Using XGBoost-SHAP for understanding the ecosystem services trade-off effects and driving mechanisms in ecologically...

Understanding the spatial and temporal variability of Ecosystem services (ES), along with the trade-offs and synergies among different services, is crucial for effective ecosystem management and sustainable regional development. This study focuses on Wensu, Xinjiang, China, as a case study to address these challenges.

ES and their trade-offs were systematically assessed from 1990 to 2020. Explainable machine learning models (XGBoost-SHAP) were employed to quantify the nonlinear effects and threshold effects of ES trade-offs, with specific attention to identifying their driving factors.

(1) From 1990 to 2020, water yield (WY) and soil conservation (SC) exhibited an inverted "N"-shaped downward trend in Wensu County: mean annual WY decreased from 22.99 mm to 21.32 mm, and SC per unit area declined from 1440.28 t/km² to 1351.3 t/km². Conversely, windbreak and sand fixation (WS) showed an "N"-shaped increase from 2.32×10⁷ t to 3.11×10⁷ t. Habitat quality (HQ) initially improved then deteriorated, with values of 0.596, 0.603, 0.519, and 0.507 sequentially. (2) Relationships between WY-WS, WY-HQ, WS-HQ, SC-WS, and SC-HQ were primarily tradeoffs, whereas WY-SC interactions were synergistic. Trade-offs for SC-HQ, WY-HQ, and WS-HQ were stronger, while WY-SC trade-offs were weaker. (3) The XGBoost-SHAP model revealed land use type (Land), precipitation (Pre), and temperature (Tem) as dominant drivers of trade-offs, demonstrating nonlinear responses and threshold effects. For instance, WY-SC trade-offs intensified when precipitation exceeded 17 mm, while temperature thresholds governed WY-HQ trade-off/synergy transitions.

This study advances the identification of nonlinear and threshold effects in ES trade-off drivers. The model's interpretability in capturing these complexities clarifies the mechanisms underlying ES dynamics. Findings are generalizable to other ecologically vulnerable regions, offering critical insights for ecosystem management and conservation strategies in comparable environments.