Bedrock and climate jointly control the phosphorus status of subtropical forests along two elevational gradients

Terrestrial carbon sequestration is constrained by the supply of soil phosphorus (P). Climate and bedrock are two key independent drivers of soil P supply. Their potential interactions with soil P supply remains poorly understood. To address this issue, we examined the P status of subtropical forests in southern China along two adjacent elevational gradients with contrasting bedrock types (granite vs. slate). The results show that the P concentration of granite was significantly lower than that of slate. Accordingly, the P concentration of mineral soil, litter, and fine-root P on the granite was significantly lower than that on the slate. In contrast, their ratios to the bedrock P concentration were higher in the granite transect than in the slate transect. Moreover, we found a stronger nutrient uplift effect on P in the P-poor transects. Although bedrock P concentrations were constant along elevation, topsoil total P, soil labile inorganic P (Pi), soil labile organic P (Po), and soil moderately labile Pi concentrations showed a significant increasing trend with elevation on both transects. Multivariate linear regression models revealed that bedrock P explained more variation of plant and soil P than did climate. Soil moderately labile Po concentration showed contrasting elevational patterns on different bedrock types, which indicated an interactive effect between bedrock P concentration and climate on the soil moderately labile Po concentration. The Pearson correlation analysis indicated that plant and soil P measures were more tightly coupled in the P-poor transect than in the P-rich transect. These results indicate that the P status of subtropical forests is determined predominantly by the bedrock P concentration and by its interaction with climate. Our results suggest that predictions of ecosystem P status and its responses to climate change can be improved significantly by incorporating local-scale parent material properties into the modeling frameworks.