Identifying the driving mechanisms of ecological environment risk in the Mongolian plateau: A study based on the XGBoost–SHAP model and GTWR
Year: 2026
Type: article
Abstract: Ecological risk poses a critical constraint on the sustainability of arid and semi-arid ecosystems. Conventional assessment approaches typically rely on static indicators or global models, which tend to overlook nonlinear responses and spatiotemporal heterogeneity in risk formation. To address this limitation, this study integrates the Ecological Risk Index (ERI) with XGBoost–SHAP and Geographically and Temporally Weighted Regression (GTWR), enabling explicit quantification of nonlinear effects and spatiotemporal non-stationarity. Using multi-source data from 2000 to 2020, we examine the spatial patterns and driving mechanisms of ecological risk across the Mongolian Plateau (MP). The results show that: (1) ecological risk exhibits a stable south–north gradient, with higher risk concentrated in western and central Mongolia and lower risk in Inner Mongolia. (2) XGBoost–SHAP reveals pronounced nonlinear relationships and clear threshold effects, and further highlights distinct national-level differences in driving mechanisms, with ecological risk in Inner Mongolia primarily controlled by natural environmental constraints, whereas in Mongolia it is more strongly influenced by land use and human activities. (3) GTWR captures significant spatial heterogeneity, with weakening precipitation effects and intensifying anthropogenic pressures in specific regions. By integrating nonlinear machine learning with spatially varying regression, this study provides a more interpretable and spatially explicit framework for ecological risk assessment, offering targeted insights for land-use optimization and ecosystem management in arid regions.
Source: Ecological Indicators
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
SDGs: Life on land, Climate action
Open Access status: gold
Estimated APC paid: $3,650
OpenAlex ID: W7166527342
