Responses of soil respiration to rainfall pulses in a natural grassland community on the semi-arid Loess Plateau of China

Pulsed rainfall affects both aboveground vegetation dynamics and belowground biogeochemical processes, such as carbon cycling, in semi-arid regions. In order to study carbon released by soil respiration (SR) after rainfall pulses in natural grassland on the Loess Plateau, a rainfall simulation experiment was conducted in a grassland community co-dominated by a C-4 herbaceous grass [Botluiochloa ischaemum (L) Keng] and a C-3 leguminous subshrub [Lespedem davurica (Laxm.) Schindl] in the loess hilly-gully region. Soil respiration rate (R-s), soil temperature (T-s), and soil volumetric water content (S-v) were measured 1 day before and 1, 2, 3, 5, and 7 days after four rainfall treatments (ambient rainfall plus a 5 mm, 10 mm, 20 mm, and 30 mm rainfall pulse) and one control treatment (only ambient rainfall) in June and August 2013. Results showed that R-s and S-v largely increased one day after simulated rainfall > 5 mm. In June, the peak R-s under 10, 20, and 30 mm rainfall was 0.80-1.03 mu mol C m(-2) s(-1) in B. ischaemum, with a 25-62% increase compared with the control treatment, and 0.74-1.0 mu mol C m(-2) s(-1) (+51-104%) in L. davurica. In August, the peak R-s was 1.23-1.73 mu mol C m(-2) s(-1) (+23-73%) and 1.52-1.70 mu mol C m(-2) s(-1) (+81-102%) in B. ischaemum and L davurica, respectively. The magnitude and duration of the increase in SR were positively related to the rainfall size, and a more considerable increase was observed in August. There was a threshold rainfall (i.e., 5-10 mm) for triggering SR increases in both months. And different responses were found between the two species, there was more substantial SR increases in L davurica in comparison to B. ischaemum. After rainfall pulses, soil moisture and soil temperature co-regulated SR. During the relatively dry season (i.e., June), SR was negatively correlated with soil temperature and the temperature sensitivity Q(10) value of SR was small (0.5-0.6), while it changed to positively in August and the Q(10) was largely increased (3.2-4.3). Conversely, soil moisture was positively related to SR in both months and explained a large portion of the variation in SR (32-43% and 42-52% in B. ischaemum and L. davurica, respectively). These findings indicated that soil moisture was the major environmental factor in controlling SR in this grassland. Overall, our study suggests that SR response following rainfall pulses is species-specific within the grassland community and tends to be controlled by soil moisture, and these should be considered in the regional carbon budget assessment in the background of vegetation rehabilitation and rainfall pattern changes.