The use of Chemometrics to Discriminate Sample Adulteration in Different Levels: the case of Peruvian Maca

Abstract

This study aimed to use chemometrics as a tool to allow for near-infrared (NIR) spectroscopy to detect adulteration in Peruvian maca powder. To adulterate the samples, pure maca powder was mixed with rice flour and rice bran, which are the most employed adulterants, using proportions of 25%, 50%, and 75% of each. After adulteration, the mixtures were submitted to NIR spectroscopy. The spectral data were used to build discriminant models through Partial Least Squares Discriminant Analysis (PLS-DA), and each model was evaluated through permutation tests to check for reliability. It was observed that all samples were discriminated against the pure ones, providing unitary sensitivities and specificities. All the models built were tested by the pairwise Wilcoxon signed-rank test, pairwise signed-rank test, and a randomization t-test, and the results indicated that the permuted and un-permuted models were not similar. The results indicated that NIR spectroscopy can be used to discriminate between pure maca powder samples and maca powder adulterated with different levels of rice flour and rice bran, achieving unitary sensitivity and specificity in the PLS-DA models.