Abstract
Vegetation regeneration enhances soil carbon sequestration in degraded karst ecosystems. However, how succession influences soil organic carbon (SOC) fractions and enzyme activities remains poorly understood in Southwest China's karst landscapes. We examined three successional stages (grassland, shrubland, and secondary forest) in Pingtang County, Guizhou Province. We quantified SOC fractions, enzyme activities, and their drivers using random forest modeling, partial correlation, and structural equation modeling.SOC content increased significantly from grassland to secondary forest, with forest soils containing 76.74% more SOC than grasslands (P < 0.05). Mineral-associated organic carbon dominated all successional stages (51.75-58.81% of total SOC), while microbial biomass carbon increased most during succession (63.84%). Particulate organic carbon remained stable across succession (P > 0.05). β-1,4-glucosidase and cellobiohydrolase activities increased with succession, while dehydrogenase and catalase activities decreased. Random forest analysis identified carbon fractions as the dominant predictor of SOC variance (38%). Partial correlation analysis confirmed significant relationships between SOC, carbon fractions, and enzyme activities. Structural equation modeling showed that carbon fractions had the strongest direct effect on SOC accumulation, while soil nutrients had the largest total effect through indirect pathways, regulating enzyme activities and carbon fraction dynamics. These findings elucidate the pathways through which soil nutrients regulate SOC accumulation during karst vegetation succession, advancing understanding of carbon dynamics in degraded karst ecosystems.