qingwen zhang, Dehua Mao, Weidong Man, Fuping Li, Y. Zhang, Fenghua Wu, Caiyao Kou, Rui Yang, Jiannan He, Xuan Yin, Mingyue Liu
Coastal wetlands play a vital role in carbon sequestration and climate change mitigation. However, the invasion of Spartina alterniflora ( S.alterniflora ) poses a significant threat to these ecosystems. In this study, we collected 114 soil samples from S.alterniflora -invaded coastal wetlands and acquired monthly remote sensing images throughout the sampling year. Time-series variables covering the entire growth stages of S.alterniflora were extracted from these images. The iterative Boruta algorithm was employed to identify sensitive variables, and machine learning algorithms (Random Forest, Boosted Regression Trees, and eXtreme Gradient Boosting) were used to predict soil organic carbon (SOC) content. A space-for-time substitution approach was then applied to assess the impact of S.alterniflora invasion age on SOC dynamics. The results show that the correlation between SOC content and remote sensing variables varied significantly across months, with June-derived variables exhibiting the highest average correlation. Independent validation further indicated that all machine learning models achieved R 2 values above 0.6, with the random forest model performing best (R 2 = 0.663, nRMSE = 0.157, RPD = 1.713). NDWI was identified as the most important predictor based on variable importance and SHAP analysis, followed by the vertical–vertical (VV) polarization and shortwave infrared (SWIR) band reflectance. Furthermore, spatial evidence revealed that SOC content increased with invasion age, peaking at a saturation point after 19 years. A slight decline was observed after 22 years, due to the greater distance from the coastline, which may have limited the exchange of water, salt, and nutrients. These findings provide spatially explicit insights into the long-term effects of biological invasion on soil carbon dynamics and establish a scientific basis for the sustainable management of coastal wetlands under invasion pressure.