Abstract:Urban areas serve as critical hotspots for biodiversity conservation. Constructing and preserving urban ecological networks from a species-centric perspective is of significant importance for mitigating threats to species survival within cities, enhancing species diversity, and achieving urban sustainable development goals. This study employs birds as indicator species, integrating complex network theory with random forest modeling to establish a bird habitat network in Shenzhen and to analyze its spatial patterns in relation to built-environment thresholds. The network comprises 135 nodes and 7 427 weighted edges, exhibiting small-world properties and high connectivity density. Random forest analysis reveals that anthropogenic disturbances contribute 40.3%, with urban proximity and nighttime light intensity showing significant effects. Among habitat factors, a distance of 960.53 m from water bodies serves as a key threshold, while a proportion of 7.08% natural grassland yields the highest efficiency in supporting diversity. Based on the findings, a management framework of 'area differentiation – node prioritization' is proposed, along with an ecological network structured as 'one core and three corridors', offering a quantitative paradigm for advancing biodiversity conservation and sustainable development in high-density urban settings.
