Climate-dependent nonlinear relationships between PM_2.5 and ecosystem productivity across China's urban...

Air pollution can alter ecosystem productivity by modifying atmospheric radiation and environmental conditions, yet its large-scale ecological impacts remain uncertain. In particular, associations between fine particulate matter (PM_2.5) and vegetation productivity may vary across climatic environments and levels of urbanization, and are often interpreted as direct aerosol effects without adequate separation from co-varying pathways. Here we examine statistical associations between PM_2.5 concentrations and gross primary productivity (GPP) across five major urban agglomerations in China spanning pronounced gradients in climate and pollution intensity. Using satellite observations and interpretable machine-learning analysis, we identify dominant drivers of ecosystem productivity and quantify nonlinear PM_2.5-GPP relationships. We find a robust relationship between PM_2.5 and GPP across all urban agglomerations, but the direction and magnitude of these relationships vary across regions in a manner jointly shaped by climate and pollution regimes. Moderate aerosol loading is associated with enhanced productivity in arid environments, whereas elevated PM_2.5 concentrations are generally associated with suppressed vegetation productivity in humid and highly urbanized regions. These responses are consistent with a nonlinear balance between diffuse radiation fertilization and radiation attenuation, with atmospheric dryness further modulating ecosystem sensitivity; however, nitrogen deposition and other co-varying pathways cannot be excluded. Importantly, annual-scale PM_2.5 transition points along the SHAP response curves vary substantially across regions, with higher values in heavily polluted environments and lower values in cleaner regions; these breakpoints mark changes in the shape of the model-attributed response rather than universal physiological thresholds separating positive and negative effects. Overall, these results reveal that the ecological associations of air pollution emerge from the coupled influences of climate, urbanization and pollution intensity, and are best interpreted as climate-dependent statistical responses rather than uniquely identified causal aerosol effects.