Ecosystem service supply–demand and socioecological drivers at different spatial scales in Zhejiang Province, China

•ES supply and demand under different scales were assessed.•There was a distinct spatial mismatch between ES supply and demand.•ES trade-offs/synergies were enhanced as the scale increased.•The impacts of social-ecological drivers on ESs varied with scale.•Some management suggestions were proposed to improve ES management. Understanding the scale effects of ecosystem service (ES) supply–demand balances and drivers is critical to hierarchical ecosystem management. However, it remains unclear how the relationships of ES supply–demand and driving factors change with the scale. In this study, we first quantified food production (FP), water yield (WY), soil conservation (SC), carbon storage (CS), and habitat quality (HQ) at pixel and county scales in 2000 and 2020 in Zhejiang Province. Then, we analyzed the ES supply–demand balances and trade-offs/synergies at different scales. Finally, we performed correlation analysis and applied a random forest model to explore the socioecological drivers of these ESs. Our work showed that the supplies of FP, WY, and SC increased, while those of CS and HQ decreased from 2000 to 2020. ESs at the pixel scale were more spatially heterogeneous than those at the county scale. FP and CS were in short supply, and the gaps between their supply and demand grew over time. Some ES supply–demand mismatches at the pixel scale disappeared at the county scale. From the pixel scale to the county scale, the correlation directions of the ES trade-offs/synergies changed slightly, but their intensities changed significantly. The temperature, altitude, percentage of forestland and normalized difference vegetation index (NDVI) had positive effects on HQ, CS and SC, while the population density (POP), gross domestic product and percentage of artificial land (PA) had negative effects. The degree of influence of most socioecological drivers on the ESs increased with increasing scale. NDVI was the most important factor for CS, while precipitation was the most important for WY. The importance of POP and PA increased with both time and scale. Ultimately, overall ES supply–demand balances should be considered at the county scale, while more accurate management measures should be implemented at the pixel scale to promote effective hierarchical ES management. This study emphasizes the necessity of considering the scale effects for ES supply–demand balances in sustainable ecosystem management.