Significance of soil temperature and moisture for soil respiration in a Chinese mountain area

We measured soil CO2 efflux rate at 11 sites with different vegetation types, elevations and soil textures in a mountain area near Taiyuan city, China, over a period of 1 year. The aim was to understand the seasonal and spatial changes of soil respiration (Rs) and its responses to soil temperature (Ts) and soil water content (Ws).During the experimental period the mean Ts of the sites at 10 cm depth ranged from about 0 to 26 °C, and the mean Ws of the surface 0–10 cm soil layer fluctuated between the levels of field water holding capacity (WHC) and less than 1/3 of WHC. The temporal course of Rs and Ts could be fitted with a three-parameter Gaussian equation, with the higher values in August and the lower values in March and December. The annual mean Rs (based on daily-weighted monthly mean Rs) was 3.08 ± 2.12 (mean ± S.D.), 3.85 ± 2.92, 3.62 ± 2.71, 2.47 ± 2.12, 3.45 ± 3.35, 3.56 ± 2.80, 3.65 ± 3.02, 4.27 ± 3.69, 4.63 ± 4.05, 3.79 ± 2.66 and 2.18 ± 1.47 μmol CO2 m−2 s−1 from site 1F (forest) to 11B (bare), respectively, and 3.51 ± 0.71 μmol CO2 m−2 s−1 across the 11 sites. Cumulative annual Rs (March to December) varied from 692.1 to 1472.0 g C m−2 yr−1, with an average of 1114.6 g C m−2 yr−1 across all 11 sites. The spatial variations (between-site and within-site) of Rs were significant, but there was no clear evidence for which factor mainly affected the spatial variation.Temporal variations of Rs were dominantly controlled by Ts during most days of the year. However, during early summer, when Ws was limiting, Rs decreased dramatically and Ws exerted control over Rs. At all sites, Ts was the primary factor driving temporal variations in Rs. The functional relationships of Rs to Ts could be described well by exponential and Lloyd and Taylor equations. The coefficients of determination R2 of Ts to Rs in 11 sites varied from 0.55 to 0.84 for the exponential equation, and from 0.51 to 0.86 for the Lloyd and Taylor equation when the drought-affected data were excluded. The relationships of Ws to Rs could be described well by linear and power equations. The R2 of Ws to Rs in 11 sites ranged from 0.27 to 0.73, which were smaller than those of Ts to Rs, when the Rs was normalized using the fit of the Q10 function with Ts at 10 °C. Both the Q10 and R10 increased when dry-affected data were removed from the data sets. The Q10 ranged from 2.37 to 5.53 in 11 sites, and the R10 of the exponential equations varied from 1.27 to 2.90 μmol CO2 m−2 s−1, slightly lower than those of Lloyd and Taylor equations ranging from 1.34 to 4.34 μmol CO2 m−2 s−1. The calculated Q10 and R10 of each site at the seasonal time scale were negatively correlated with Ts and positively correlated with Ws. For all data sets, four two-variable equations including linear and non-linear ones could be used to model relationships of Rs to both Ts and Ws together, with the R2 ranging from a minimum of 0.58 to a maximum of 0.86 for individual site. Our research results can bear important implications for the study of CO2 efflux in similar semiarid regions.