Ecological risk from anthropogenic stressors and their key drivers exhibit scale dependence

Robust and spatially explicit ecological risk assessments (ERAs) are essential for informing effective risk mitigation strategies. However, existing studies have insufficiently explored spatial variability in the interactions among multi-level drivers of ecological risk. Integrating Bayesian Belief Networks (BBNs) with geographically weighted regression (GWR), this study proposes a spatially explicit framework for assessing ecological risk from multiple anthropogenic stressors, and applies it to the Beijing–Tianjin–Hebei (BTH) region of North China. Results indicate that cumulative stressor risk, aggregated from agricultural production, non-cropping livelihood activities, and urbanization, exhibits pronounced spatial clustering, with urbanization acting as a key leverage point in shaping cumulative risk. These high-risk clusters delineate priority areas for conservation and targeted mitigation. These findings highlight the importance of explicitly accounting for compound stressors and incorporating finer-scale socioeconomic indicators to enhance the robustness and policy relevance of ERAs. Spatial analyses further reveal strong scale dependence in ecological risk and its drivers, while the effects of anthropogenic stressors on risk exhibit pronounced spatial heterogeneity, reflecting the combined influence of localized socioeconomic processes and broader regional linkages. Based on these insights, differentiated risk mitigation strategies tailored to dominant local stressors are proposed. Overall, this study demonstrates the value of integrating probabilistic modeling with spatially adaptive regression techniques for capturing compound stressor effects and spatial heterogeneity in ERAs, thereby enhancing the policy relevance of ERAs for sustainable land use planning and regional risk governance.