A microbially driven and depth-explicit soil organic carbon model constrained by carbon isotopes to reduce parameter equifinality

Abstract. Over the past years, microbially driven models have been developed to improve simulations of soil organic carbon (SOC) and have been put forward as an improvement to assess the fate of SOC stocks under environmental change. While these models include a better mechanistic representation of SOC cycling compared to cascading-reservoir-based approaches, the complexity of these models implies that data on SOC stocks are insufficient to constrain the additional model parameters. In this study, we constructed a novel depth-explicit SOC model (SOILcarb – Simulation of Organic carbon and its Isotopes by Linking carbon dynamics in the rhizosphere and bulk soil) that incorporates multiple processes influencing the δ13C and Δ14C values of SOC. This was used to assess if including data on the δ13C and Δ14C values of SOC during parameter optimisation reduces model equifinality, the phenomenon that multiple parameter combinations lead to a similar model output. To do so, we used SOILcarb to simulate depth profiles of total SOC and its δ13C and Δ14C values. The results show that when the model is calibrated based on only SOC stock data, the residence time of subsoil organic carbon (OC) is not simulated correctly, thus effectively making the model of limited use to predict SOC stocks driven by, for example, environmental changes. Including data on δ13C in the calibration process reduced model equifinality only marginally. In contrast, including data on Δ14C in the calibration process resulted in simulations of the residence time of subsoil OC being consistent with measurements while reducing equifinality only for model parameters related to the residence time of OC associated with soil minerals. Multiple model parameters could not be constrained even when data on both δ13C and Δ14C were included. Our results show that equifinality is an important phenomenon to consider when developing novel SOC models or when applying established ones. Reducing uncertainty caused by this phenomenon is necessary to increase confidence in predictions of the soil carbon–climate feedback in a world subject to environmental change.