Carbon dioxide emissions from Ultisol under different land uses in mid–subtropical China

Soil organic carbon (SOC) pools and turnover time are highly sensitive to site variables and land-use change. In order to ascertain mass distributions of different SOC pools influenced by land use and evaluate their relationships with carbon mineralization and different soil properties, a 62-day laboratory incubation under a range of moisture (20 to 100% of water-holding capacity) and temperature conditions (5–45 °C) was conducted to measure CO2 evolution from four land-use types, and data from the incubation study were fitted to a three-pool first-order model that separated mineralizable soil organic carbon into active, slow and resistant carbon pools. The results indicate that: (1) The active carbon pool comprised 1.2–3.5% of SOC with a mean residence time of 49–65 days, (2) The slow carbon pool comprised 25.3–60% of SOC with a mean residence time of 2–27 yr, (3) The resistant pool accounted for 36.5–73.9% of SOC in four land-use types. The woodland had the highest resistant carbon pool and lowest active carbon pool, which indicates it to be more stable than other land-use types. Soil carbon derived respiration was well correlated with all the organic carbon pools as well as C:N ratio and soil derived respiration. SOC was positively correlated with slow and resistant carbon pool, and soil derived respiration. CO2 emissions increased (r2 = 0.273–0.544, P < 0.0001) with increasing temperature (up to 40 °C), with emissions reduced at the lowest and highest soil moisture contents. The Q10 values (the factor by which respiration rate increase for a 10 °C increase in temperature) (from 5 °C to 45 °C, at 60% of water-holding capacity) of 1.9 ± 0.2, 2.2 ± 0.3, 2.3 ± 0.3, and 3.2 ± 0.5 were observed for paddy, orchard, woodland and upland, respectively, and decreased with decreasing moisture content when soil water content was less than its optimum value, but an opposite trend was shown when soil retained water at contents higher than the optimum water content. Inclusion of both moisture and temperature in our multiple polynomial models resulted in much better predictions of CO2 (r2 = 0.70–0.78, P < 0.0001) emissions than predictions obtained using temperature (r2 = 0.27–0.54, P < 0.01) or moisture (r2 = 0.29–0.45, P < 0.01) alone. Our study indicates that vegetation type and/or management practices which control soil biological and biochemical properties, were important predictors of C fluxes in this short-term experiment.