Spatial pattern and influencing factors of topsoil organic carbon contents in hilly and mountainous regions of Eastern Sichuan in China

The spatial distribution of soil organic carbon (SOC) and its influencing factors are very crucial to the soil properties and sustainable land use. Among them, the hilly area of eastern Sichuan is dominated in the food security and ecological protection. However, some agricultural practices (such as over-cultivation and excessive use of fertilizers) have led to the decline in the SOC content. There are the increasing risks on the soil health and the long-term sustainable agriculture. This study aims to explore the SOC distribution in the hilly areas of eastern Sichuan, according to the unique terrain and soil texture characteristics. 1 301 soil samples were collected from five fields in the various districts and counties, indicating the representative coverage of all secondary cultivated land categories. The soil samples were taken at a depth of 15–30 cm with the distribution of organic carbon in agricultural soils. A systematic analysis was implemented on the influencing factors on SOC content. Two categories were classified into: external environmental random and soil endowment factors. External environmental factors included the meteorological data, such as annual mean temperature and precipitation. While the soil endowment factors were encompassed the physicochemical inherent properties of the soil, including texture, parent material, and nutrient content characteristics. Topographical data (such as slope and elevation) was derived from DEM (Digital Elevation Model) calculations, in order to explore the impact of terrain on the SOC distribution. A combination of statistical methods, spatial analysis, and random forest models was employed to assess the correlation between these influencing factors and the SOC distribution. Results indicate that: 1) SOC content was ranged from 4.91 to 15.26 g/kg in the hilly areas of eastern Sichuan, with an average of 7.39 g/kg and a coefficient of variation of 21.70%. It infers that there was the moderate spatial variability across the region. 2) The Gaussian model revealed a nugget effect of 0.173, with a nugget ratio of 61.5%, indicating the moderate spatial variation at small scales. 3) The Kriging interpolation was used to predict a spatial pattern of SOC distribution. A "north high, south low" trend was found with the SOC levels gradually increasing from the southern to northern regions. This spatial variation was attributed to the external environmental random factors and soil endowment factors. 4) Climate and topography shared the significant correlations with the SOC content (P < 0.01). Among them, annual mean temperature displayed a strong negative correlation (−0.72) with the SOC content, while the correlations with precipitation (0.34) and slope (0.30) were relatively weak. 5) Soil texture and physicochemical properties were the most significant influencing factors on the SOC content, overshadowing the influence of topographical and climatic conditions. The findings were emphasized that the soil management practices (particularly on the soil texture and nutrient levels) shared the greatest impact on the SOC distribution. 6) The various models were tested. Among them, the random forest regression model performed the best to predict the SOC distribution. Therefore, the nutrient content was identified as the most critical influencing factor on SOC, accounting for 61.3% of the spatial variation. In contrast, topographical factors were contributed only 1.7%, indicating the relatively minor role of terrain in determining SOC levels. These findings suggest that the soil nutrient management can be expected to enhance the SOC content in the sustainable agricultural practices in the hilly areas of eastern Sichuan. This finding can provide the promising insights for the influencing factors on SOC distribution. The importance of targeted land management was highlighted to enhance the soil health for the sustainable agricultural productivity.