Abstract: The carbon sequestration mechanism of the “wheat stubble-green manure” mixed return system in dryland areas of the Loess Plateau remains unclear. We selected local typical Triticum aestivum straw and Glycine ussuriensis straw to simulate the cereal/legume straw co-return model, and established four straw addition treatments with low (6.3 g·kg^-1) and high (11.0 g·kg^-1) organic carbon content in dark loessial soil: no straw addition control (CK), T. aestivum straw single application (S), G. ussuriensis straw single application (G), and equal mixture of both (SG). Using laboratory incubation experiments combined with ¹³C natural abundance tracing technology, we distinguished the original SOC and newly formed SOC, and measured soil CO₂ release dynamics, active carbon-nitrogen fractions, carbon acquisition enzyme activity, and net sequestration. The results showed that: 1) In low organic carbon soil, the cumulative CO₂ release for CK, S, G, and SG treatments was 371.3, 3228.9, 3403.1, and 6126.1 mg C·kg^-1, respectively. In high organic carbon soil, the values were 640.4, 3744.8, 3848.6, and 7000.6 mg C·kg^-1, respectively. 2) In low organic carbon soil, the net carbon sequestration for CK, S, G, and SG treatments was -0.40, 1.16, 0.20, and 0.89 g·kg^-1, respectively. In high organic carbon soil, the values were -0.98, 0.48, -0.05, and 1.02 g·kg^-1, respectively. 3) Carbon acquisition enzyme activity and active carbon-nitrogen fractions were generally higher in high organic carbon soil than in low organic carbon soil, with SG treatment showing the highest enzyme activity and active carbon-nitrogen fractions. 4) Path analysis indicated that newly formed SOC was the key positive factor determining net carbon sequestration, and significantly positively correlated with the input of straw-derived cellulose, hemicellulose, and lignin carbon. Initial soil carbon-nitrogen pro-perties had a direct negative effect on the formation of new SOC. Plant residue mass primarily promoted new carbon formation by enhancing microbial activity and carbon acquisition enzyme activity. In summary, low organic carbon soil had high carbon sequestration potential, with T. aestivum straw single application being the optimal treatment. High organic carbon soil was suitable for cereal/legume mixed application strategies. This study would provide a theoretical basis for differentiated straw return management in dryland areas.

Key words: black loessial soil, initial soil organic carbon, cereal/legume straw co-return, soil organic carbon sequestration, ¹³C natural abundance tracing