Geographical origin discrimination of wheat kernel and white flour using near-infrared reflectance spectroscopy fingerprinting coupled with chemometrics

Wheat kernel and flour with three genotypes across 4 years procured from three different geographical areas of China were analysed using near-infrared reflectance (NIR) spectroscopy coupled with chemometrics to better classify wheat according to the origin, production year and genotypes, respectively. For this purpose, principle component analysis-linear discriminant analysis and multi-way anova were applied to the NIR data. The best classification percentages were obtained for flour matrix both for geographical origin and production years with the correct percentages of 100% and 73%, respectively. For genotypes, wheat whole kernel showed better classification percentage (98.2%). All the samples were validated using external validation procedure and the obtained percentages were found satisfactory with the average prediction abilities of >85% in all regions indicating the suitability of the developed model. Multivariate anova showed that NIR fingerprints of wheat kernels and flours were significantly influenced by regions, years, genotypes and their interactions. In conclusion, white flour showed better performance in discriminating the geographical origin as compared to wheat whole kernel.