Green spaces in subtropical cities are important for providing ecological services that support human well-being and serve as reservoirs for diverse microbial communities, which in turn support ecosystem functions. However, studies on the characteristics of the phyllosphere microbial community and public health risks associated with putative pathogens in various urban green spaces exposed to anthropogenic stress remain limited. To address this gap, we collected leaf samples from forests, greenbelts, parklands, and wetlands across Wuhan, China, and analyzed the bacterial and fungal communities via next-generation sequencing (NGS) techniques. For bacterial and fungal communities, alpha diversity was significantly greater in low-traffic zones than in high-traffic zones. Beta diversity analysis revealed distinct clustering of bacterial and fungal communities according to the urban green space type. Anthropogenic factors (foot traffic) influence green space type to shape microbial community structure, function, and stability, with shifts significantly associated with soil physicochemical properties via Mantel tests and redundancy analysis. The relative abundance of Enterobacter and Enterococcus was significantly greater in high-intensity parklands (HIPS) and high-intensity greenbelts (HIGS) (41.84, 38.32%), respectively. Our findings provide important information for the sustainable management of urban green spaces by regulating microbial communities, offering new insights into ecosystem health and human well-being.

Wildfire susceptibility mapping supports proactive forest management, and estimated predictive performance may vary with spatial dependence and the control-point sampling strategy. We developed an interpretable random-forest framework to map wildfire occurrence probability across Hubei Province, China, by integrating multi-source environmental (meteorological, topographic, and vegetation) and socio-economic predictors. To enhance methodological robustness and address high-dimensional data complexity, the Boruta algorithm was employed for rigorous feature selection, identifying the most significant drivers while filtering out random noise. The model showed strong discrimination on held-out data (AUC = 0.942, accuracy = 87.9%), and variable importance highlighted sunshine duration, elevation, relative humidity, and maximum temperature as dominant predictors. Predicted wildfire probability exhibited a clear east–west gradient; high and very high susceptibility classes covered 22% of forested land while containing 82% of historical fires, indicating priority zones for targeted prevention and resource allocation. These results demonstrate that combining multi-source predictors with machine-learning interpretability can produce actionable susceptibility maps for regional fire-risk management.