Soil fertility context shapes biodiversity-carbon associations across forest strata in a subtropical karst landscape.

Forests in karst landscapes occur across strong edaphic gradients that may alter biodiversity-carbon relationships, yet these relationships are commonly evaluated using tree aboveground biomass carbon, potentially obscuring contrasting responses among forest strata and biomass compartments. We examined how soil fertility context modifies multidimensional biodiversity-carbon associations across aboveground and allometrically estimated belowground compartments in a subtropical karst forest landscape of China. Fifty-four plots were classified as low-fertility (LF) or high-fertility (HF) based on soil nutrient conditions. We quantified tree and shrub aboveground carbon and allometrically estimated root carbon, and assessed plot-level taxonomic diversity, functional diversity, community-weighted trait composition, structural diversity, and abiotic conditions. Associations were evaluated using correlation, regression, relative-importance, and piecewise structural equation analyses. Tree aboveground and estimated tree root carbon were higher under LF conditions, whereas both shrub carbon components were higher under HF conditions. Associations between plot-level biodiversity attributes and carbon stocks varied strongly among woody strata and fertility contexts. Under HF conditions, plot-level species diversity was positively associated with tree carbon but negatively associated with shrub carbon, while structural diversity was positively related to estimated tree root carbon but negatively related to shrub carbon. Under LF conditions, plot-level functional diversity showed negative fitted paths to tree carbon but positive paths to shrub carbon. Predictor rankings and pathway configurations differed across all four carbon compartments, and the measured predictors explained more variation under HF than LF conditions. These findings show that soil fertility is associated with contrasting carbon-storage patterns between tree and shrub strata and that aggregated metrics can conceal reversals in biodiversity-carbon relationships. Resolving forest strata and aboveground and allometrically estimated root-carbon components is essential for understanding plant diversity and carbon storage and informing vegetation restoration in karst landscapes.