QUALITY ATTRIBUTES ANALYSIS AND AUTHENTICATION OF ARGENTINEAN WINE VINEGARS BASED ON MULTIPLE SPECTRAL SOURCE FUSION STRATEGIES

WAGNER MARCELO; ORTIZ DANIELA; ZALDARRIAGA HEREDIA JORGELINA; MONTEMERLO ANTONELLA; AZCARATE SILVANA; CAMIÑA JOSE

Congreso; 20º Encontro Nacional de Química Analítica, 8º Congresso Ibero-Americano de Química Analítica; 2022

Wine vinegars are highly valued mainly regarding the implemented acetification process. From a technological point of view, there are two production methods: submerged cultivation method, which is used in the industry to quickly obtain low-quality vinegars, and traditional method, in which the acetic fermentation is carried out inside wooden barrels, which are filled with wine to ferment, giving rise to high-quality vinegars that are more appreciated for their greater organoleptic complexity1. Particularly in Argentina, most wine vinegars are made through industrial fermentation processes and few of them are produced in a traditional way but without added value2. Currently, the search for fast, accurate and low-cost determinations makes UV-Vis and near-infrared (NIR) spectroscopies one of the most popular alternatives to determine the concentrations of chemical compounds as well as for authenticity studies3.The aim of this work was to develop an analytical methodology based on UV-vis and NIR measurements in combination with multi-block analysis to quantify a multi-parametric analysis for the quality control of wine vinegars and, additionally, to differentiate them according to the implemented production method. Different determinations that are normally used for the quality control of these products were carried out applying standard methods. For vinegars, the following results were obtained: total acidity 3.70-6.82%, fixed acidity 0.11-1.08%, volatile acidity 3.63-6.70%, pH 2.49 - 3.07 and polyphenol concentration 0.1 1.98 g/L eq. Ac. Gallic.Spectra of each sample were obtained from direct measurement for NIR (944-1670nm) and by dilution for UV-Vis (178-890nm). With the information generated, predictive models were built for the quantification of each parameter through partial least squares (PLS) regression analysis with each data set obtained by each individual technique. Likewise, in order to exploit the particularities of the different data blocks, different data fusion strategies (low and mid-level) were performed and compared. The predictive capacity of each model was evaluated from the calibration regression coefficients (R2), the mean square errors of calibration (RMSEC), cross-validation (RMSECV) and prediction (RMSEP), as well as the percentage error of prediction (REP). The best results were achieved when the mid-level data fusion strategy with SO-PLS was applied. All the parameters analyzed showed REP values below 12%. From spectral data, a classification approach to differentiate and classify vinegars produced by traditional and industrial systems was conducted. Principal component analysis (PCA) and SO-PLS discriminant analysis (PLS-DA) were performed on the data set as exploratory and classification analysis, respectively. From PCA, a clear differentiation of the samples of wine vinegars produced industrially from those produced traditionally was observed, and the classification analysis using SO-PLS-DA yielded a correct classification rate of 96%. The synergistic combination of spectral information and chemometrics can be a powerful tool to perform a quality control of wine vinegars and to verify the type of acetification carried out for its production. However, future studies involving a larger number of samples are needed to validate the results and conclusions presented here, in order to allow the implementation of this methodology in the certification of these products.