Abstract: With the change of global climate, forest soil organic carbon, as an important part of the carbon pool, has become one of the key research topics of forest carbon cycle. We studied soil organic carbon, carbon density and their influencing factors of eight natural Chinese pine forests with different ages in Songshan, Beijing using the variance analysis and variance decomposition method. We found that (1) the average soil carbon content of the pine forests in the studied area was 20.61 g·kg^-1. The soil carbon density was 153.67 t·hm^-2, which was lower than the average value of Chinese forest ecosystems (193.55 t·hm^-2). At the same aged stands, the soil organic carbon content decreased with the increase of soil depth significantly (P<0.05). (2) In the 0-50 cm soil layers, there existed significant differences in soil organic carbon content among different aged stands (P<0.05). However, in the 50-100 cm soil layer, the difference was not significant (P>0.05). As the stand age increased, soil carbon density increased significantly (P<0.05). Soil organic carbon in the shallow surface layers (0-30 cm) accounted for 81.1%, 83.6%, 82.5%, and 81.7% of total soil organic carbon in middleaged forest, nearmature forest, mature forest, and overmature forest, respectively. (3) The carbon content and carbon density of each layer were significantly positively correlated with soil water content (P<0.001, P<0.05), and the carbon content of each layer was significantly negatively correlated with soil bulk density (P<0.05). The average carbon content (ACC), and average carbon density (ACD) of various soil layers in each plot were significantly correlated with topographic factors, stand characteristic factors and soil factors. Terrain model, stand characteristic model, and soil model explained the degree of variance of ACD and ACC differently. Overall, the stand characteristic model can explain the ACD variance well, while the terrain factor model and the soil factor model can not. The stand characteristic model and the soil model together can explain the variance of ACC well, while the terrain factor model can not.

Key words: ecosystem services, carrier, ecosystem services flow, cost-effective.