Soil organic carbon in arid and humid grasslands of China: Spatial heterogeneity, driving factors, and future changes

Grassland ecosystems play a crucial role in the global carbon cycle, yet the spatial patterns and drivers of soil organic carbon density (SOCD) across aridity gradients remain insufficiently understood. Here, we used the Carbon Density Dataset of China's Terrestrial Ecosystems (2010s) and applied random forest models to simulate SOCD at two soil depths (SOCD_{0-20 cm} and SOCD_{0-100 cm}) in arid and humid grasslands. We further quantified the contributions of environmental drivers and projected future SOCD dynamics under Shared Socioeconomic Pathways (SSPs). Results showed that SOCD was lower in arid than in humid regions, but due to their vast extent, arid regions contained larger total soil carbon stocks. The 0-20 cm layer accounted for approximately 50% of the SOCD in the 0-100 cm profile. In arid regions, SOCD_{0-20 cm} was mainly controlled by mean annual temperature and soil moisture, while SOCD_{0-100 cm} depended on clay content and total nitrogen. In humid regions, SOCD_{0-20 cm} was strongly influenced by temperature and NDVI, whereas SOCD_{0-100 cm} was regulated by clay and nitrogen. Under future SSP scenarios, SOCD is projected to decline in arid regions but increase in humid regions, indicating an enhanced carbon sink potential. SOCD_{0-20 cm} showed higher variability and lower stability than SOCD_{0-100 cm}. These findings clarify the spatial heterogeneity and depth-dependent controls of grassland SOCD and provide a scientific basis for region-specific soil carbon management, contributing to China's "dual carbon" goals and sustainable grassland policies.