A data-model fusion technique to evaluate the temperature sensitivity of SOC decomposition

Background/Question/Methods

A good understanding of temperature sensitivity (Q₁₀) of soil organic carbon (SOC) decomposition, especially the relative sensitivity of labile and non-labile C decomposition is crucial for predicting C dynamics under a projected warming climate. Despite much effort, uncertainty still exists in terms of the relative sensitivity of labile and recalcitrant C pools. Although different methods have been proposed to assess the relative temperature sensitivity of labile and recalcitrant C decomposition based on incubation data, there is no preferable method available so far. A new model was developed in this study. The main objectives were to calculate Q₁₀ using a data-model fusion technique, and to evaluate dynamics of Q₁₀s with C recalcitrance. This model allows estimating turnover rates for different C pools, the proportion of each C pool and Q₁₀s by using Bayesian probability inversion and a Markov Chain Monte Carlo (MCMC) technique to simultaneously assimilate information provided by incubation data under different temperatures. A 2-pool and a 3-pool based model were compared. A published dataset, which consists of 12 soils from 6 sites along a mean annual temperature gradient (2–25.6°C), with three incubation temperatures (15, 25 and 35 °C) and an incubation period of 588 days was assimilated.

Results/Conclusions

With the 2-pool model all parameters were well constrained, but Q₁₀ between 15-25°C for the more resistant pool could not be well constrained for most sites. When the 3-pool model was used, parameters and Q₁₀ relating to the nominal passive pool were not well constrained between 15-25°C. Accordingly, only the Q₁₀ values between 25-35°C were used for further analysis in this study. The mean residence time (MRT) for the labile and more resistant pool varied from 0.17-0.53 and 16.9-54.1 years, respectively, based on the 2-pool model. According to the 3-pool model, MRT of the nominal fast, slow and passive pools varied from 0.07 - 0.3, 1.42-5.78 and 27.98-319.01 years, respectively. The mean (±95%CI) Q₁₀ values for the more labile and more resistant pool were 2.1±0.28 and 2.9±0.41 respectively when estimated by a 2-pool model. Q₁₀ values for the nominal labile, slow and passive pools were 1.7±0.21, 2.9±0.74 and 3.8±0.47, respectively, when estimated by a 3-pool model. Q₁₀ was found to be a function of mean residence time of C pools by a natural logarithmic way. The results support that temperature sensitivity of C decomposition increases with C recalcitrance.