Biogeochemical models simulate soil nitrogen (N) turnover and are often used to assess N losses through denitrification. Though models simulate a complete N budget, only specific N pools/fluxes (i.e. N2O, NO3-, NH3, NOx) are usually published, because the full budget cannot be validated with measured data. Field studies rarely include full N balances, especially N2 fluxes, which are difficult to quantify. Limiting publication of modeling results based on available field data is a missed opportunity to improve the understanding of modeled processes. We suggest that the modeler community support publication of all simulated N pools and processes in future studies.

The analysis of natural abundance isotopes in biogenic N 2O molecules can give precious information such as the nature of their precursor. However, many uncertainties exist and further validations are necessary to confirm this method as a reliable tracer of biogeochemical cycles. In particular, current methodologies (such as the isotopocule map approach) can only estimate the combined contributions of several processes at once. In this study, we aimed to develop a new methodology capable of individually discriminating the main sources of N 2O production in soil by combining natural abundance isotopes with the use of a 15N tracer ( 15N Gas Flux method). To achieve this, we conducted parallel laboratory incubations of an agricultural soil, during which we optimized the denitrification conditions through increase of moisture and amendment of nitrate; where this nitrate was either labelled or unlabeled with 15N atoms. A new linear system combined with Monte Carlo simulation enabled the determination of N 2O source partitioning, where bacterial denitrification was identified as the dominant process (87.6%), compared to fungal denitrification (9.4%), nitrification (1.5%) and nitrifier denitrification (1.6%). This new system has been compared to a recently developed stable isotope modelling tool applying Bayesian statistics (FRAME). The results agreed generally well at the exception of lower bacterial denitrification (80%) and higher nitrifier-denitrification (9%) contributions found with the FRAME model. This new approach provides a perspective for a wider application, potentially enabling the source partitioning of nitrous oxide emissions in agroecosystems.