Decade Dynamic Monitoring of Exotic Mangroves in Guangdong Province

Sonneratia apetala (S. apetala) and Laguncularia racemosa (L. racemosa) are typical exotic mangrove species in Guangdong Province, China. Their rapid spread brings potential invasive risks to the ecological balance and biodiversity of native mangrove ecosystems. Thus, accurately quantifying their distribution changes over the past ten years is key to regional ecological conservation and coastal zone management.

To tackle the classification problems caused by medium-low resolution remote sensing imagery and small-sample datasets, this study develops a hybrid spatio-temporal dual-channel (HSTD) method. This method integrates temporal and spatial feature information, which allows for accurate classification and dynamic monitoring of these two exotic mangrove species in Guangdong Province.

The experimental results show that the HSTD method significantly improves the classification performance for exotic mangroves, with an Intersection over Union (IoU) of 0.739. Its overall accuracy (OA) is 3.8% higher than that of standalone deep learning models and 9.7% higher than traditional machine learning models. Notably, compared with similar products, the proposed model can identify L. racemosa and scattered S. apetala patches more comprehensively.

In 2025, the total area of S. apetala in Guangdong Province reached 3509.13 ha, while that of L. racemosa was 81.01 ha. The two species showed an asymmetric overlapping distribution pattern. From 2016 to 2025, both exotic mangrove species presented an overall expanding trend: S. apetala had a cumulative area growth of 3.5%, while L. racemosa achieved an annual average growth rate of 5.4%—2.6 times that of S. apetala.

This study clarifies the spatio-temporal evolution patterns of S. apetala and L. racemosa in Guangdong Province, and provides important technical support and decision-making basis for the differentiated management and control of local exotic mangrove species.