Vertical distribution of soil organic and inorganic carbon under different vegetation covers in two toposequences of the Liudaogou watershed on the Loess Plateau, China

Understanding the carbon (C) cycle of the terrestrial ecosystem and estimating the C sequestration potential in soils are critical to evaluate ecosystem function and security. The vertical distribution of soil organic carbon (SOC) and inorganic carbon (SIC) as well as their influencing factors under different types of vegetation in two toposequences of the Liudaogou watershed on the Loess Plateau of China were investigated. Results showed that the measured soil properties were differentiated into two groups: (1) one behaving relatively conservatively (pH, bulk density, silt content, and capillary and total porosity) with coefficient of variations (CVs) < 10% and (2) the other encompassing more labile components (SOC, SIC, noncapillary porosity, clay, and sand and soil water content) with CVs > 10%. Soil organic C under different vegetation covers in the two toposequences (0.62 to 5.12 g kg⁻¹) decreased rapidly with depth in the top 0 to 50 cm soil layer and then remained relatively stable over the 50 to 200 cm depth, which can be described by the exponential model. In contrast, SIC (4.11 to 18.69 g kg⁻¹) was much higher than SOC and showed distinct distribution patterns, which increased initially with depth and then decreased downwards except for the grassland in the west-facing slope (W-grassland) exhibiting an opposite trend. There were no significant differences of SOC either under different vegetation covers or in different slopes over the whole profile (p > 0.05); however, significant differences of SIC were observed for both different vegetation covers in the same slope and for the same vegetation cover in different slopes (e.g., W-grassland versus W-forestland, W-grassland versus NE-grassland) (p < 0.05). This study demonstrated that different soil properties could simultaneously affect SOC and SIC contents; thus attempts to model SOC and SIC based on the independent effects of individual soil property without consideration of their interactions would result in unsatisfactory prediction. We proposed models that quantitatively described SOC and SIC contents as a combined product of different soil properties for each vegetation cover in both northeast-facing and west-facing slopes using multiple regression analysis, which could improve the prediction of SOC and SIC contents for the study area from the simple linear regression models.