Prediction of wine color attributes from the phenolic profiles of red grapes (Vitis vinifera).J Agric Food Chem. 2008 Feb 13; 56(3):1105-15.JA
Knowledge about the relation between grape and wine phenolics is of key interest for the wine industry with respect to being able to predict wine quality from analyses of grapes. Prediction of the phenolic composition and color of experimentally produced red wines from the detailed phenolic composition of the corresponding grapes was investigated using a multivariate approach. Grape extracts and wines were produced from 55 different grape samples, covering 8 different Vitis vinifera cultivars: Alicante, Merlot, Syrah, Cinsault, Grenache, Carignan, Cabernet Sauvignon, and Mourvedre. The phenolic composition of the grapes and wines showed that the average ratios between wine and grape phenolics ranged from 0.25 to 7.9 for the different phenolic compounds. Most interestingly, the average ratios were low for anthocyanins (0.31) and tannins (0.32), intermediate for (+)-catechin (0.75) and polymeric pigments (0.98), and high for gallic acid (7.9). Individual wine phenolics in general correlated well with several grape phenolics, indicating that a multivariate approach might be advantageous for prediction of wine phenolics from grape phenolics analysis. However the use of multivariate prediction of individual wine phenolics from the complete grape phenolic composition only improved the prediction of wine polymeric pigments, whereas wine anthocyanins were predicted with the same precision as from the direct relation with grape anthocyanins. Prediction of color attributes of pH normalized experimental wines from the phenolic profiles of grapes was accomplished using a multivariate approach. The correlation between predicted and measured total wine color was high (r = 0.958) but was very similar to the correlation coefficient obtained for the direct relation between grape anthocyanins and total wine color (r = 0.961). Color due to copigmentation, color due to anthocyanins, and color intensity were also predicted well.