Large-scale spatial variability and distribution of soil organic carbon across the entire Loess Plateau, China

Soil organic carbon (SOC) plays an important role in soil productivity and the global carbon cycle. However, little is known about the regional distribution of SOC across the entire Loess Plateau region of China. We investigated 382 sampling sites across the region (620 000km2) and collected 764 soil samples from the topsoil (020cm) and subsoil (2040cm). Standard statistics were used to identify the regional SOC content and the relationships with 11 selected environmental variables. Concentrations of SOC varied within a wide range throughout the region from 0.38 to 54.03gkg-1, with mean values of 10.34 and 6.78gkg-1 for the topsoil and subsoil, respectively. Coefficient of variation values showed moderate variation for SOC in both soil layers. Significant correlations were detected between SOC and these environmental variables, notably with soil total nitrogen (TN), soil pH, and clay content. Multiple linear regression analysis indicated that TN, clay content, soil pH, elevation, and temperature had greatest effects on regional SOC variability among all the selected soil and site variables. Geostatistical analysis showed that the maximum autocorrelation ranges were 384 and 393km for SOC in the topsoil and subsoil, respectively. Nugget-to-sill ratios were 0.52 and 0.50, which also indicated moderate spatial dependence. Maps of SOC distribution produced by the geostatistical method showed that the overall spatial pattern was characterised by an area of low SOC content surrounded by bands with higher values, which generally increased towards the region's boundaries. The distribution pattern corresponded to that of the major regional landforms, which also influenced land use, whereby the sandy Ordos Plateau is surrounded by relatively fertile plains and valleys, where the human population density is highest, and the regional boundary is mountainous. The spatial data of SOC could be useful as an important initial state in regional SOC modelling and possibly be used in calibration and prediction processes in the remote sensing method to estimate SOC content for large-scale areas. CSIRO 2012