Our previous research showed an increasing trend of rainfall in a temperate arid region of China. In this study, to address potential effects of additional rain on carbon emissions of an arid ecosystem, we conducted a field manipulative experiment with five simulated rain addition treatments (four rain addition treatments and one for comparison). Each treatment included four repetitive plots, each with an area of approximately 113 m², in Dun Huang, Gansu. The design simulated rain increases of 0% (CK), 50%, 100%, 200% and 300% of the long-term (1978-2008) average annual precipitation (40 mm) at the study site. We used the Li-8100 automated soil CO₂ flux system in a Nitraria sphaerocarpa community during the growing season, from May to September 2010. Our results showed a linear positive correlation in soil carbon emissions between 09:00-11:00 and 24:00 hours for vegetated and bare soil (bare soil R²=0.31-0.76, P<0.001; vegetated soil R²=0.85-0.96, P<0.001). The rain increases accelerated vegetated and bare soil respiration in the plots. Compared with the control group, soil carbon emissions of bare soil increased by 8.83, 11.03, 26.64, 31.69 g/m²,and the emissions of vegetated soil increased by 30.02, 45.34, 98.08, 133.90 g/m² with the 0% (CK), 50%, 100%, 200% and 300% treatments, respectively. Soil carbon emissions of vegetated soil were three times greater than those of bare soil during the growing season, which was statistically significant. There was a significant linear correlation between soil carbon emissions and precipitation during the growing season, in both bare soil and vegetated soil. With a 1 mm increase in rainfall, soil carbon emissions in vegetated and bare soils increased 1.12 g/m² and 0.27 g/m², respectively. With a given community cover, every 1 mm rainfall increase produced an additional 0.69 g/m² of soil carbon emission, relative to the control treatment.
We developed a multiple linear regression model for community coverage, precipitation and soil carbon emissions of a community of Nitraria sphaerocarpa, and found that soil carbon emissions of the control group were 2.97-6.55 kg from May to September. During the growing season, when community coverage was eight percent, soil carbon emissions of the community increased by 0.62, 2.18, 5.30 and 8.42 kg with the 0% (CK), 50%, 100%, 200% and 300% treatments, respectively. With community coverage at nineteen percent, the emissions respectively increased by 1.56, 3.12, 6.24 and 9.36 kg for those same treatments. For community coverage of thirty-eight percent, the respective emission increases were 2.50, 4.06, 7.18 and 10.30 kg. This study is important for estimating the ecosystem carbon budget in the arid region of Northwest China with future rainfall increases.