Loess plateau mismatch between ecosystem service supply and demand driven by climate change and human activity

Balancing the increasing demand for ecosystem services (ESs) with their limited supply is a critical global challenge for human society. However, the drivers shaping the spatial heterogeneity of the comprehensive supply-demand balance (CSDB) remain unclear. To address this, we developed an integrated framework synthesizing multi-source datasets to assess ES supply-demand balance, analyze coupling coordination and trade-offs/synergies, and identify driving factors. Focusing on the Loess Plateau (LP), from 2000 to 2020, we analyzed the spatiotemporal dynamics of four ESs: water yield (WY), carbon sequestration (CS), soil conservation (SC), and food production (FP). Furthermore, interpretable machine learning was employed to identify dominant CSDB drivers. Results show: (1) While the overall supply of all four ESs increased, demand was significantly concentrated in urban areas and their peripheries. This led to persistent and intensifying deficits in WY, CS, and FP in urban centers, whereas forest-covered regions largely maintained a balance or surplus. SC remained in surplus across most areas, with only minor localized deficits in the Plateau Wind-Sand Region (PWSR). Consequently, the CSDB exhibited a pattern of high surplus in forest zones and deteriorating deficits in urban areas. (2) The Coupling Coordination Degree (CCD) of multi-service supply-demand balances was generally higher in the central region and decreased from southeast to northwest. Approximately 51% of the area was in a state of serious imbalance. Significant differences were observed between supply-demand balance pairs: WY-CS exhibited strong synergy; WY-SC, CS-SC, and CS-FP showed weak synergy; while WY-FP and SC-FP were dominated by trade-offs. (3) The spatial heterogeneity of the CSDB was primarily explained by human activity-related variables in the model (contribution >70% in most subregions), with population density showing a substantial negative contribution. Precipitation served as a secondary key factor, promoting CSDB surplus. Additionally, five supply-demand balance clusters were identified to propose spatially differentiated governance pathways. This study provides a scientific basis and theoretical foundation for alleviating ES supply-demand imbalances, optimizing regional resource allocation strategies, and promoting sustainable ecosystem management practices on the LP. • Pioneered explainable ML to quantify ESs dynamics, trade-offs, and drivers in LP, enhancing interpretability. • Uncovered 2000–2020 ESs' spatiotemporal heterogeneity, LP mismatch between ecosystem service supply and demand driven human-climate factors. • Crafted strategies via five ESs clusters for sustainable resource allocation optimization in LP.