Estimated carbon residence times in three forest ecosystems of eastern China: Applications of probabilistic inversion

Carbon residence time is one critical parameter for predicting future land carbon sink dynamics but has not been well quantified for many plant and soil pools. This study applied a probabilistic inverse analysis of multiple observations to estimate mean residence times of carbon among three forest ecosystems in eastern China. Three assimilation experiments were conducted with either net ecosystem exchange data from eddy-flux measurements or six biometric and soil data (i.e., foliage biomass, fine root biomass, woody biomass, litterfall, soil organic carbon, and soil respiration) or all data to evaluate their relative effectiveness on estimation of carbon residence times of different pools in a terrestrial ecosystem model. Estimated mean residence times of carbon ranged from 2 to 10 months for metabolic litter and microbial biomass pools, from 1 to 3 years for foliage, fine root biomass, and structural litter pools, and from 17 to 1361 years for woody biomass, slow and passive soil organic matter pools at three forest sites. The residence times of carbon were longer for leaf, litter and microbes pools but shorter for fine root and wood pools in the young evergreen coniferous plantation at Qianyanzhou site than the two mature mixed forests at Changbaishan and Dinghushan sites. Carbon residence times were well constrained for three plant pools and slow soil organic matter by biometric and soil data, whereas residence times for metabolic and structural litter, and microbial biomass pools were constrained by daily net ecosystem exchange data. Overall, our study demonstrated that biometric, soil and net ecosystem exchange data are complementary in constraining mean residence times of an ecosystem model.