Bulk Functionality Diversification by Unsupervised Single‐Kernel Near‐Infrared (SKNIR) Sorting of Wheat

Abstract

ABSTRACTWe explored the effects of fractioning heterogeneous bulk wheat by fast unsupervised single‐kernel near‐infrared (SKNIR) sorting according to an internal complex NIR functionality trait using a fast prototype kernel sorter designed for postharvest bulk sorting. Sorting into three functionality fractions was performed on low quality lots from an organic field experiment from two growth years and two locations. Sorted lots were mixtures originally diversified by three different preceding catch crops. The resulting 12 fractions, as well as the 12 original wheat lots were characterized by 20 standard quality variables of grains and flours. The data was analyzed by principal component analysis (PCA) and analysis of variance (ANOVA). Within each year and location/cultivar, the SKNIR fractionation had significant positive effect on bulk grain density, protein, wet gluten content, Zeleny sedimentation volume, farinograph water absorption, farinograph softening, falling number, gelatinization temperature, and hardness index. Using the NIR fingerprint directly for sorting without calibration to a univariate reference showed that the resulting fractions were based on the major variance in the entire physicochemical quality trait within each lot as expressed by NIR. This novel unsupervised approach may become a powerful tool for sorting according to complex functionality traits, thus increasing overall quality, applicability, and value of the sorted crop.