Purpose In this study, a process-oriented biogeochemistry model, denitrification-decomposition (DNDC), was employed and adapted to interpret and integrate the field observations that the tested ecosystem was a weak sink of atmospheric carbon dioxide (CO 2 ) in 2004 but a strong source in 2005 during the growing seasons. Then we applied the model to predict long-term impacts of climate change on carbon (C) dynamics in the semiarid grassland. Materials and methods To adapt DNDC for the targeted grassland, we modified the default values of several grass parameters such as maximum biomass production, biomass partitions, plant tissue C/N ratio, and accumulative thermal degree days based on local observations. Daily weather data for 2004 and 2005 in conjunction with soil properties and management practices for the location were utilized as inputs to simulate the grass growth and soil C dynamics. The modeled C fluxes were compared with the eddy tower data. Sensitivity tests were conducted with a baseline and twelve alternative climate scenarios of 100 years for the target grassland.
Results and discussionThe observed and modeled CO 2 fluxes data were well in agreement (P<0.0001), both showing that the grassland shifted from a sink to a source of atmospheric CO 2 from a wet year (2004) to a dry year (2005) over growing season. Simulations of 100 years found that, under the fenced conditions, (1) the tested ecosystem would gain C with the baseline climate conditions at a rate of 200 kg C/ha/year; (2) the warmer and drier climate scenario made the worst case having the lowest grass production with 72 kg C/ha/year lost from the soil carbon pool; and (3) the cooler and wetter climate scenario made the best case having the highest biomass production with 790 kg C/ha/year sequestered in the soil during the simulated 100 years. Conclusions DNDC model could be used for the prediction of C dynamics in this semiarid grassland ecosystem. Since the ecosystem production is precipitation-limited, a cooler or wetter future climate would substantially elevate the C sequestration capacity of the grassland. However, the C sequestration potential could significantly decrease and even become negative to turn the ecosystem to a source of atmospheric CO 2 if the climate turned to be warmer and/ or drier in the coming 100 years.