Quantifying and simulating carbon and nitrogen mineralization from diverse exogenous organic matters

The potential contributions of exogenous organic matters (EOMs) to soil organic C and mineral N supply depend on their C and N mineralization, which can be assessed in laboratory incubations. Such incubations are essential to calibrate decomposition models, because not all EOMs can be tested in the field. However, EOM incubations are resource-intensive. Therefore, easily measurable EOM characteristics that can be useful to predict EOM behaviour are needed. We quantified C and N mineralization during the incubation of 663 EOMs from five groups (animal manures, composts, sewage sludges, digestates and others). This represents one of the largest and diversified set of EOM incubations. The C and N mineralization varied widely between and within EOM subgroups. We simulated C and N mineralization with a simple generic decomposition model. Three calibration methods were compared. Individual EOM calibration of the model yielded good model performances, while the use of a unique parameter set per EOM subgroup decreased the model performance, and the use of two EOM characteristics to estimate model parameters gave an intermediate model performance (average RMSE-C values of 32, 99 and 65 mg C g(-1) added C and average RMSE-N values of 50, 126 and 110 mg N g(-1) added N, respectively). Because of the EOM variability, individual EOM calibration based on incubation remains the recommended method for predicting most accurately the C and N mineralization of EOMs. However, the two alternative calibration methods are sufficient for the simulation of EOMs without incubation data to obtain reasonable model performances.

Automated summary

This study quantified the C and N mineralization of 663 EOMs from five groups and found significant variability in mineralization within and between EOM subgroups. Individual EOM calibration of the model produced the best performance, while a unique parameter set per EOM subgroup decreased model performance and using two EOM characteristics for parameter estimation yielded intermediate results.