Answers are grounded in this paper and cite their sources.
Understanding the spatiotemporal dynamics of island ecosystem health under global change is essential for achieving regional sustainability. However, few studies have systematically quantified how natural and anthropogenic drivers jointly shape island ecosystem health, especially in tropical coastal regions. This study addresses this gap by developing an integrated and interpretable framework that combines the Vigor–Organization–Resilience–Services (VORS) model with explainable machine learning to assess the human–natural coupling mechanisms of Hainan Island, a typical tropical biodiversity hotspot in the northern South China Sea. Multi-source remote sensing and socioeconomic datasets (2015–2020) were used to construct the Ecosystem Health Index (EHI) through analytic hierarchy weighting and Random Forest modeling. To enhance model transparency and interpretability, Increase in Mean Squared Error (IncMSE), Break Down (BD) plots, and Partial Dependence Plots (PDP) were employed to quantify the relative importance and nonlinear effects of multiple drivers. Results show that Hainan Island's overall ecosystem health remained at a moderately high level (EHI >0.5) during 2015–2020, with stable conditions in the central mountainous rainforest region and localized degradation in urbanized coastal zones. Human activity intensity—particularly GDP and nighttime light—exhibited a progressively increasing negative influence, with GDP importance rising from 0.163 (2015) to 0.238 (2018) and NLD importance more than doubling over the same period. Climatic factors such as temperature and precipitation presented threshold-like nonlinear responses, and these thresholds remained temporally stable across all six years. Two-variable interaction analysis revealed that the negative ecological effect of GDP is strongly modulated by elevation, with lowland areas (<200 m) experiencing approximately twice the ecological cost of economic development compared to highland areas (>600 m), and this disparity intensified over the study period. The explainable modeling framework reveals that anthropogenic pressures have surpassed certain ecological resilience limits in low-elevation coastal areas, whereas natural regulation remains dominant in inland rainforests. These findings provide a robust analytical pathway for diagnosing island ecosystem vulnerability and guiding spatially adaptive restoration and management strategies under the “dual pressures” of climate change and socioeconomic development.
No takes yet. Share an insight, caveat, or question.
Fang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: