Abstract
Abstract Background Climate warming is altering precipitation regimes and increasing the frequency of extreme rainfall events. Soil respiration (Rs) is sensitive to these changes and remains a major uncertainty factor in predicting forest carbon cycling. However, it is not yet clear whether the response of Rs to soil temperature (Ts) becomes more dependent on soil moisture content (SMC) as SMC declines within the observed range, and how this moisture dependence is reflected in the temporal variability and temperature sensitivity of Rs. In this study, we applied structural equation modeling (SEM) and empirical models to analyze two-year continuous Rs observation data in a temperate deciduous broadleaved forest to evaluate how SMC modifies the temperature response of Rs across soil moisture conditions, with particular emphasis on relatively lower SMC conditions. Results Ts showed a stronger overall association with Rs than SMC, whereas the temperature response of Rs varied across SMC conditions. Within the SEM framework, Ts and SMC were identified as the main variables directly associated with Rs, and the path coefficient of Ts was greater than that of SMC, suggesting that Ts had a stronger direct association with Rs variability. Ts explained the overall seasonal pattern of Rs, but the residuals of the Ts-only model were significantly related to SMC and tended to be more negative under relatively lower SMC conditions. The apparent temperature sensitivity of Rs increased from the 5–10% SMC range to the 20–25% SMC range and then declined in the 25–30% SMC range (Q 10 = 3.7 at 5–10% SMC, 4.9 at 20–25% SMC, and 2.3 at 25–30% SMC). Conclusions Together, these findings indicate that Ts explained a substantial proportion of the overall seasonal variation in Rs, whereas under relatively lower SMC conditions, the increase in Rs with increasing Ts was constrained, indicating greater moisture dependence of the temperature response of Rs under relatively lower soil moisture conditions. In addition, the variation in apparent temperature sensitivity across SMC conditions suggests that a single fixed temperature response may not adequately represent Rs under changing soil moisture conditions. Therefore, Rs models should explicitly account for SMC-dependent variation in the temperature response of Rs.