Background Soil carbon pool is the largest carbon pool in terrestrial ecosystem, and its dynamic change and the distribution of reserve play an important role in the quality of soil. Therefore, the accurate estimation of soil organic carbon (SOC) storage, analysis of the spatial variability and uncovering the main factors that control the spatial distribution of SOC may provide basic data and theoretical basis for regional ecological construction and scientific and rational use of land resources.

Methods This research was conducted in Aohan banner of Chifeng city, Inner Mongolia, based on soil survey data. The scale effects of spatial variability in surface SOC (0-20 cm soil depth) was studied by multi-scale sampling of small watershed, medium region and county scale. The classical statistics and geostatistics with software GS+7.0 ArcGIS10.0 were used to analyze the spatial variability of surface SOC content.

Results 1) The mean contents of SOC at the county, medium region and watershed scales were 7.49, 7.57 and 7.54 g/kg, respectively. The change range of organic carbon content in the 3 scales was found to be between 1.43-19.48, 1.31-19.89, and 1.23-19.95 g/kg, respectively. There was no significant difference in the surface soil organic carbon fractions between the 3 scales, all of which were at a relatively low level. SOC contents showed a moderate spatial correlation with the range between 980 m and 2 100 m at the 3 scales.2) The Gauss model had better fitting effect on the soil organic carbon fraction of the small watershed scale, and the spherical model had better effect on the soil organic carbon content in the county and middle scale. The model fitting residuals was relatively small (1.96-2.43), and the determination coefficient was between 0.50-0.75, which showed that the selected model described the spatial structure characteristics of soil organic carbon content on each scale more accurately. The C₀/(C₀+C) reflected the variation characteristics of soil organic carbon data, and C₀/(C₀+C) on the 3 scales was between 27.66%-36.61%. It shows that the soil organic carbon in each scale was in the medium intensity spatial correlation. With the study scales enlarging, the spatial autocorrelation scale of SOC increased gradually. There was a significant difference of the relationship between influential factors and SOC content at different scales.3) Soil and vegetation presented the greatest influence on the distribution of SOC content at county scale, and vegetation and elevation showed the greatest influence on the distribution of SOC content at medium region scale, and elevation and slope had the greatest influence on the distribution of SOC content at small watershed scale.

Conclusions The results may help us to gain a better knowledge of the spatial distribution pattern of SOC and the dominant factors that impact this pattern at different scales.