Near infrared spectroscopy coupled to chemometrics for the authentication of donkey milk

Donkey milk (DM) is an emerging foodstuff, gaining increasing attention due to its nutritional properties. As a high value-added product, it can be subjected to adulteration with cheaper milks, e.g., cow milk. The present work addresses the possibility of developing a fast and economic method for the authentication of pure DM using Near Infrared (NIR) Spectroscopy. For this purpose, 147 samples (67 pure DM and 80 mixtures) were analyzed fresh; additionally, all samples underwent successive freezing-thawing cycles. Partial Least Squares-Discriminant Analysis (PLS-DA) was used to differentiate pure and adulterated samples. PLS-DA model built on the fresh milk correctly classified all training and test samples (100% accuracy). When model was applied to the freeze-thawed individuals, it showed high accuracy (79.7% after 3 cycles), suggesting that the spectroscopic signature of adulteration prevails on that of freezing-thawing, though the impact of the latter was proved to be significant by ANOVA-Simultaneous Component Analysis (ASCA).