Quality mapping: An exploratory approach for evaluating the chemicals driving in-mouth quality in red wines

Zusammenfassung

Non-volatile molecules have been reported to be important molecules driving wine sensory properties; however the wine industry is far from knowing their exact role in wine quality perception. The present work aims to establish correlations between (1) the non-volatile composition of a set of red wines and their in-mouth sensory attributes and (2) non-volatile composition and the quality scores given by a panel of experts on the same set of wines. A group of 21 wine experts rated the in-mouth quality of 16 Spanish red wines with different non-volatile composition on an unstructured scale anchored with "nil" on the left and "high" on the right. To avoid aroma and color influence in quality assessment, the evaluation was carried out using nose clips and dark glasses. In parallel, a descriptive analysis method was used to evaluate the intensity of in-mouth sensory properties; further non-volatile molecules with known sensory activity were quantified. Results show an exponential relationship between the in-mouth attribute sourness and both tartaric and succinic acids. Succinic acid and to a lesser extent the alcohol content present exponential as well as t-aconitic acid are exponentially correlated with astringency as are procyanidins B1 and B3 and the total polyphenolic index, which are linearly correlated. Results show that experts have heterogeneous in-mouth quality concepts. Experts are divided into three main clusters, and their quality judgments were regressed onto the significant factors derived from the PCA calculated on the 22 non-volatile molecules with sensory activity. Seventy-one per cent of expert quality responses were fitted by the elliptical model, while 29% were fitted by the vector model. Experts belonging to cluster 1 (fitted by the vector model: "the more the better") find that high contents in protein-precipitable PAs and in c-aconitic acid result in high quality exemplars. Clusters 2 and 3 quality scores were significantly predicted by elliptical models and their anti-ideal and ideal quality points were respectively defined. © 2014 by Nova Science Publishers, Inc. All rights reserved.