Evaluation of winter wheat varieties’ responses to nitrogen supply supported by 10-band multispectral aerial imaging: reproducibility over crop seasons and sites

• Three red-edge bands improved crop trait prediction across wheat varieties • Vegetation indices predicted yield and N traits with high accuracy (R² ≥ 0.89) • RF identified NDRE740 and MCARI705 as key predictors of stage-specific N traits • Study supports data-driven, variety-specific N fertilization strategies Nitrogen (N) fertilization trials require resources to investigate the effects of N on yields, N use efficiency, and soil fertility. Multispectral imaging with drones could deliver more affordable and accurate information to optimize N fertilizers. The current literature lacks evidence of the robustness of vegetation indices (VIs) across sites, seasons, and varieties, with studies relying on multispectral sensors that had five bands or less and one red-edge (RE) band centered around 717 or 730 nm. Additional RE bands may allow for the estimation of VIs relevant for N use with higher accuracy and reproducibility. This study examines (i) whether additional RE bands increase the accuracy and consistency of grain and straw yield and N in the biomass, and (ii) whether the relationships between these parameters and VIs change according to crop stage and winter wheat varieties. The study was conducted at two Swiss sites over three seasons, with N fertilizer treatments varying in timing and rate. We found strong relationships (R 2 > 0.7) between VIs and N content in biomass, and grain and straw yield. The NDRE with the RE band centered at 740 nm was the most accurate predictor of yield from measurements conducted at heading (R 2 = 0.92) and during grain filling (R 2 = 0.89), while the MCARI with the RE band centered at 705 nm, measured before heading, was a stronger predictor of total N at anthesis (R 2 = 0.9). Therefore, the additional RE bands improved the accuracy of VIs for optimizing the N fertilization of wheat.