Calibrating soil nitrogen turnover in biogeochemical models with 15N tracing data

Process-based biogeochemical models dynamically simulate the coupled Carbon (C), nitrogen (N), and water cycles in plant-soil-atmosphere systems, making them useful for understanding and upscaling the effects of management and climate change on N flows in agricultural systems. These models use conceptual pools such as plant tissue, microbial biomass and organic matter with different decomposability, which exchange material according to the selected model structure. The evaluation of model design and the validation of internal turnover fluxes is challenging as datasets typically include only yields and measurements of trace gases leaving the system, with limited information on how specific C and N pools contribute to distinct flows. Regarding the N cycle in particular, microbial immobilization and mineralization have remained largely unconstrained in biogeochemical models. Isotopically enriched, 15 N-labeled, mineral fertilizer or slurry can be used as a tracer to quantify N processes and transformations, including plant uptake, soil retention, and losses of fertilizer N. Here, we combine data from a 15 N-labelling experiment with the isotope model SIMONE to better constrain N turnover in the biogeochemical model LandscapeDNDC. The results show that LandscapeDNDC overestimated the allocation of applied fertilizer N into the plants while underestimating the retention in the soil pool and contribution to N loss pathways. At the same time, the model underestimated uptake of N from mineralized soil organic matter (SOM). Assuming that the internal mineralization-immobilization-humification cycle was underestimated, the corresponding turnover rates were calibrated based on fertilizer 15 N recovery in plants and soil. This calibration not only improved the simulation of plant uptake of 15 N for all seasons – the original main objective - but also led to better agreement with mineralization and immobilization rates inferred from 15 N measurements in soil and plants. Our results demonstrate how the combination of isotope modeling with 15 N recovery data can be used to effectively constrain biogeochemical process parameters.