Assessing thresholds and spatial effects of drivers on multi-scale ecosystem service supply-demand matching

Analyzing multi-scale ecosystem service (ES) supply-demand matching, identifying social-ecological drivers (SEDs) and their thresholds, and exploring spatial differentiation are crucial for ES regulation. However, neglecting scale effects in SED thresholds may impede policy precision. This study emphasizes these scale effects by identifying driver thresholds across spatial heterogeneity at multiple scales. We analyzed ES supply-demand matching across four scales (1 km and 3 km grids, town, county) in Bashang from 2000 to 2020. Drivers were selected using a social-ecological framework. A generalized additive model (GAM) identified key thresholds of the ES supply-demand ratio (ESDR), and multi-scale geographically weighted regression (MGWR) revealed spatial heterogeneity. We quantified three supply-demand ratios: carbon sequestration (CS_SDR), soil conservation (SC_SDR), and water yield (WY_SDR). Results show that as scale increased, ESDR patterns became more regionally clustered, masking local deficits and yielding apparent regional surplus. At smaller scales, key drivers were primarily ecological, whereas at larger scales, economic and social factors dominated. Nonlinear thresholds were more pronounced at larger scales. In 2020, drivers operated at smaller scales than in 2000 and 2010. At the 1 km grid, precipitation drove CS_SDR and WY_SDR, while NDVI dominated SC_SDR, each with a single threshold. At the county scale, dominant factors shifted to slope for CS_SDR, PM10 for SC_SDR, and NLI for WY_SDR, with one or two thresholds. Notably, the same driver could have opposite effects in different regions. This study clarifies multi-scale driving mechanisms of ES supply-demand and supports multi-tiered ecological security optimization and cross-scale sustainable ES management. • Different ecosystem service demand-supply matching (ESDR) is scale-specific. • The social-ecological driver (SED) framework for ESDR is constructed according to regional context. • At grid scale, ecological factors are more important for ESDR; at administrative scale, it is social factors. • Larger spatial scales lead to clearer thresholds of social-ecological drivers for ESDR. • The same driver at different time nodes acts on different spatial scales and exhibits obvious spatial heterogeneity.