Nonlinear associations between comprehensive ecosystem services and four compound climate extremes: a comparative analysis across China from 2000 to 2020

Ongoing climate change has intensified compound climate extremes, increasing the long-term exposure of ecosystems to compound climatic stress and posing growing risks to ecosystem services (ESs) through complex and spatially heterogeneous associations. Understanding these associations, and whether CES-level relationships exhibit nonlinear features, is essential for formulating effective landscape management and adaptation strategies under increasing climatic pressure. This study aims to (1) characterize the spatiotemporal dynamics of four types of compound climate extreme exposure, six key ESs, and comprehensive ecosystem services (CESs) from 2000 to 2020, and (2) compare the nonlinear associations and spatial explanatory effects of different compound climate extremes on CESs. Daily data were used to identify compound extreme days, while five-year aggregated indicators were used to characterize recent cumulative exposure corresponding to CESs observation years. We integrated long-term high-resolution datasets with random forest (RF) and the optimal parameter geographic detector (OPGD). RF was used to assess the relative national-scale contribution of compound climate extremes to CESs prediction and to identify CES-level nonlinear association patterns, whereas OPGD was used to quantify spatial explanatory power and interaction effects associated with CESs heterogeneity. The spatial patterns of the six ESs and CESs remained relatively stable during 2000–2020, whereas the total amount of CESs showed an initial increase followed by a decline. In contrast, compound climate extremes showed clear regional divergence. RF-based analyses suggested apparent nonlinear association patterns at the equal-weighted CES level. Among the four extreme events, WD showed the most pronounced nonlinear pattern in the PDPs, with a curve-based transition from a weak positive association at lower exposure levels to a stronger negative association at higher exposure levels, most apparent around 10 days of recent cumulative WD exposure under the aggregation framework used in this study. Across both the national-scale contribution and spatial analyses, dry-related compound climate extremes showed stronger associations with CESs than CW and WW. These findings indicate that recent cumulative exposure to compound climate extremes is associated with CES-level nonlinear and spatially heterogeneous patterns across contrasting regional climate–landscape contexts in China. This study provides an empirical basis for ecosystem management and region-specific adaptation under escalating climate risk.