Modeling quality of premium spanish red wines from gas chromatography-olfactometry data

Abstract

The aroma compositions of 25 premium Spanish red wines have been screened by quantitative gas chromatography-olfactometry and have been related to the quality scores of the wines. The study has shown that up to 65 odorants can be present in the aroma profiles of those wines, 32 of which have been detected in less than half of the samples. One new odorant is reported for the first time in wine [(Z)-2-nonenal], and only 11 odorants, most of them weak and infrequent, remain unknown. Quality was not positively correlated with any single compound or with any olfactometric vector built by the summation of odorants with similar odors. However, an olfactometric vector built by the summation of the olfactometric scores of defective odorants, such as 2-methoxy-3,5-dimethylpyrazine, 4-ethylphenol, 3-ethylphenol, 2,4,6-trichloroanisole, and o-cresol was significant and negatively related to quality. Quality could be satisfactorily explained by a simple partial least-squares model (79% explained variance in cross-validation) with just three X-variables: the aforementioned defective vector, a second vector grouping 9 other compounds with negative aroma nuances, and the fruity vector, grouping 15 compounds with fruit-sweet descriptors. This result shows that the quality of these red wines is primarily related to the presence of defective or negative odorants, and secondarily to the presence of a relatively large number of fruit-sweet odorants. Remarkably, only in a few low-quality samples could defective aroma nuances be detected, which suggests that defective and negative odorants exert a strong aroma suppression effect on fruity aroma.