal evolution and driving mechanisms of ecological risk in the Zhejiang section of the Qiantang River Basin (QRB-ZJ), a composite geomorphic watershed spanning mountains, hills, plains, and estuaries. Using multi-temporal land use data (1980–2020) derived from remote sensing, we constructed a Landscape Ecological Risk Index (LERI) and applied spatial autocorrelation, the GeoDetector, and an interpretable XGBoost–SHAP model to identify key drivers. The results indicate that (1) the expansion of construction land and the reduction in cultivated land have significantly increased fragmentation and reduced connectivity; (2) ecological risk exhibits a phased pattern of “increase–stability–decline,” with high-risk clusters concentrated along urban and coastal belts; and (3) land use intensity and socio-economic activity are dominant drivers, while topography and precipitation modulate spatial heterogeneity. A “fragmentation–exposure–connectivity weakening” risk chain is identified in low-elevation coastal zones, indicating that blue–green network integrity, rather than water proximity, determines ecological resilience. The proposed multi-model framework offers a transferable approach for assessing and mitigating ecological risk under long-term land use transitions and provides scientific support for sustainable basin governance and spatial planning in coastal China and similar geomorphic systems worldwide.