Köchy, M., and A. Freibauer (2011)
Soil contains three quarters to four fifths of the terrestrial organic carbon mass, almost three times the amount of carbon in the atmosphere. Due to its size, small changes in the soil carbon mass can have large effects on atmospheric C concentration and hence on global warming. Previous models have not included wetlands and did not account for the thawing of permafrost soils which both have huge C masses. In addition, previous models did not include effects of land use change. We constructed a three-pool, temporally implicit model based on categorized driver variables associated by probabilities (Bayesian Network). The model was successful in simulating existing C stocks (C mass · area-1 · depth-1). Differences could be attributed to inconsistencies among input data classification with regard to historic C stocks, C input (net primary productivity [NPP] after harvest), and wetland status. Simulation based on projected changes of climate and land use showed that globally, climate effects on NPP have the strongest impact on soil C. Land use, in contrast, has the greatest effect locally via NPP because the expected global extent of land use change involving a change in growth form is small. Direct climate effects on decomposition rates were greatest in the humid tropics because of greater absolute increases in decomposition fractions with higher temperatures. Regions outside the tropics are most important for the total global C mass because of the greater land area. Global C mass is expected to increase if NPP increases due to CO2 fertilization. C mass is expected to remain similar or decrease if NPP is limited by N availability. For future assessments of global C stocks we suggest including the effect of high water tables on decomposition fractions.