[Spatial Differentiation Characteristics and Driving Factors of Soil Organic Carbon in Various Landscape Complexes]

Investigating the spatial differentiation characteristics of soil organic carbon （SOC） in regional agricultural land and analyzing its driving factors are important for screening auxiliary variables for SOC prediction in agricultural land and the accurate prediction of soil carbon stock. This study considered SOC in different types of agricultural land landscape complexes in Beijing as the research object. The differences in SOC content and its stock in different landscape complexes were explored based on the long-term positional monitoring data on the quality of cultivated land in Beijing and the field sampling and testing data. Utilizing multi-source and open-source data as environmental variables that affected SOC spatial differentiation, we explored the quantitative and spatial relationships between SOC and climate, topography, soil parent material, land use, and biomass factors in different landscape complexes through GeoDetector and geographically weighted regression modeling. Additionally, we constructed a structural equation model to reveal the pathways that influence each driving factor on SOC in terms of direct and indirect effects. Ultimately, the major controlling factors of SOC were identified in the study area. The results showed that： ① The mean values of ω（SOC） for various types of landscape complexes in the study area ranged from 6.23 to 28.26 g·kg^-1, with a variation coefficient of 3.80% to 33.92%, showing spatial heterogeneity. ② Climate, topography, soil parent material, land type, and biomass factors contributed to SOC at highly significant levels （P < 0.01）, and all factors were synergistic on SOC after their interaction. All factors could explain the spatial variation of SOC from 0.691 to 0.704, with stable explanatory validity. ③ Temperature, topography, and land type showed a highly significant direct effect （P < 0.01） on SOC in the study area. Among them, temperature was negatively correlated with SOC content. In different land types, SOC content was higher in landscape complexes located in forested land and lower in cultivated land. Topographic factors had the most excellent direct effect on SOC （effect value = 0.698, P < 0.001）, with higher SOC content in the mountains and lower in the plains, and topographic factors could also exert an indirect effect on SOC through differences in temperature and land type （P < 0.01）. Soil parent material and normalized difference vegetation index correlated significantly with topographic factors but had non-significant （P ≥ 0.05） direct effects on SOC content. Overall, topographic factors are essential factors influencing SOC spatial heterogeneity in the study area. It can be used as a critical cofactor to provide a reference for accurately assessing soil carbon stock on agricultural land in the study area.