Applying Bayesian network models to identify and optimize ecological security zoning—from the perspective of ecosystem service tradeoffs

Facing intensified climate change and anthropogenic pressures, the Luo River Basin has experienced ecological space shrinkage and ecosystem degradation. This study quantitatively assessed water yield (WY), soil conservation (SC), and carbon sequestration (CS) services in the basin, analyzed their spatiotemporal tradeoffs using root mean square error (RMSE), identified key driving factors through a Bayesian network model (BBN), and constructed four scenarios for ecological security identification and risk zoning, with high-risk zones determined by the optimal state subset. Results show (1) All ecosystem services (ESs) showed increasing trends with distinct "west-high-east-low" and "south-high-north-low" spatial patterns, where high-value clusters occurred in peripheral high-altitude and central regions, while low-value zones followed the Luo River and its tributaries. (2) Drivers exhibited heterogeneous impacts on ES tradeoffs, showing state-dependent effects but temporal stability. (3) Bayesian multi-scenario analysis indicates that the scope of high-risk areas continued to expand with the superposition of driving factors, reaching 4.48 × 103 km2 in Scenario 3, an increase of 3.55 × 10³ km2 compared to Scenario 1. Priority restoration areas were identified in Lushi, Song County, Luoning County, Yiyang County, Yichuan, and Gongyi County. Precision strategies for the Luo River basin support ecological restoration and regional sustainable development.