Selective transport of soil organic and inorganic carbon in eroded sediment in response to raindrop sizes and inflow rates in rainstorms

The soil carbon (C) reservoir involves soil organic and inorganic carbon stocks. The influence of soil erosion processes on soil organic carbon (SOC) migration has been investigated in the literature, but the influence on soil inorganic carbon (SIC) remains uncertain. Variations in the proportion of SOC and SIC in eroded sediment can influence C cycling, so we conducted a detailed laboratory study to investigate the selective transport of SOC and SIC in eroded sediment in treatments with different raindrop sizes and inflow rates. We used the boosted regression trees (BRT) models to estimate the SOC and SIC contents according to the ratio of flow depth and raindrop diameter (h/D), the maximum shear stress (τ*max), the runoff rate, the stream power (ω), the sediment concentration (SC), the soil loss rate (SLR), and various particle-related factors, including D50, D[4,3], D[3,2], clay content, silt content and sand content of non-dispersed sediment. The BRT model explained 42.1% and 56.1% of the variation in the SOC and SIC contents in the sediment, respectively. The relative importance of particle-related factors was approximately 50%, demonstrating that the particle size distribution (PSD) of sediment was a crucial factor for both SOC and SIC transport. The SOC content was sensitive to the presence of fine particles, while the SIC content was largely determined by coarse particles because smaller particles with a higher specific surface area tended to adsorb more SOC whereas coarser particles contained more carbonates in the form of binding agents. In sediment under intensive rainfall on steep slopes, the SOC tended to be depleted, while the SIC was slightly enriched. Our study demonstrates that both SIC and SOC were selectively transported by water erosion and fills gaps in this field of research. Moreover, this work implies that the PSD of the sediment provides important information about the variation in SOC and SIC and should be considered when modelling the global C cycle.