Multivariate geostatistical methods for analysing the contribution of urban lakes and neighbouring greenery to mitigating PM2.5 under stressor indicators

• Lakes and greenery contribute to PM 2.5 removal and mitigate the impact of stressors. • Stressor indicators impact on PM 2.5 removal by greenery varied at different scales. • A negative correlation for large lakes dominated stressor-PM spatial interaction. • Interactive stressors contributed more to PM 2.5 mitigation in the GAM model. Maintaining the ecological integrity and functionality of ecosystems is one of the major challenges faced in the sustainable management of natural capital. Urban areas play a key role in this setting, providing multiple ecosystem services for a rapidly growing urban population worldwide while under constant pressure from several interacting natural and anthropogenic stressors. This paper targets the critical knowledge gap concerning how different stressor indicators – traffic (TD), building density (BD), building height (BH), percentage of impervious surface (PLAND_I) and land surface temperature (LST) – and their interactions affect the removal of particulate matter (PM 2.5 ) by urban blue-green infrastructure at multiple spatial scales. With spatial ranges varying with the area of blue-green infrastructure, the results showed that within the small-scale spatial range (90–450 m) the stressors LST, TD, PLAND_I and their interactions had a significant impact on PM 2.5 ; in the mesoscale spatial range (240–600 m), the stressors LST, TD, BH, PLAND_I, BD and their interactions had a strong impact on PM 2.5 ; while in the large-scale spatial range (390–1200 m), stressors of TD, PLAND_I, LST, BH and their interactions had a significant impact on PM 2.5 . Additionally, several spatial structures of PM 2.5 -stressor interactions were found, especially with larger lakes (exceeding 67.6 ha), which were dominated by negative correlations, which is singularity attributed to their greater capacity for PM 2.5 accumulation and microclimatic regulation and may mitigate the influence of stress factors. Importantly, this study confirmed that interactive stressors contributed more to the GAM model. Thus, overlooking interactive stressors may lead to an overestimation of PM removal by urban blue-green infrastructure. Regarding the spatial interactions of stressors-PM at multiple scales, spatial range conditions can change the properties of the blue-green infrastructure that determine the effective PM accumulation and identify the crucial stressor indices. A framework was developed to address the stressors’ mode of action and the extent to which the combined stressors affect PM mitigation. It allows the scientific community and relevant stakeholders to evaluate which stressors and their interactions in relation to PM removal share a common spatial pattern, and to assess independently the spatial covariation between stressors and PM removal at different spatial scales. It also demonstrates the possibility of using these stressor indicators as potential predictors of the impacts of land-use intensity on PM mitigation.