Spatiotemporal evolution and drivers of eco-environmental quality for sustainable rural tourism in China’s Yangtze River Basin: an explainable AI approach

Abstract

Balancing tourism-driven growth (SDG 8.9) with terrestrial ecosystem conservation (SDG 15) is vital for sustainable watershed management. Focusing on 409 Key Rural Tourism Villages in China’s Yangtze River Basin, this study integrates Google Earth Engine and MODIS data to assess eco-environmental quality (EEQ) from 2000 to 2024 using the Remote Sensing Ecological Index (RSEI), employing an XGBoost-SHAP framework to decode driving mechanisms. Results indicate that these villages exhibit a clustered spatial distribution pattern, with three core agglomerations. The mean RSEI across all villages exceeds 0.73, indicating overall improvement and resilient recovery after 2020. However, a distinct north-south spatial disparity emerged: improved villages were predominantly concentrated north of the Yangtze River mainstem, whereas degraded villages clustered in the Taihu, Poyang, and Dongting Lake basins to the south. Elevation, land-use change, and temperature are the primary drivers, with a critical altitudinal threshold near 2,500 metres, where synergistic effects turn antagonistic. Although the XGBoost model achieved strong predictive performance (R2 = 0.853, RMSE = 0.071), 14.7% of the variance remains unexplained, suggesting the influence of additional unmeasured factors. These findings advance SDG 15 localisation by establishing spatial baselines and threshold-guided carrying-capacity regulations for watershed tourism governance.