Portable NIR Spectroscopy-Chemometric Identification of Chemically Differentiated Yerba Mate ( Ilex paraguariensis ) Clones

Yerba mate plants are part of the ombrophilous mixed forest with araucaria trees. Genetically improved yerba mate clones are classified into two groups of interest to a selective consumer public, decaffeinated and high caffeine plants. Quality control strategies for this type of food are essential, and multivariate tools can help in this procedure. Partial least squares-discriminant analysis (PLS-DA) was used to identify chemically differentiated yerba mate plants together with reflectance measurement (900–1700 nm) near-infrared spectroscopy (NIRS) in direct analysis of plant material. Yerba mate plants were cultivated in the semi-hydroponic system under five plant shading levels (0%, 40%, 51%, 76%, and 82%). Robustness of the mathematical model was verified for plants with all these shading levels. The PLS-DA model showed a sensitivity of 96.52% for the training set and 93.33% for the test set. Specificity greater than 97.12% was found for both sets, with an efficiency rate of 96.82% for the training set and 95.31% for the test set. Wilcoxon signed classification, sign classification in pairs, and randomization t -tests showed an excellent model fit. Principal component analysis of the NIR spectra demonstrated that shading affected the chemical composition more in high-caffeine clones than in decaffeinated ones. This indicates that the caffeine synthesis in yerba mate plants represents an adaptative strategy to elevated light conditions. Graphical Abstract