Optimizing nitrogen rates for winter wheat using in-season crop N status indicators

Abstract

Conventionally, split nitrogen (N) applications at tillering and stem elongation enhance winter wheat yield, protein content, and nitrogen use efficiency. Vegetation indices, such as the Normalized Difference Vegetation Index (NDVI), Normalized Difference Red Edge index (NDRE), and leaf chlorophyll content (LCC) can be used as crop N status indicators (CNSIs) to easily underline the N deficiency. The aim of this study, conducted across 4 growing seasons in North-West Italy, was to create a model for regulating wheat fertilization rates and improve crop yield. The model relies on CNSIs measurements collected during the initial stages of stem elongation, aiming to achieve predetermined yield targets. In each year, the experimental design was a factorial combination of four N rates (0, 33, 66, and 99 kg N ha⁻¹) at tillering and five at stem elongations (0, 33, 66, 99 and 132 kg N ha⁻¹). The Aubusson cultivar, characterized by intermediate yield potential and protein content, was used to calibrate and validate the model in a 3-year trial (2018–2020), while the model was also applied to cv LG Ayrton (high yield potential) and Izalco (high protein content) in the 2020–21 season. Yield and protein content trends in function of N rate were parabolic or sigmoidal respectively and both tillering and stem elongation rate contributed to increase the grain yield and protein content. Furthermore, the significant interaction between tillering and stem elongation fertilization on grain yield suggested the possibility of correcting the N deficiency after tillering fertilization with a further application. A calibration function for a variable rate application was established related to the CNSIs; all of them were good predictors but NDRE showed a higher overall correlation (R² = 0.479) with grain yield than NDVI (R²= 0.461) or the LCC values (R²= 0.236) considering all the 3 years of experiments. The model’s intercept was reduced according to the decrease in the grain yield goal. The model's validation was accomplished by comparing the outcomes predicted by the model yields with the measured. The yield’s Root Mean Square Error (RMSE) values were low for cv. Aubusson (0.85, on average) in all 3 years, while the RMSE was higher in 2021 for LG Ayrton (1.90) and Izalco (1.35), in a production situation with a higher yield potential. The results suggest that the topdressing N fertilization rate could be accurately determined from measured CNSI values for a site-specific N fertilization management, but they also highlight the requirement of a model adaptation for different genotypes and environments.