Revealing Multiscale Characteristics of Ecosystem Service Flows: Application to the Yangtze River Economic Belt

Assessing ecosystem service (ES) supply–demand relationships and identifying their driving forces are essential for ecological security and sustainable ecosystem development. Using ES supply–demand mismatches as a basis, this study characterized the spatiotemporal evolution of ES supply and demand from 2000 to 2023. Additionally, a SHAP-informed Stacking Bayesian optimization model was employed to identify key drivers of supply–demand imbalances. Building on this, threshold-aware spatial optimization of ecosystem service flows was performed using an improved minimum-cost algorithm within an NSGA-II multi-objective framework. The results showed that: (1) The YREB’s supply–demand balance (SDB) exhibited significant spatial heterogeneity. Water SDB declined with fluctuations, decreasing from 5.343 × 10 11 m 3 to 4.433 × 10 11 m 3 , whereas carbon SDB shifted from a surplus (+1.514 × 10 9 t) to a deficit (−1.673 × 10 9 t) during the study period. Crop SDB rose from 1.361 × 10 8 to 1.450 × 10 8 t across the study period. (2) Nighttime light intensity (NLI) was the dominant factor for water SDB and carbon SDB, while cropland area was the key driver for crop SDB. (3) Over 2000–2023, water SDB flow increased from 8.5 × 10 9 m 3 to 1.43 × 10 10 m 3 . Carbon SDB flows more than tripled from 9.576 × 10 7 tons to 2.89 × 10 8 tons. Crop SDB flow increased nearly twelvefold over 2000–2023, from 3.3 × 10 5 t to 3.93 × 10 6 t. The findings provide scientific support for coordinating ecological conservation and high-quality development across the Yangtze River Economic Belt.