Near-infrared reflectance spectroscopy for the rapid discrimination of kernels and flours of different wheat species

Since products made of ancient wheat species are enjoying increasing demand by consumers, reliable and rapid methods for the automatable differentiation of flours and kernels of costly ancient species from less expensive bread wheat are required. In the present study, we demonstrate near-infrared (NIR) reflectance spectroscopy to represent a rapid and powerful method for product authentication. Multi-class partial least square discrimination analyses (PLSDA) were based on NIR spectra of kernels and flours from bread wheat (n = 705), spelt (n = 673), durum (n = 75), emmer (n = 75), and einkorn (n = 73), showing accuracy values of 80-100%. Two-class classification analyses allowed the clear-cut differentiation of species of the same degree of ploidy (durum vs. emmer and bread wheat vs. spelt) by the PLSDA model and validation without misclassifications. Most importantly, the detection of adulterations of spelt flours with bread wheat flours was feasible. Two spectral ranges (1370-1450 nm and 1850-1930 nm) were identified to exert the highest discriminative power between bread wheat and spelt. Since NIR spectrometers are routinely being used in the cereal industry for the determination of, e.g., protein and water content of wheat kernels and flour, the implementation of our approach may instantly allow for the authenticity control of wheat kernels and flours therefrom.