Diagnosing Ecosystem Service Imbalance: A Machine Learning‐Driven Dissection of Supply–Demand Bundle Dynamics in the Dongting Lake Basin

ABSTRACT The Dongting Lake Basin, a critical wetland ecosystem in the middle Yangtze River, has experienced pronounced changes in ecosystem service (ES) provision and consumption under the combined stresses of rapid urbanization and hydrological shifts. However, the understanding of the spatiotemporal evolution and driving mechanisms of supply–demand‐based ecosystem service bundles in river–lake connected systems remains insufficient. Selecting the years 2000, 2010, and 2020 as temporal benchmarks, this study constructed a supply–demand indicator system for six ESs—water yield (WY), soil retention (SR), habitat quality (HQ), urban cooling (UC), carbon sequestration (CS), and flood regulation (FR)—to reveal their spatiotemporal dynamics. We uncover a fundamental spatial mismatch: high‐value supply zones are concentrated in peripheral ecological lands, while intensifying demand is radially clustered in expanding urban agglomerations. Average WY increased from 554.6 mm in 2000 to 780.4 mm in 2020. Mean SR supply in high‐value zones rises from 177.67 to 233.49 t/ha. In contrast, average CS declined from 555.75 to 490.06 gC/m 2 , while UC and FR also exhibited decreasing trends. On the demand side, urbanization drove a marked expansion of high‐demand zones. WY and CS demands were highly clustered in urban agglomerations, with mean CS demand increasing from 9.08 × 10 4 t to 26.78 × 10 4 t. UC and HQ demands first rose and then declined, with high‐demand areas expanding centrifugally along with construction land. Through Self‐Organizing Map (SOM) model, we identified five ES bundles: low‐intensity composite, urbanization‐dominated, wetland‐buffer, ecological protection‐prioritized, and agricultural production‐oriented types. The XGBoost‐SHAP interpretable model further revealed that nighttime light intensity, land‐use structure, NDVI, and climatic factors have bundle‐specific influences, underscoring the joint forcing of natural settings and human activities in shaping ES spatial patterns. This study provides spatially‐explicit and quantitative evidence and policy insights for balancing ES supply and demand advancing fine‐scale land‐space governance in the Dongting Lake Basin.