Urban heat islands (UHI) have become a major constraint on ecological security and livability in lake-dense regions. However, their multi-source driving mechanisms and spatial heterogeneity require further investigation from a from ecological security gradients. To address this issue, this study constructed an ecological security pattern using a stress–structure–service (3S) process chain. Building on ecological security gradients, it further explored how interactions among ecological potential stress (EPS), land ecological structure (LES), and ecosystem services (ESS) influenced UHI formation. Using the Dongting Lake region as a case study, remote sensing data, machine learning, and a multiscale geographically weighted regression (MGWR) model were integrated to identify nonlinear UHI drivers and their multiscale spatial responses. The results showed clear spatial differentiation in ecological security patterns. Low-level zones were strongly affected by human activities, while high-level zones formed ecological barriers dominated by water bodies, wetlands, and forests. UHI showed a lake-cooled, urban-heated, and lake-ring cropland thermal transitional pattern, with heat intensity gradually weakening as ecological security levels increased. XGBoost-SHAP analysis showed that EPS intensified at low thresholds and dominated warming. LES exerted bidirectional regulation in medium and high gradients, while ESS formed an ecosystem-dependent nonlinear mitigation mechanism in high gradients. UHI drivers were multiscale and spatially heterogeneous, with warming factors exerting strong local effects and cooling factors maintaining broad-scale stability. Along the ecological security gradient, UHI drivers transitioned from human disturbance dominance to ecological regulation dominance, indicating a trend toward system-level self-regulation. This study offers scientific support for climate-adaptive planning and ecological security management in lake-dense regions.
Xiong et al. (Mon,) studied this question.