Using spatially variable nitrogen application and crop responses to evaluate crop nitrogen use efficiency

Abstract Low nitrogen use efficiency (NUE) is ubiquitous in agricultural systems, with mounting global scale consequences for both atmospheric aspects of climate and downstream ecosystems. Since NUE-related soil characteristics such as water holding capacity organic matter are likely to vary at small scales (< 1 ha), understanding the influence on NUE subfield 32 ha) could increase fertilizer NUE. Here, we quantify four conventionally managed dryland winter-wheat fields Montana following multiple years sub-field variation experimental N applications. To inform farmer decisions that incorporates NUE, developed a generalizable model predict by comparing six candidate models, using ecological biogeochemical data gathered from open-source repositories normal farm operations, including yield protein monitoring data. While varied across years, was highest areas low availability estimated mineralization (SON). At levels applied N, distinct responses among suggest capacities supply non-fertilizer plant-available suggesting supplies more available locations higher total reducing any rate. Comparing modelling approaches, random forest regression provided predictions least error relative observed Subfield predictive models can help optimize agronomic maximizing economic net return which provides valuable approach optimization use.