In recent years, the application of remote sensing techniques is gaining a growing interest and importance in agriculture.
Researchers often combine data from near-infrared and red spectral bands according to their specific objectives. These types of combinations present the disadvantage of lack of sensitivity due to using a single or limited group of bands. In this work on-farm canopy spectral reflectance (CSR) data, composing of ten spectral bands (SBs) plus four spectral vegetation indices (SVIs), is considered in a joint manner to set up a methodology capable to identify genotype by environment interaction (GxE) in wheat. Spectral data are analysed over five wheat genotypes grown in five different environments.
Historically breeders have recognized the potentially negative implications of GxE in selection and cultivar deployment and have focused on developing tools and resources to quantify it.
We propose to perform a statistical batch processing, applying twoway analysis of variance to multiple spectral data, with genotype and environment as fixed factors. Results prove that this methodology performs well in both directions, capturing differences between genotypes within a single environment, and between environments for a single genotype, representing a step forward to converting spectral data into knowledge for the subject of GxE.