Exploring the impact of land use on bird diversity in high-density urban areas using explainable machine learning models.

Amid rapid urbanization, land use shifts in cities globally have profound effects on ecosystems and biodiversity. Birds, as a crucial component of urban biodiversity, are highly sensitive to environmental changes and often serve as indicator species for biodiversity. This study, using Shenzhen as a case study, integrates machine learning techniques with spatial statistical methods. Firstly, a multi-layer perceptron (MLP) model was employed to globally simulate bird richness based on citizen science data. Subsequently, a geographic weighted random forest (GW-RF) model was used to construct the complex relationship between bird diversity and land use. Additionally, SHAP analysis evaluates the effects of urban factors and development patterns on bird diversity. The findings reveal that anthropogenic disturbances and habitat factors significantly influence bird diversity. Furthermore, the impact of land landscape patterns on bird diversity exhibits notable spatial heterogeneity, with landscape patterns within ecological spaces and developed land showing marked differences in their effects on bird diversity. The study's findings clarify the intricate effects of urbanization on bird diversity, pinpointing specific ecological conservation areas. It underscores the importance of ecological conservation in guiding urban development, advocating for strategic restoration to bolster urban sustainability and optimize land use for the protection of ecological diversity.