NIR SPECTROSCOPY AND CHEMOMETRICS FOR RAPID AND NON-DESTRUCTIVE DETECTION OF ADULTERATION IN BLACK MASHUA POWDER

Black mashua (Tropaeolum tuberosum Ruiz & Pavón) is an Andean tuber of interest for its bioactive and nutraceutical potential. Its growing demand and commercialization in powder form increase the risk of adulteration through substitution with lower-cost raw materials of similar composition, such as yellow mashua. In this context, near-infrared (NIR) spectroscopy, combined with chemometric tools, offers a rapid and non-destructive alternative for the authentication of powdered foods. The objective of this study was to evaluate the performance of NIR spectroscopy in detecting the adulteration of black mashua with yellow mashua, considering different spectral regions, pretreatments, and classification algorithms. Thirty samples of pure black mashua and 30 samples for each level of adulteration with yellow mashua were prepared, distributed across three ranges: low (2.5 and 5%), medium (7.5, 10, and 12.5%), and high (15 and 17.5%). Three spectra were acquired from each sample and averaged to obtain a representative spectrum. Spectra were acquired using two NIR instruments over the ranges of 900–1700 nm and 1350–2550 nm. Spectra were evaluated without pretreatment (raw) and pretreated with Savitzky–Golay smoothing (SG, 11:2) and Standard Normal Variate (SNV). Data were split into 75% for training and 25% for validation. Classification was performed using Linear Discriminant Analysis (LDA), Support Vector Machine (SVM), Ensemble Subspace Discriminant (ESD), and Wide Neural Network (WNN), with 20 repetitions per model. Performance was assessed using Accuracy, Sensitivity, Specificity, Precision, and F1-score. The 1350–2550 nm region showed the best overall performance. ESD showed the best average performance with untreated spectra, achieving 89.73% Accuracyv, 88.35% Sensitivityv, 96.55% Specificityv, 90.01% Precisionv, and 88.55% F1-scorev. The higher performance observed in the 1350–2550 nm region suggests that this region contains more discriminant information for differentiating black mashua from samples adulterated with yellow mashua. Furthermore, the best performance being obtained with raw data suggests that the pretreatments attenuated useful discriminant information. In conclusion, NIR spectroscopy in the 1350–2550 nm region, combined with ESD, showed satisfactory performance for detecting the adulteration of black mashua with yellow mashua. These results support the potential of this methodology as a rapid and non-destructive tool for authenticity control of black mashua powder.