Soil-derived dissolved organic matter in Inland wetlands along A temperate river: Insights from spectroscopic characteristics coupled with machine learning methods

Dissolved organic carbon (DOC) is a critical component of wetland carbon cycling, yet the sources and characteristics of soil-derived dissolved organic matter (DOM) that influence DOC dynamics in inland river wetlands remain poorly understood. This study investigated the spatial variations in DOM sources and composition, and their influence on DOC dynamics in 12 river wetlands across China's second-largest inland river basin, the Heihe River Basin. Using UV–Vis spectroscopy and fluorescence excitation-emission matrix (EEM) spectroscopy coupled with parallel factor analysis (PARAFAC), we found that wetland soil DOM predominantly consisted of protein-like components and had low aromaticity and hydrophobicity. Protein-like fluorescence was more pronounced in downstream wetlands than in midstream and upstream wetlands, and it was higher in subsoil compared to topsoil. Spectroscopic indices indicated low aromaticity and humification but high bioavailability, particularly in downstream wetlands. Machine learning analysis revealed that DOM spectral characteristics had a stronger direct influence on DOC dynamics than geographical and soil properties of the sites. The biological and humification indices of DOM were significantly negatively and positively correlated with DOC concentration, respectively. In addition, DOM characteristics were significantly correlated with elevation, climate, and soil properties. Our findings highlight the critical role of environmentally driven DOM characteristics in controlling DOC fate, providing valuable insights for predicting soil carbon biogeochemistry and informing carbon management strategies in inland wetlands.