Risk-Informed Ecological Network Optimization in a Semi-Arid Coal Mining Landscape

Coal mining landscapes require restoration strategies that account not only for where ecological risk is concentrated but also for how risk constrains landscape connectivity. However, landscape ecological risk assessment is still commonly used as a zoning tool, with weak links to resistance surface parameterization and node-level restoration. Using the Shenmu coal mining area in northern China as a case study, we developed a risk-informed ecological network framework based on multi-source spatial data from 1995 to 2020. The framework combined landscape ecological risk assessment, GeoDetector-based driver analysis, ecological source screening, resistance surface construction, minimum cumulative resistance modeling, a gravity model, and circuit theory-based node diagnosis. Landscape dominance showed the highest explanatory power within the tested factor set (q = 0.06083), followed by land use type, water body proximity, and landscape fragmentation, while most factor interactions showed bivariate or nonlinear enhancement. Risk zoning delineated ecological conservation (467.62 km2), enhancement (1434.86 km2), and restoration areas (2566.49 km2). The framework identified 10 ecological sources; 18 potential corridors with a total length of 213.18 km; and 89 key nodes, including 52 pinch points, 4 barrier points, and 33 fracture points. The main contribution of this framework lies not in combining established ecological network tools, but in transferring ecological risk information into resistance surface parameterization and linking different types of critical nodes to differentiated restoration priorities. These outputs should be interpreted as model-based structural and potential functional connectivity priorities, rather than as direct evidence of realized species movement.