The amount of CO2 respired from the soil annually was approximately 945, 785 and 920 g C m-2 for 2001, 2002 and 2003, respectively. Although Ts exerted a significant control on the spring recovery of Rs, most of the annual variability in total Rs was attributed to variations in volumetric soil water content (θ). At the seasonal time scale, Rs increased exponentially with Ts. In 2001, the soil CO2 efflux at 10oC (R10) and temperature sensitivity of Rs (Q10) estimated from Ts at 2 cm were 3.9 μmol m-2 s-1 and 3.4, respectively. A residual analysis revealed that a strong reduction in Rs occurred when θ in the root zone dropped below 0.10 m3 m-3. Separating the growing season Rs data set between high and low θ (based on the above threshold value) showed that both the R10 (4.4 and 2.8, respectively) and Q10 (3.1 and 2.5, respectively) were reduced during periods of low θ. To avoid the confounding effects of long-term changes in θ, Ts and phenology, values of R10 and Q10 were calculated daily during the growing season of 2001 using nighttime datasets of Rs and Ts at 2 cm. These daily Q10 values slowly decreased from 3.1 in May to 1.9 in September. In contrast, the daily R10 values closely followed the seasonal patterns of Ts and θ. This suggests that the biomass of biologically active soil, rather than its temperature sensitivity, responded to changes in environmental forcing. The results also showed that total CO2 production occurred mostly in the shallow soil layers (above 5 cm) during spring. The gradual warming deep in the soil and the low θ near the soil surface late in the growing season increased the relative contribution of deep soil layers to total CO2 production.