Prediction of nitrogen mineralization in organically fertilized growing media for soil‐less production

Abstract

BackgroundOrganic fertilizers derived from recycled materials and by‐products are currently investigated as a way of freeing ourselves from synthetic chemical mineral fertilizers within the framework of the agroecological transition. These organic fertilizers have to undergo a mineralization process mainly carried out by microbes, so that the mineral elements can be consumed by the plants.AimsThe challenge consists in providing tools to predict available N coming from mineralization of organic fertilizers to better control the doses and the frequency of application.MethodsWe developed and compared two predictive models of N mineralization of organic fertilizers, a multivariate statistical model and a first‐order kinetic model. Temperature (4, 20, 28, and 40°C) and humidity (−3.2, −10, and −31.6 kPa) were modulated and confronted to the response of four different growing media (GM) types and two organic fertilizers during a 49‐day experiment. The input parameters tested for the statistical model were the amount of N in the fertilizer, the initial N content, temperature and humidity of the GM.ResultsBoth models satisfactorily predicted the mineral N content, even if they tended to overestimate it for low concentrations (mostly corresponding to low temperature, 4°C) and the first‐order kinetic model overestimated it for the highest mineral N content (1000–1300 mg N kg−1). The two models were used to predict mineral N content on an independent dataset acquired under in situ conditions. The errors of prediction (RMSE) ranged between 220 and 256 mg N kg−1 according to the multivariate and first‐order models, respectively.ConclusionsTwo models have demonstrated satisfactory their ability to estimate the mineral nitrogen content in GM they need to be validated in more GM‐fertilizer couples and in the presence of plant.