<p>Understanding the spatial and temporal variations of ecosystem service bundles (ESBs) and their underlying social-ecological factors is crucial for effective ecosystem management. Previous studies have focused on the supply side of ESBs, neglecting the integration of supply-demand relationships into bundle analysis. This study quantifies the ecosystem service supply-demand ratio (ESDR) for six key services—crop production (CP), water yield (WY), carbon storage (CS), air purification (AP), habitat quality (HQ), and outdoor recreation (OR)—in the Yangtze River Delta (YRD) region of China for the years 2005 and 2020. Using a self-organizing feature map, we identified four distinct ESDR bundles (ESDBs) with homogeneous supply-demand states. Gradient boosting decision tree (GBDT) models were further adopted to explore the relative importance and nonlinear effects of social-ecological drivers on ESDRs in each bundle. Results indicate significant decreases in ESDR values for CP, CS, AP, HQ, and OR. ESDB1 was predominantly located in mountainous areas, ESDB2 was concentrated in agricultural regions with extensive cultivated land, ESDB3 was situated in the central YRD, and ESDB4 was primarily distributed in urban areas. Additionally, cultural services responded mostly to socio-economic variables, while regulating services responded mostly to climate and topographic variables. The supply-demand relationships in each bundle exhibited potential threshold impacts and nonlinear trajectories in response to key drivers. The study offers specific recommendations for ecosystem management, suggesting prioritized ecological protection in ESDB1, optimized land-use planning in ESDB2, a balanced approach between ecosystem services and economic activities in ESDB3, and enhanced ecological space quality in ESDB4.</p>