Construction of ecological security pattern based on ecological protection red line, machine learning algorithms, and circuit theory: A case study of Kunming city, China

Constructing an effective ecological security pattern (ESP) is crucial for maintaining the connectivity and functionality of regional ecosystems. To address the uncertainty in ecological source (ECS) boundary identification and the subjectivity in constructing resistance surfaces in previous studies, this research introduces the use of the ecological protection red line (EPRL) and a minimum area threshold method to delineate ECSs. Simultaneously, machine learning algorithms optimize the resistance surface, and circuit theory identifies priority restoration areas for ecological corridors (ECRs). Finally, this study constructs the ESP of Kunming City. The results indicate that: (1) 31 ECSs were identified, covering an area of 3114.63 km², accounting for 71.68 % of the total area of the EPRL. These areas mainly consist of woodlands and waters, primarily located in the northern part of Kunming and around Dianchi Lake; (2) The ecological resistance surface was constructed using the XGBoost model, achieving an AUC value of 0.85 and outperforming other models. The spatial pattern of ecological resistance in Kunming shows lower resistance in the northern areas and higher resistance in the southern areas, with high resistance regions concentrated in areas of intensive human activity around Dianchi Lake; (3) A total of 77 ECRs were identified, with a total length of 1555.42 km. Furthermore, 59 priority restoration areas of ECRs were delineated, including 26 ecological pinch points and 33 ecological barriers, covering a total area of 275.44 km². These areas are mainly concentrated along the shores of Dianchi Lake, where urban expansion has severely impeded ecological connectivity. The study developed an integrated ESP incorporating ECSs, ECRs, and nodes and proposed targeted optimization strategies to effectively enhance the connectivity and integration of the regional ecological network.