Rapid urbanization intensifies landscape fragmentation, making the optimization of ecological networks essential for maintaining landscape connectivity. Effective optimization needs a diagnostic understanding of a network’s internal cluster structure. However, clustering methods cannot simultaneously reconcile spatial adjacency with interaction strength, often yielding results that are either geographically fragmented or ecologically incomplete. The objective is to generate spatially coherent clusters while preserving critical functional connections, thereby enabling the identification of structural weaknesses for targeted conservation. The main contribution of this study is developing a spatially constrained method of ecological network clustering and optimization, which involves constructing a Maximum Spanning Tree (MaxST) under Voronoi-based adjacency rules and employing a Genetic Algorithm (GA) to optimally partition this structure. The framework begins by constructing a MaxST under strict spatial adjacency rules to form a contiguous network backbone. A GA then optimally partitions this backbone by maximizing modularity, incorporating interaction strengths from both adjacent and non-adjacent corridors. The method successfully identified ecologically meaningful and spatially contiguous clusters. It revealed that corridors between clusters, despite their greater length, constituted the network's most vulnerable links due to their lower interaction strength. The analysis pinpointed high-priority structural weaknesses for intervention, including key ecological pinch points, barrier areas, and critical bridging corridors. This study introduces a practical method that explicitly addresses the spatial-functional trade-off in ecological network analysis. It shifts the conservation focus from isolated landscape elements to diagnosing and repairing cluster-based structural weaknesses, offering a targeted pathway for enhancing connectivity in fragmented urban landscapes.
Nie et al. (Mon,) studied this question.