A model-based methodology to derive optimum nitrogen rates for rainfed crops - a case study for corn using STICS in Canada

While the application of nitrogen (N) fertilizer increases crop productivity, the amount of N not taken up by the crop is released as reactive N causing adverse environmental effects, which are associated with short-term climate variations and consist of nitrous oxide emissions, ammonia volatilization, and nitrate leaching. Finding the optimum rate of N application accounting for both economic and environmental aspects is challenging because it varies according to soil texture and climatic conditions. In this paper, a model-based methodology is developed to identify the ecophysiological optimum N rates at which the applied N leads to minimum N excess with little loss in maximum achievable yield. A slope-based method is proposed to identify the optimum nitrogen use efficiency (NUEopt) and to determine the corresponding optimum N rate (Nopt) for each growing season in given agroclimatic regions according to soil properties. It uses the yield predicted by a process-based crop model, long time series of climate data, and a newly proposed Mitscherlich Baule-plateau (MB-P) function. The NUEopt is identified by evaluating the linearity of the relationship between yield and Nopt, and the reduction in yield compared to maximum achievable yield for a given soil and various growing seasons. We illustrate the methodology and its performance by presenting corn yield response to N rate applications predicted by the STICS crop model for dominant soils with contrasting properties with 48-61 years of daily climate data from five major regions located along an agroclimatic gradient of the Mixedwood Plains ecozone in Canada (42.3°N 83°W - 46.8°N 71°W), which is the main grain corn production ecozone of Canada. Our case study in this ecozone indicates that a newly proposed yield MB-P function outperforms the two commonly used functions, i.e., linear-plateau, and MB functions. The proposed methodology provides valuable information such as the likelihood of achieving a yield in a given region, the recommended N rate for given expected yield, the consequent N excess or deficit, and the reduction in yield compared to maximum achievable yield. The proposed methodology can be used in any region where a crop model has been adapted, and soil properties and time series of climate data are available. This approach sets the stage for an in-depth environmental assessment on reactive N within an integrated decision support system.