Cross-border mangrove dynamics and management in the Beibu Gulf: Long-term remote sensing observation using object-oriented deep learning

•Cross-border mangrove dynamics in the Beibu Gulf were studied.•Object-oriented deep learning classification methods are developed.•Mangroves in the Beibu Gulf coastal zone increased by 64.92% in China and 36.19% in Vietnam.•Aquaculture has significantly hindered mangrove expansion in the Vietnam part of the Beibu Gulf. As a unique land-sea transitional wetland, mangroves possess high ecosystem service value. Monitoring mangroves in cross-border areas is crucial for shaping bilateral protection policies and guiding effective management decisions. This study examines the spatio-temporal dynamics of mangroves in the Beibu Gulf, situated in the China-Vietnam cross-border region, utilizing long-term Landsat imagery and an object-oriented deep learning classification technique from 1991 to 2021. Combining land cover change, landscape indices, and spatial analysis, the spatio-temporal dynamics of mangroves were quantitatively evaluated. Results reveal that the object-oriented deep learning classification method significantly enhances mapping accuracy and integrality of the classification results. During 1991−2021, the mangrove area in the Beibu Gulf coastal zone (BGCZ) increased by 92.11 km2, with annual increases of 0.82 km2 in the China part and 2.25 km2 in the Vietnam part. Mangroves connectivity improved in both the China and Vietnam parts, though the degree of disturbance was more pronounced in the Vietnam part compared to the China part. Mangrove dynamics are influenced by numerous factors, with aquaculture being the primary cause of mangrove fragmentation in the BGCZ. Varied management mechanisms and coastal zone development models in China and Vietnam have exerted influence on the dynamic evolution of mangroves. This study offers a significant scientific foundation for future cross-border mangrove conservation and management.