Sensory Impact of Extended Maceration and Regulated Deficit Irrigation on Washington State Cabernet Sauvignon Wines

Descriptive analysis.

The main goal of this study was to assess the sensory impact of extended maceration applied to Cabernet Sauvignon grapes produced using four different RDI protocols. To that end, a descriptive analysis with a trained panel was conducted to determine specific sensory effects of both the maceration and the RDI treatments. Potential synergistic or antagonistic sensory effects of extended maceration on the fruit of the different RDI treatments were also explored. A fixed-effect two-way ANOVA with interaction was performed on the sensory data (Table 5). The RDI treatment influenced the sensory profile of the wines, with particular impact on color, taste (bitterness), and mouthfeel (astringency) components. However, the effect of the maceration treatment prevailed in all the sensory attributes, a trend also previously reported for the wine chemical composition.

A significant RDI treatment × winemaking interaction was observed for the purple and red color components, black fruit aroma, and bitterness. The significance of the interaction for these attributes called for reevaluation of the treatment effects and first-order interactions by one-way ANOVA (Table 6). The purple component was more (negatively) affected in the 70% ET_c treatment than in the other RDI treatments upon application of EM. Additionally, a two-way ANOVA applied to the previously reported anthocyanin content of the wines at day 250 (coincident with the sensory analysis) revealed a small yet significant RDI × winemaking effect for anthocyanins (p = 0.044) (Casassa et al. 2013b). This effect followed the same direction observed in the visual purple component, which explains the observed sensory interaction.

There was also a significant RDI × winemaking interaction for the black fruit aroma component (Table 5). A one-way ANOVA indicated that the perception of the black fruit aroma decreased proportionally more in the 25% ET_c treatment than in the other treatments upon application of EM, although the reason for this decrease is unknown. However, extended maceration in general decreased the perception of the black fruit aroma (Table 6). Recently, Callejón et al. (2012) reported that in microfermentations (1 L) of Cabernet Sauvignon wines, some norisoprenoids such as β-damascenone could be selectively bound by the skins and other fermentation solids during prolonged maceration, thereby decreasing their volatility. In turn, β-damascenone has been shown to synergistically increase the perception of the black fruit note (Pineau et al. 2007), which may explain the lower ratings of the black fruit attribute in extended maceration wines. Lastly, bitterness was also comparatively more accentuated in the 70% ET_c treatment upon EM than in the wines of the other RDI treatments.

Principal component analysis.

To further explore the comparative influence of both the maceration and the RDI treatments on the sensory profile of the wines, the full data set was submitted to principal component analysis (PCA), including the wine replicates and the repeated measures performed during the formal sensory sessions. The PCA biplot and confidence ellipses were constructed with 95% certainty according to the Hotteling’s test (Husson et al. 2005), which provides significance testing. Only the first two principal components with eigenvalues greater than or equal to one were retained (Figure 1). The PCA plot showed a bidimensional model that explained 94% of the observed variability. Dimension 1, which explained ~77% of the variability, was a function of the maceration treatment, which suggests a comparatively higher impact of the EM treatment over the RDI treatments on the sensory profile of the wines. Irrespective of the RDI treatment, control wines clustered in the negative region of dimension 1, whereas EM wines grouped on the positive region of dimension 1 (Figure A). There were strong correlations between EM wines and the red and brown color components, oxidized character, astringency and bitterness, all located in the positive side of dimension 1 (Figure 1B). It has been previously shown that extended maceration for 30 days or more increased astringency ratings by 22% in Cabernet Sauvignon (Scudamore-Smith et al. 1990) and by 34% in Merlot (Casassa et al. 2013a). Similarly, in the present study, astringency ratings were, on average, 25% higher in EM wines relative to the control wines (Table 5).

• Download figure
• Open in new tab

Figure 1

Principal component analysis of descriptive sensory data of Cabernet Sauvignon wines evaluated by a trained panel (n = 15): (A) confidence ellipses based on multivariate distribution of Hotelling’s test for p < 0.05 indicating 95% confidence intervals and (B) sensory loadings. RDI treatments: 100% ET_c, 70% ET_c, and 25% ET_c represent replenishment of 100%, 70%, and 25%, respectively, of full-vine ET_c from fruit set through harvest; 25/100% ET_c represents 25% ET_c from fruit set to veraison followed by 100% ET_c from veraison to harvest. Winemaking treatments: C, control (10 days skin contact); EM, extended maceration (30 days skin contact).

Bitterness perception was also affected by extended maceration. Although previous reports of the effects of extended maceration on bitterness perception are not conclusive (Yokotsuka et al. 2000, Casassa et al. 2013a), the present work has shown an average increase of 26% in the bitterness ratings of EM wines relative to those of the control wines (Table 5). This increase can also be further corroborated by inspection of the relative position of the ellipses for the EM wines and the loadings for bitterness in the PCA analysis. Monomeric flavan-3-ols content was generally higher in EM wines (Table 4). Because catechin and epicatechin are known to primarily elicit bitterness (Lesschaeve and Noble 2005), higher amounts of these monomers in EM wines may explain the observed higher bitterness ratings. In addition, there is a potential cooperative contribution to overall perceived bitterness perception by the judges from the oligomeric and protein-precipitable tannin content in EM wines (Table 4), but this hypothesis merits further research.

The PCA sensory loadings on the negative side of dimension 1 (Figure 1B) strongly associated the control wines with the purple component and saturation visual descriptors and the black and red fruit aroma descriptors. Overall CIELab chromatic differences between any given control and EM wines were detected at day 250 (Table 2), and these differences favored the control wines. These results suggest that purple component and saturation were the two color components directly related to higher perceived color in the control wines.

Comparatively higher ratings for fruit aroma attributes in the control wines (or lower ratings of both fruit aroma components in EM wines) (Table 5) can be attributed (but may not be limited) either to a genuine increase in the headspace concentration of these two aroma attributes in the control wines or to a masking effect resulting from the oxidized character in EM wines. Many compounds, including heptenal, methional, and phenylacetaldehyde, have been associated with the perception of oxidation in red wines (San Juan et al. 2012). However, in the context of the present study, acetaldehyde formation by metal-catalyzed oxidation of ethanol during extended maceration seems more plausible (Danilewicz 2003). Acetaldehyde at levels above its detection threshold bears a negative sensory connotation, and this compound was used as a standard for the oxidized character during the training of the sensory panel (Table 1). Furthermore, quinones resulting from the coupled oxidation of polyphenols can readily bind nucleophiles such as mercaptans and thiols bearing fruity notes (Nikolantonaki et al. 2010), thus lowering perceived fruitiness. However, the generation of an oxidized character and a decrease in the fruit aroma of the wine may not necessarily be a causative result of the application of extended maceration, as other studies have found no significant increases of the oxidative character upon extended maceration (Casassa et al. 2013a) or no effect of it in the wine’s fruit flavor (Harbertson et al. 2009). In the present study, the oxidized character did develop after the extended maceration, although the ratings for the oxidized character in these wines averaged 3 cm on the 15 cm unstructured scale (Table 5), which is in the low perception range of the scale. However low this oxidative character, it was enough to reduce the ratings of red and black fruit in the EM wines by 26% and 21%, respectively, relative to the control wines.

The sensory impact of the RDI treatment was a function of dimension 2, which explained 17% of the residual variability of dimension 1. Wines from the 100% ET_c treatment clustered at the lower end of dimension 2 (Figure 1A), indicating the lowest perception for color, fruity aromas, and mouthfeel attributes. The 100% ET_c treatment was included in the experimental design to explore the relationship between the application of an irrigation protocol based on replenishment of the full evapotranspirative demand and the generation of vegetative aromas in the resulting wines under the premise that a more vigorous canopy would lead to an enhancement of vegetative aromas (Allen and Lacey 1993). However, the vegetal component was confirmed to be absent in the 100% ET_c wines (and in all the wines) and thus was not selected as a discrimination term during the panel evaluations. This result suggests that, under some combinations of cultivar, clone, climate, and soil, supplying the full evaporative demand does not necessarily lead to vegetative flavors in the resulting wines.

The wines of the 70% ET_c and 25/100% ET_c treatments were distributed in the medial region of dimension 2 defined by the aromatic descriptors red and black fruit aroma (control wines) and the chromatic descriptors red and brown component (EM wines) (Figure 1). Although not necessarily causative, the application of different RDI protocols have been linked to enhanced concentration of fruity norisoprenoids in the resulting wines (Qian et al. 2009, Ou et al. 2010). The interconversion of carotenoids to odorant C₁₃-norisoprenoids has yet to be conclusively established, but it is believed that reduced vine vigor may result in increased sun exposure and berry temperature in the fruiting zone, thereby leading to carotenoid degradation with the concomitant increase in C₁₃-norisoprenoids precursors (Kondouras et al. 2006, Bindon et al. 2007). While we did not identify these compounds analytically, the generalized trend for the sensory results presented here is consistent with these previous studies. Finally, the 25% ET_c treatment generated wines with higher purple component and saturation in the control wines, consistent with comparatively higher content of anthocyanins and small polymeric pigments in these wines (Table 3). When the 25% ET_c treatment was processed with EM, the sensory profile shifted toward a pronounced oxidized character, bitterness, and astringency.

Also of significance was the overlap of the 70% ET_c control and 25/100% ET_c control wines on the left side of dimension 1 (Figure 1A). The ellipses represent empirical descriptions of the variability of the sensory evaluations (Husson et al. 2005), and if the ellipses are superimposed, then the wines are not significantly different from a sensory standpoint. Conversely, the general absence of overlap in six out of the eight treatments highlights the magnitude of the sensory differences. Furthermore, the compression of the ellipses is an indication of the reliability of the panel data.

Correlation between chemical and sensory data.

The relationship between the significant chemical features of the wines and their sensory properties was examined by canonical correlation analysis (Figure 2). In addition to an agglomerative hierarchical cluster analysis of the chemical and sensory data sets, the correlation coefficients between any given pair of variables in the two retained dimensions are also provided (González et al. 2012). Shared cluster membership indicates the strength of the relationship between the canonical variates for both the chemical and sensory variables. For the chemical and sensory data sets, the cluster containing the highest number of variables was labeled cluster 1, the cluster with the next highest number of variables was labeled cluster 2, and so forth. For the chemical analyses, cluster 1 was composed of all the anthocyanin derivatives (except cyanidin), skin-derived proanthocyanidins, total flavonols, quercetin derivatives, small polymeric pigments, the red CIELab component (a*), and saturation (C*). Cluster 2 was composed of flavan-3-ols, oligomeric and polymeric proanthocyanidins, seed-derived proanthocyanidins, protein precipitable tannins, iron reactive phenolics, lightness (L*), and cyanidin derivatives. Cluster 3 was composed of large polymeric pigments, the yellow CIELab component (b*), and the hue angle (H*). For the sensory data set, three main clusters were identified. Cluster 1 was composed of astringency, bitterness, oxidized character, and red and brown color components. Cluster 2 was composed of red and black fruit aromas, whereas cluster 3 consisted of saturation and purple component. To understand which chemical variables were positively or negatively correlated with the sensory variables, the cluster image map provides the observed correlation coefficient values between any pair of variables. Positive correlation coefficients close to 0.93 were observed between protein precipitable tannins, iron-reactive phenolics, and oligomeric proanthocyanidins as well as astringency, bitterness, and brown component. These sensory attributes defined the sensory profile of extended maceration wines, thus suggesting that for the wines analyzed here, bitterness and astringency are highly correlated with the content of oligomeric proanthocyanidins (mean degree of polymerization up to 5 flavan-3-ols subunits) and also with protein-precipitable tannins, seed-derived proanthocyanidins, and overall phenolic concentration. Likewise, positive correlation coefficients were observed between the color sensory attributes saturation and purple component and the chemical attributes flavonols, quercetin derivatives, small polymeric pigments, a*, C*, and malvidin and delphinidin derivatives. As control wines were chromatically (Table 2) and sensorially (Table 5) more colored than EM wines, the aforementioned chemical variables may most likely be responsible for the perceived color features of control wines. In particular, a positive correlation between small polymeric pigments, C*, and monomeric anthocyanins has been previously reported in Merlot wines obtained after 10 days of skin contact (Casassa et al. 2013a), consistent with the results reported here. Positive correlations were also observed between flavan-3-ol content and the perception of both astringency and bitterness, suggesting that at the concentration of these monomers in EM wines (overall mean: 346 ± 37 mg/L, n = 8), the perception of both bitterness and astringency could be at least partially explained by the simultaneous occurrence of a comparatively higher concentration of flavan-3-ols and oligomeric proanthocyanidins. Nevertheless, since astringency and bitterness were also correlated with multiple phenolic measures (e.g., oligomeric proanthocyanidins, protein-precipitable tannins, iron-reactive phenolics), the occurrence of a single causative effect for both sensory features is unlikely.

• Download figure
• Open in new tab

Figure 2

Canonical correlation analysis showing clustered image maps of the correlation between significant sensory and chemical attributes considering all wines (n = 16) in the experiment. The dark red and blue colors indicate positive and negative correlation coefficient values, respectively, whereas light green colors indicate near zero correlation coefficient values. Cluster membership within sensory and chemical data is indicated as different shades of grey and black. L*: lightness; PAs: proanthocyanidins; b*: yellow component; H*: hue; a*: red component; C*: saturation.