Spatiotemporal Patterns and Socio-Ecological Drivers of Coastal Wetland Landscape Fragmentation in Yancheng, Jiangsu

The coastal wetlands of Yancheng, Jiangsu Province, serve as a crucial overwintering and stopover site for rare waterbirds such as red-crowned cranes. The changes in their landscape pattern are directly related to the protection of regional wetland ecological functions and biodiversity. Using land use data from 2010 to 2022 to extract landscape pattern indicators for each period, this study adopts the Covariance–Analytic Hierarchy Process (Cov-AHP) to construct a comprehensive Landscape Fragmentation Index (LFI) for coastal wetlands, considering aspects including patch density, boundary complexity, spatial connectivity, and landscape diversity. Combined with multi-source indicators of Ecology–Economy–Society (EES), the Generalized Additive Model (GAM) and Geographically Weighted Regression (GWR) are adopted to systematically analyze the nonlinear response relationships and spatially heterogeneous driving mechanisms of landscape fragmentation. GWR is employed to reveal the spatial heterogeneity of the influence of each driving factor on fragmentation by mapping local regression coefficients. The results show that: (1) During the study period, the overall landscape fragmentation of the coastal wetlands in Yancheng, Jiangsu Province, exhibited a slow increase trend, with a spatial gradient pattern of “higher in the north and lower in the south, and higher in coastal areas than in inland areas,” reflecting the combined effects of varying levels of economic development and human activity intensity across different administrative regions and along the coast-inland gradient. (2) Based on the deviance explained by the GAMs, social factors generally had higher explanatory power for LFI than ecological and economic factors. Specifically, human population density and the proportion of construction land showed a significant positive correlation with LFI, while NDVI and the proportion of farmland exhibited obvious nonlinear effects under different fragmentation levels. (3) The GWR results indicated that the regression coefficients of the main driving factors were highly spatially non-stationary, and regions with high coastal development intensity had the most significant promoting effect on landscape fragmentation. The local coefficient maps further reveal that GDP and NDVI exhibit the strongest spatial heterogeneity, with their effects shifting from positive to negative across different sub-regions. The study demonstrates that the integrated framework of Cov-AHP combined with GAM and GWR can effectively characterize the spatiotemporal dynamics and multi-dimensional driving mechanisms of coastal wetland landscape fragmentation, providing a reference for the protection of coastal wetlands in Yancheng, Jiangsu Province, and the conservation of waterbird habitats represented by red-crowned cranes.