A novel method for evaluating flavanols in grape seeds by near infrared hyperspectral imaging

Abstract

Chemical composition of seeds changes during grape ripening and this affects the sensory properties of wine. In order to control the features of wines, the condition of seeds is becoming an important factor for deciding the moment of harvesting by winemakers. Sensory analysis is not easy to carry out and chemical analysis needs lengthy procedures, reagents, and it is destructive and time-consuming. In the present work, near infrared hyperspectral imaging has been used to determine flavanols in seeds of red (cv. Tempranillo) and white (cv. Zalema) grapes (Vitis vinifera L.). As reference measurements, the flavanol content was estimated using the p-dimethylaminocinnamaldehyde (DMACA) method. Not only total flavanol content was evaluated but also the quantity of flavanols that would be extracted into the wine during winemaking. A like-wine model solution was used for this purpose. Calibrations were performed by partial least squares regression and they provide coefficients of determination R2=0.73 for total flavanol content and R2=0.85 for predicting flavanols extracted with model solution. Values up to R2=0.88 were reached when cultivars were considered individually.