Machine learning-based detection of mangrove dynamics in a subtropical bay: reasons and outcomes

Mangrove forests constitute one of the most carbon-dense ecosystems in tropical and subtropical intertidal zones, providing significant ecological and economic benefits worldwide. Nevertheless, these vital ecosystems have experienced substantial degradation in recent decades, primarily attributable to escalating anthropogenic pressures and rising sea levels. Here, this study employed remote sensing imagery (1987–2023) and machine learning to examine mangrove forest dynamics in Fangcheng Bay (FCB), a subtropical bay in China's Beibu Gulf. Our analysis demonstrated a remarkable 182.38 % expansion in FCB's mangrove coverage over the 36-year period (1987–2023), with total area increasing from 233.19 ha to 658.49 ha. The West Bay (WB) and East Bay (EB) exhibited respective increases of 52.98 % and 274.47 %. Meanwhile, landward mangroves declined while seaward expansion occurred at an average shoreline progression rate of 1.28 m/yr. Furtherly, our analysis indicates that neither sea level rise nor estuarine declining suspended sediment concentration significantly influenced mangrove expansion. Tidal current-driven sediment deposition created optimal growth conditions by continuously replenishing mangrove tidal flats. These findings elucidate the drivers and patterns of FCB's mangrove dynamics amid rapid urbanization, offering critical implications for global mangrove conservation in comparable bay systems.