Revealing the nonlinear driving mechanisms and scenario simulation pattern of landscape ecological risk in Jilin Province using XGBoost-GeoShapley and PLUS model

Deciphering the spatiotemporal dynamics of land use and landscape ecological risk (LER) is critical for advancing regional sustainable development. Focusing on Jilin Province, China, this study quantifies land use transitions and associated LER patterns from 1990 to 2020. By integrating landscape metrics, spatial statistics, and geographic explainable artificial intelligence (XGBoost-GeoShapley, eXtreme Gradient Boosting and GeoShapley), we systematically characterize the spatiotemporal heterogeneity of LER and explore the nonlinear driving mechanisms. Moreover, we integrate environmental and socioeconomic data from multiple sources with the Patch-generating Land Use Simulation (PLUS) model to forecast land use and LER trajectories across various future scenarios. The findings reveal that: (1) Between 1990 and 2020, land use in Jilin Province consisted primarily of farmland and forest land, which collectively represented more than 80% of the overall surface area; LER were mainly classified as low and medium risk, making up over 75% of the total, with high-risk zones predominantly located in the western part of the province. (2) Projections for LER in 2030 through multi-scenario simulations suggest that, within the natural development framework, the high-risk and medium-high risk areas make up 14.0%, continuing the historical rise in LER; in contrast, the ecological protection scenario reveals that high-risk areas only constitute 0.8%, the lowest among all four scenarios; under the farmland protection framework, high-risk areas represent 7.4%; while in the economic development scenario, both medium high-risk and high-risk classifications account for over 35%, which is the highest risk level across the four scenarios. (3) The effects of the driving factors are nonlinear and involve complex interactions, with the 500–1000 m elevation range serving as a key threshold interval for the region. These factors also exhibit significant spatial effects, and the driving role of a single factor varies across different geographical environments. The findings provide guidance for sustainable development and regional ecological security in Jilin Province.