Combining Olfactory and Gustatory Clues in the Judgment of Aging ...

Combining Olfactory and Gustatory Clues in the Judgment of Aging Potential of Red Wine by Wine Professionals Jennifer Langlois,1* Jordi Ballester,1,2 Eva Campo,1 Catherine Dacremont,1 and Dominique Peyron1,2 Abstract: The study was aimed at determining the relative importance of olfactory and gustatory information for the vin de garde concept, which refers to wine that is intended for aging. First, the study investigated how olfactory information contributed to the assessment of the aging potential of young wine. Burgundy wine professionals categorized 26 red wines (vintage 2005) according to their aging potential in two experimental conditions: orthonasal evaluation and overall evaluation. Second, the aromatic notes associated with wines perceived with an aging potential were examined. A trained descriptive panel described the odor of the 26 red wines to determine whether the wines judged by wine professionals to have an aging potential (orthonasal evaluation) were characterized by specific aromatic notes. Third, the study investigated how taste and mouthfeel contributed, together with odor, to judgments of aging potential. The trained descriptive panel also assessed taste (sweetness, acidity, and bitterness) and mouthfeel (alcohol and astringency) intensity of the wines. Results showed that odor provides some clues for professional judgment of aging potential. These olfactory clues are sufficient to assess the aging potential for some wines, but not all. Wines judged to have aging potential by orthonasal evaluation tended to have woody, caramel, roasted, and prune notes. Taste and mouthfeel also provided clues for professional judgment of aging potential. Red wines judged to have aging potential by overall evaluation tended to have high astringency and low acidity. Thus, the combination of gustatory and olfactory information allowed professionals to judge the aging potential of wines. The vin de garde concept involves a wide variety of complex perceptual clues (color, odor, taste, and mouthfeel). Key words: aging potential, red wine, wine professionals, trained panelists

A key trend in the wine market is toward simple and fruity wines that can be consumed immediately (Araujo 2005, Gibson 2006, Spurr 2006). Consumers are not necessarily able to age wine, as the practice requires both a monetary investment and cellaring facilities (Araujo 2005). However, there is also a market for high-quality wines with great aging potential, the so-called vins de garde. They are wines made to be kept several years or even decades in cellar and expected to improve during aging (Ribéreau-Gayon 2000). They have a prestigious image linked to tradition and refined winemaking in agreement with local, consistent, and equitable practices. They are highly valued because it is assumed that a vin de garde from a great vintage will become a magnificent wine (Johnson and Robinson 2002). Not all wines have the ability to age well and consumers often want to know whether a wine is suitable for aging.

Consequently, wine professionals have to evaluate aging potential, that is, whether a young wine will become a high-quality wine over time. The characteristics on which wine professionals base their judgments regarding the aging potential of a young red wine are not clearly defined, as illustrated by French enologist Denis Dubourdieu: “How to define initially a vin de garde? We know for the white, we don’t know for the red!” (Dupont 2005). Comments from popular wine literature may illustrate this lack of consensus: “A young vivid wine with fruity aromas: that is the profile of red vin de garde” (Dupont 2005) and “Only wines perfectly composed, i.e., with a good structure presenting a balance between tannins and acidity, have a sufficient longevity to achieve this ideal period of development” (Dovaz 2004). Despite this diversity of opinion, French wine professionals have been in close agreement when judging the aging potential of red wines (Jaffré et al. 2007, 2009), demonstrating a shared vin de garde concept. When wine professionals judge the aging potential of a young wine, they consider existing characteristics and future characteristics that they may infer from their knowledge of the wine aging process, including physicochemical changes that will impact color, taste, mouthfeel, and aroma. For instance, phenolic compounds, which are related to color, astringency, and bitterness in wine, undergo structural transformations with time during aging. These modifications progressively modify the sensory properties of wine: the rather bright red color of young wines will decrease in intensity with a simultaneous increase in lightness (Pérez-Magariño

Centre des Sciences du Goût et de l’Alimentation, UMR 6265 CNRS, 1324 INRA, Université de Bourgogne, AgroSup Dijon, 15 rue Hugues Picardet, 21000 Dijon, France; and 2IUVV Jules Guyot, Université de Bourgogne, 1 rue Claude Ladrey, 21078 Dijon, France. *Corresponding author (email: [email protected]; fax: +33 80 68 16 01) Acknowledgments: The authors thank the enologists, winemakers, technological managers, enology professors, and all panelists who participated in this work for their availability. Manuscript submitted Mar 2009, revised July 2009, accepted Sept 2009 Copyright © 2010 by the American Society for Enology and Viticulture. All rights reserved. 1

15 Am. J. Enol. Vitic. 61:1 (2010)

16 – Langlois et al.

and González-San José 2006) and a change toward brickorange hue (Monagas et al. 2007). A decrease of astringency related to phenolic compound modifications is also observed during aging (Mateus et al. 2004, Vidal et al. 2004). A recent study investigated professional judgments of aging potential by contrasting the structure and color of red wines judged with and without an aging potential (Jaffré et al. 2009); wines with an aging potential were characterized by a higher astringency level and a saturated color. In addition to color and structure, aroma also evolves during wine aging. There is a decrease of fresh fruity notes (Jackson 2002) and an evolution toward more complex and finer aromas, with a blending of aromatic notes that leads to a homogeneous and harmonious odor (Peynaud and Blouin 1999). When wine professionals discuss the sensory characteristics of wines with an aging potential that will evolve during storage, they primarily use terms relative to tannins and wine structure, but also terms related to aroma (Jaffré et al. 2007). Professionals were asked to assess the aging potential of wines according to three conditions: visual evaluation only, mouthfeel evaluation only, and overall evaluation (visual, olfactory and mouthfeel, according to the usual procedure used by wine professionals) (Jaffré et al. 2009). Results indicated that the combination of visual evaluation and mouthfeel evaluation was not sufficient to predict the outcome of the overall evaluation. Altogether, these results suggest that beyond astringency and color, aromatic notes could be involved in aging potential judgments by wine professionals. The present study was designed to determine the relative importance of odor, taste, and mouthfeel for the vin de garde concept, with three subobjectives. The first was to determine how olfactory information contributed to the assessment of aging potential when wine is young. Wine professionals assessed the aging potential of red wines, first by orthonasal evaluation only (odor) and second by overall evaluation (color, odor, and f lavor.) Young red wines from Burgundy and several wines from other regions were used in the assessment. We hypothesized that some specific aromatic notes may contribute to the assessment of aging potential by wine professionals, but that orthonasal judgments will lead to different outcomes than overall judgments. Second, the aromatic notes associated with wines with aging potential were studied. A trained descriptive panel described the odor of wines with and without aging potential as assessed by wine professionals by orthonasal evaluation. The hypothesis was that wines with aging potential when judged by orthonasal evaluation are characterized by specific aromatic notes. Third, the study investigated how taste and mouthfeel, together with odor, contributed to judgments of aging potential. The trained descriptive panel also assessed the intensity of taste (sweetness, acidity, and bitterness) and mouthfeel (alcohol and astringency) of the wines. Intensity scores were combined with orthonasal assessments of aging potential in a multiple linear regression to explain overall assessments of aging potential. The hypothesis was that the discrepancy

between aging potential judgments made by orthonasal evaluation and overall evaluation can be partially explained by taste and/or mouthfeel characteristics of the wines.

Materials and Methods Wines. Twenty-six commercial red wines, vintage 2005, were selected: 20 Burgundy wines with different appellation levels, two Côtes du Rhône wines, two Bordeaux wines, and two wines from Spain (Table 1). Half of the wines from each origin were considered to have aging potential, according to the judgment of our providers (winemakers, enologists, cellar specialists, and wine merchants). Wines were tested approximately two years after harvest, in June/July 2007). Judgment of aging potential by wine professionals. Wine professionals performed a categorization task regarding wine aging potential in two experimental conditions: orthonasal evaluation, and overall evaluation, in two separate sessions. Twenty-three professionals in the wine field in Burgundy (6 women and 17 men ranging from 32 to 74 years, median age 42) took part in the experiment. There were enologists, winemakers, technical managers, and enology professors. Categorization by orthonasal evaluation. In a first session, wine professionals performed a categorization task

Table 1 Wines and their a priori allocation. Wine

Code

With aging potential Bourgogne Bourgogne Hautes Côtes de Nuits, Les Genevrières Chambolle-Musigny Gevrey-Chambertin Nuits St Georges, Les Allots Côte Chalonnaise Grande Réserve Mercurey 1er cru, Les Velleys Mercurey 1er cru, Perrières Beaune 1er cru, Les Teurons Bordeaux Lalande de Pomerol Côtes du Rhône Crozes, Hermitage Spain Toro, Corral de Campanas

AP1 AP2 AP3 AP4 AP5a AP6a AP7a AP8a AP9a AP10a AP11 AP12 AP13

With poor aging potential Bourgogne Bourgogne Hautes Côtes de Beaune Savigny les Beaune 1er cru, Les Lavières Bourgogne Bourgogne Hautes Côtes de Nuits Côte Chalonnaise Réserve Rully, Les Chauchoux Chorey-les-Beaune Bordeaux Château de Lucques Côtes du Rhône Villages Cairanne, Le Ventabren Spain Montemar

NP14 NP15 NP16 NP17 NP18a NP19a NP20a NP21a NP22a NP23a NP24 NP25 NP26

a

Wines duplicated for the descriptive analysis.

Am. J. Enol. Vitic. 61:1 (2010)

Judgment of Aging Potential of Red Wines – 17

according to the aging potential of wines by smelling only (orthonasal). All wines were served in dark wineglasses to eliminate information related to aspect and color. A 20 mL sample was poured into each glass, covered with a petri dish, and coded with a random three-digit code. This experimental condition was performed first because the other sensory information memorized during a previous overall evaluation of the wines could influence orthonasal evaluation. Categorization by overall evaluation. In a second session, wine professionals performed a categorization task according to the aging potential of wines by an overall evaluation (visual, olfactory, and mouthfeel). Assessment was performed using clear wineglasses according to the usual wine-tasting procedure of the professionals. A 40 mL sample was poured into each glass, covered with a petri dish, and coded with a random three-digit code. Experimental procedure. For each categorization task, wine professionals assessed the 26 wines, presented according to a Latin Square arrangement, and answered yes or no to the question: “For you, is this wine a wine with an aging potential?” They also indicated the confidence level (not sure, moderately sure, or very sure) of their answer. Wine professionals were informed only that the wines to evaluate were red wines of the same vintage. All wines were served at room temperature (~20°C) and were evaluated in individual booths. For practical reasons, several sessions were organized with subgroups of typically three to four participants. Between sessions, the air of an opened bottle was extracted with a vacuum pump and wines were stored at 6°C for a maximum of two days. Descriptive analysis by trained panel. Descriptive analysis is classically used to evaluate sensory properties of food and beverages. It relies on the ability of a trained panel to identify and quantify the relevant sensory properties of a set of products. This approach, based on attributes rating, has been used to describe wines in numerous studies (e.g., Cristovam et al. 2000, Guinard and Cliff 1987, Heymann and Noble 1987, Noble et al. 1984, Perrin et al. 2007). This approach was used for taste and mouthfeel perceptions in the present study. For odor description, the method was based on attribute citation frequencies previously used to evaluate wine odor (Campo et al. 2008, McCloskey et al. 1996, Piombino et al. 2004). Panelists select from a list the most appropriate odor attributes to characterize wine aroma. This approach seems well adapted for wines because of the potentially wide diversity of aromatic notes (Lawless 1999). It is an alternative to the classic descriptive analysis especially suitable when the issue is to identify the aromatic notes that characterize a set of wines (Campo et al. 2010). Participants and training. Thirty-three participants (15 women and 18 men ranging from 20 to 68 years, median age 27) were recruited from students and staff at the Université de Bourgogne. They volunteered to be trained in wine description one hour per week for at least eight months. Panelists participated in at least 20 one-hour training sessions (one session per week). At the beginning of the

training, panelists were provided with 115 terms similar to a previously developed list (Campo et al. 2008), which was obtained by compiling terms from other lists used in the description of wines from different varieties and geographical origins (Aznar et al. 2003, Le Fur et al. 2003, Noble et al. 1987, Peynaud and Blouin 1999, Piombino et al. 2004). Terms were arranged by odor families—fruity, floral, spicy, plant, animal, burnt, undergrowth, and other—following the Wine Aroma Wheel (Noble et al. 1987). The training consisted of two steps, general (14 sessions) and specific (5 sessions), during which panelists worked in subgroups. During general training, panelists became familiar with odor attributes and with intensity rating of taste and mouthfeel attributes. A typical session began with odor references for a subset of odor attributes. For each attribute, panelists were provided with several references, natural products and/ or commercially available aromas, and they chose the most suitable by consensus. Panelists then described three to four wines by choosing the most relevant odor attributes in the list and by rating sweetness, acidity, bitterness, astringency, and alcohol intensity on a six-point scale (0 = absence, 1 = very low, 2 = low, 3 = moderate, 4 = high, and 5 = very high). The session ended with a discussion during which the panel leader compared the intensity scores and highlighted the most frequently cited terms for each wine. Wines for general training were selected to present intense and easily recognizable aromatic notes and included red, white, and rosé wines of diverse grape varieties and origins. During specific training, panelists became familiar with the type of wines in the present study. They described odor and rated sweetness, acidity, bitterness, astringency, and alcohol intensity of 16 red wines, 14 of which were Burgundy wines, vintage 2005. During this training step, panelists discussed and modified the initial attributes list, eliminating irrelevant, ambiguous, and redundant terms and adding pertinent terms. At the end of the training the final list included 115 attributes organized hierarchically (Table 2). Experimental procedure. Trained panelists carried out descriptions of the 26 wines in individual booths. In order to assess panel reproducibility, 12 wines were duplicated (Table 1). The 38 samples were divided into five blocks of seven or eight wines with approximately the same ratio of wines with/without an aging potential in each. Trained panelists took part in five one-hour sessions assessing one block per session. Samples of 20 mL wine were presented according to a Latin Square arrangement, in dark wineglasses covered with petri dishes, and coded with random three-digit codes. All wines samples were served at room temperature (~20°C). In each block, the samples were first evaluated orthonasally (sniffing after swirling) to describe odor of wines and, after a 5-min break, the same samples were evaluated in mouth to assess taste and mouthfeel intensity of wines. For odor description, panelists were asked to choose a maximum of five attributes from the list of 115 according to the following instructions: “Eight wine samples coded with three-digit number are presented. You have to smell

Am. J. Enol. Vitic. 61:1 (2010)


these samples in given order, from left to right, and for each sample to check in the list the most appropriate attributes to describe each sample (five attributes maximum per wine). You can indicate one or more attributes that are not in the list.” For taste and mouthfeel assessments, panelists rated the intensity of sweetness, acidity, bitterness, astringency, and alcohol on a six-point scale (0 = absence, 1 = very low, and 5 = very high) for each sample. Panelists were not informed of the specific objective of the study. However, after the specific training they knew they tasted red wines mainly from Burgundy. Panel performance. For odor description, individual performances were examined from the 12 duplicated wines with an average reproducibility index (Campo et al. 2008). This reproducibility index represents the proportion of common attributes used to describe the same wine at the two replications. It varies from 0, no common attribute, to 1, perfect reproducibility. According to the distribution of these indexes for the 33 panelists (Figure 1), the values ranged from 0.20 to 0.64 (mean 0.39), indicating a satisfactory reproducibility of the panel (Campo et al. 2008). To

Table 2 Final list of 115 attributes (in French for the experiment), including odor families and odor notes, used for descriptive analysis. Odor family Fruity White fruit Yellow fruit Dried fruit Nuts Tropical Citrus Red fruit Black fruit Other

Odor note Quince, apple, pear Apricot/peach, melon Date, dried fig, prune Almond, walnut, hazelnut Pineapple, banana, passionfruit, lychee, mango, coconut Bergamot, lemon, orange, grapefruit Cherry, strawberry, raspberry, redcurrant Blackcurrant, blueberry, blackberry Cooked fruit, candied fruit, muscat, kernel/ bitter almond

Floral

Wattle, chamomile, honeysuckle, orange blossom, geranium, jasmine, lilac, rose, lime blossom, violet

Spicy

Anise/fennel, cinnamon, clove, curry, juniper berry, ginger, laurel, nutmeg, pepper, licorice, thyme, vanilla

Plant Vegetal Other

assess the panel performance regarding intensity ratings, a two-way analysis of variance (ANOVA) was performed; panelists, wines, and their interaction (fixed model) were run on intensity scores for sweetness, acidity, bitterness, astringency, and alcohol for the 12 replicated wines. No significant panelists × wines interaction was found, indicating a good agreement between panelists. Wine factors were significant for every taste and mouthfeel attribute (acid F = 4.14, p < 0.001; sweet F = 4.92, p < 0.001; astringent F = 12.19, p < 0.001; alcohol F = 2.38, p < 0.01) but not for bitterness (F = 1.56, p = 0.11), indicating satisfactory discrimination abilities of the panel.

Results Contribution of olfactory information to assessment of aging potential. For each experimental condition, orthonasal and overall evaluation, data were summarized in a co-occurrence matrix counting the number of times two particular wines were categorized in the same way (both with aging potential or both without aging potential). The matrices were analyzed by multidimensional scaling (MDS) with a nonmetric analysis. Coordinates of wines in MDS configurations were submitted to hierarchical cluster analysis with the Ward method. For the two categorization tasks, wines were separated in two clusters (Table 3). For the overall evaluation, the average percentage of the answers “yes, it is a wine with an aging potential” was 68% for cluster 1 and 28% for cluster 2. For the orthonasal evaluation, this average percentage was 71% and 34% for clusters 1 and 2, respectively. Thus, in the two conditions, cluster 1 represented the wines categorized with an aging potential and cluster 2 the wines categorized without aging potential. In the overall evaluation condition, cluster 1 included 13 wines: seven Burgundy wines, and the six wines from other areas. In the orthonasal evaluation condition, cluster 1 included only seven Burgundy wines, six of which (AP4, AP5, AP6, AP8, AP10, and NP17) were also assessed with an aging potential by overall evaluation. These results

Artichoke, asparagus, cabbage, celery, green bean, bell pepper, olive hay/dried leaf, herbaceous, pine/resin, menthol/fresh

Undergrowth

Mushroom, humus/earth, moldy

Burnt/woody Burnt Woody

Caramel, roasted coffee/chocolate, toast Fresh wood, smoke

Animal

Leather, meat extract, musk/civet, cat urine, wet dog

Other

Alcohol, kirsch, butter, lactic, yeast, cookie, fruit drop, honey, cider, rubber, cardboard, incense, ink, bitumen, flint, dust, sulfur, cold tobacco, sweat

Figure 1 Distribution of reproducibility of individual panelists, as measured by the average of reproducibility indexes applied to attributes given to the same wine on replicate presentations (n = 33).

Am. J. Enol. Vitic. 61:1 (2010)


indicate that olfactory clues are sufficient to assess the aging potential for some wines, but not all, and suggest that olfactory information participates in the definition of the sensory space of vin de garde, together with other sensory characteristics. Aromatic notes associated with wines judged with an aging potential. For the odor description performed by the trained panel, only attributes cited by at least five panelists (15% of the panel) for at least one wine were considered for data analysis (Campo et al. 2008, Piombino et al. 2004), leading to consideration of 29 attributes. Attribute citation frequencies were analyzed through correspondence analysis (CA) using SPAD software (ver. 5.5, Cisia-Ceresta, Montreuil, France). For replicated wines, the average citation frequency of the two replications was considered to perform the CA. A first analysis showed that wine NP26 (from Spain), described by attributes from the fruity family (banana, fruit drop, pear, and strawberry), was isolated from the other wines on axis 1, indicating a wine very different from the others in terms of odor characteristics. Thus, a second analysis was carried out without this wine. Wines and attributes were projected into the 1-2 subspace (Figure 2). The first dimension, explaining 17% of the total variance, opposes fruity attributes (cooked fruits, fruit drop, banana, and cherry) to bell pepper, green bean, wet dog, and humus/earth attributes. The second dimension, explaining 13% of the total variance, opposes the caramel, vanilla, toast, and leather attributes to alcohol, humus/ earth, pear, licorice, fruit drop, and strawberry. Hierarchical cluster analysis with the Ward method was applied to all factorial coordinates of the wines in the CA space us-

Table 3 Clusters yielded by the hierarchical cluster analysis using wines coordinates from three-dimensional MDS space for the categorization task by overall evaluation and for the categorization task by orthonasal evaluation. Cluster 1: wines judged with an aging potential by wine professionals; cluster 2: wines judged without (w/o) aging potential by wine professionals. Overall evaluation

ing SPAD software (ver. 5.5; Cisia-Ceresta). The terms that best characterized each of the clusters were identified by using the test-value parameter (Morineau et al. 1995); the attributes that showed statistically significant test-values ( p < 0.05) for each cluster were kept. This hierarchical cluster analysis yielded two stable clusters separated by axis 1: on the right side wines with fruity notes and on the left side wines with vegetal and animal notes. However, the partition in eight clusters was the best partition option to account for odor characteristics of wines with an aging potential (Table 4). Cluster 2 included nine wines, six of which (AP4, AP5, AP6, AP8, AP10, and NP20) were categorized orthonasally by wine professionals as having aging potential. These wines were significantly described by woody, caramel, and prune attributes. Two fruity attributes, blackberry and cherry, also characterized these wines but were only significant at 10% level. Wine NP17, also categorized orthonasally with an aging potential by wine professionals, was alone in cluster 8 and was described by bell pepper and roasted coffee/chocolate. Vanilla note also characterized this wine but was only significant at 10% level. The importance of burnt (caramel, roasted coffee/chocolate) and woody (woody, vanilla) notes for wines with an aging potential may be related to the idea that most red wines with a good potential for aging are matured in oak barrels (Jackson 2002). Storage in oak barrels usually speeds wine maturation and adds complementary oaky, vanilla, or spicy/smoky flavors (Garde-Cerdán and Ancín-Azpilicueta 2006, Jackson 2002). Actually, the woody character was often mentioned when wine professionals were questioned regarding the sensory characteristics of wines with an aging potential (Jaffré et al. 2007). Some fruity notes (prune, blackberry, and cherry) also characterize wines with an aging potential, in agreement with some evocations of vin de garde in popular wine literature, such as: “A modern style,

Orthonasal evaluation

Cluster 1: with potential

Cluster 2: w/o potential

Cluster 1: with potential

Cluster 2: w/o potential

AP3 AP4 AP5 AP6 AP8 AP10 AP11 AP12 AP13 NP17 NP24 NP25 NP26

AP1 AP2 AP7 AP9 NP14 NP15 NP16 NP18 NP19 NP20 NP21 NP22 NP23

AP4 AP5 AP6 AP8 AP10 NP17 NP20

AP1 AP2 AP3 AP7 AP9 AP11 AP12 AP13 NP14 NP15 NP16 NP18 NP19 NP21 NP22 NP23 NP24 NP25 NP26

Figure 2 Projection of attributes and wines in the correspondence analysis space (dimensions 1 and 2). Underscored wines: wines judged by wine professionals with an aging potential according to the categorization task by orthonasal evaluation.

Am. J. Enol. Vitic. 61:1 (2010)


showing a core of polished cherry and blackberry notes, shaded by vanilla. The tannins are beautifully integrated. Elegant and long. Approachable now, but will improve” (Wine Spectator, 31 May 2008, pp. 106-107). However, woody notes appear rather predominant in wines judged with an aging potential, which may be linked to perceptual olfactory interactions between fruity and woody odor compounds. The qualitative dominance of the woody note in three woody-fruity mixtures (isoamyl acetate/whisky lactone, ethyl butyrate/whisky lactone, and ethyl butyrate/ guaiacol) has been shown when the perceived intensities of each unmixed compound were equal (Atanasova et al. 2005). This dominance is in accordance with the observation of a masking effect on some wine odorants when wood predominated (Moio et al. 1993). Taste, mouthfeel, and the difference between orthonasal and overall judgments of aging potential. A two-way analysis of variance (ANOVA) with panelists and wines as fixed factors was performed on intensity scores of sweetness, acidity, bitterness, astringency, and alcohol. For replicated wines, the average intensity score was used to perform the ANOVA. Results indicated that the wines were different for every taste and mouthfeel attribute (acid F = 7.27, p < 0.0001; sweet F = 4.97, p < 0.0001; astringent F = 9.48, p < 0.0001; and bitter F = 2.53, p < 0.0001) except for alcohol (F = 1.09, p = 0.35). In order to combine taste and mouthfeel intensities with assessments of aging potential, an aging potential score was computed from the results of the categorization tasks by the professionals, similar to the six-point certainty scale of Jaffré et al. (2009). For each wine, an individual yes/ no answer was combined with its confidence level. Scores were: 1 = not sure, 2 = moderately sure, 3 = very sure, with a positive sign for “yes” answers and a negative sign for “no” answers. Thus, the aging potential scores ranked from -3 (“no potential for aging” and “very sure”) to 3 (“yes, potential for aging” and “very sure”). Individual scores were averaged over the group of professionals. These average scores were computed separately for overall evaluation and orthonasal evaluation. A multiple linear regression analysis was performed with the aging potential scores for overall evaluation as the dependent variable and intensity scores of taste and mouthfeel attributes and the aging potential scores for the orthonasal evaluation as the independent variables, using the stepwise method (SAS software, ver. 9.1, SAS Institute, Cary, NC). (Variables are added one by one to the model whenever the F statistic for the added variable is significant at the 0.05 level. After adding a variable, the stepwise method looks at each variable already included in the model and deletes any variable that does not produce an F statistic significant at the 0.05 level.) The multiple linear regression provided a significant model (R² = 0.66; p < 0.05) including acid, astringent, and aging potential judgment by orthonasal evaluation (Table 5). The aging potential score for overall evaluation was related negatively to acidity and positively to astringency

Table 4 Clusters yielded by the hierarchical cluster analysis using all factorial coordinates of the wines in the CA space, characterization by attributes showing statistically significant test values. Wine/cluster Cluster 1 AP1 NP25

Cluster 2 AP2 AP3 AP4a AP5a AP6a AP8a AP10a NP20a NP23 Cluster 3 AP12 Cluster 4 NP15 Cluster 5 AP13 NP14 NP18 NP19 Cluster 6 AP9 NP16 NP22

Cluster 7 AP11 NP21 NP24

Cluster 8 NP17a

a

Attributes

Test value

P

Raspberry Banana Strawberry Roasted coffee/chocolate Humus/earth Green bean

2.48 2.38 1.80 -1.66 -1.69 -2.09

0.01 0.01 0.04 0.05 0.04 0.02

Woody Caramel Prune Humus/earth Bell pepper Dust Strawberry

2.00 1.62 1.60 -1.69 -2.03 -2.18 -2.37

0.02 0.05 0.05 0.04 0.02 0.01 0.01

Caramel

1.87

0.03

Dust Blackcurrant Leather

2.26 1.89 1.61

0.01 0.03 0.05

Fruit drop Pear Alcohol Caramel Leather

2.93 2.20 1.85 -2.23 -2.24

0.00 0.01 0.03 0.01 0.01

Humus/earth Green bean Redcurrant Dust Caramel Cooked fruits

2.79 2.71 2.19 1.63 -1.82 -1.89

0.00 0.00 0.01 0.05 0.03 0.03

Bell pepper Wet dog Green bean Pepper Cherry Roasted coffee/chocolate Banana

3.48 2.48 1.95 -1.62 -1.71 -1.71 -1.86

0.00 0.01 0.02 0.05 0.04 0.04 0.03

Bell pepper Roasted coffee/chocolate Alcohol Cherry

2.56 2.51 -1.63 -1.86

0.00 0.01 0.05 0.03

Wines judged by wine professionals with an aging potential according to the categorization task by orthonasal evaluation.

Table 5 Statistical analysis of multiple linear regression (R² = 0.66, p < 0.05). Variable Intercept Acid Astringent Aging potentiala a

Parameter estimate 2.61 -2.40 1.42 0.43

Judgment by orthonasal evaluation.

Am. J. Enol. Vitic. 61:1 (2010)

F value

Pr>F

0.71 6.16 14.64 4.38

0.408 0.022 0.001 0.049


and to the aging potential score for the orthonasal evaluation. This result confirms that red wines judged with an aging potential when they are young tend to be astringent, as reported elsewhere (Jaffré et al. 2009). The result also agrees with the assessment that wines expected to be consumed while young tend to have lighter f lavors, whereas those requiring extended aging are initially tannic and bitter (Jackson 2002).

Discussion The present study revealed that young red wines judged with an aging potential tend to have woody, caramel, roasted, vanilla, and prune notes. The descriptive approach based on attribute citation frequency was suitable for studying vin de garde aroma. The product map obtained by CA explains a relatively low percentage of variance on the first principal components, ~31% on the first two dimensions. This low value is partially due to the high number of attributes considered for analysis, leading to increased variability in the data. One way to increase the variance explained by the first components is to use a more conservative criterion for attribute selection, for instance by selecting only attributes used by at least one-third of the panel for at least one wine. But, this method may lead to trivial descriptions by missing subtle aromatic notes that may not have been detected by a high number of panelists or that might have been reported by a variety of attributes, such as lemon, orange, and grapefruit to refer to the same citrus note. Thus, a compromise has to be found between the richness of the description and the noise level in the data. In this experiment, we chose to favor a detailed description of wines. Even though the total variance explained by the first dimensions of the CA space is relatively low, the map is meaningful and leads to comprehensive interpretation both when evaluating the dimensions of the product space and explaining the clustering regarding aging potential. Taste and mouthfeel contribute together with odor to aging potential judgment by wine professionals. Young red wines judged to have aging potential are also characterized by high astringency and low acidity levels. These results are in agreement with a study that surveyed wine professionals about changes in sensory characteristics during aging for wines with aging potential (Jaffré et al. 2007). The professionals often mentioned terms relative to tannin, wine structure, and aroma. Moreover, visual and in-mouth information seem to be complementary for aging potential judgment by wine professionals (Jaffré et al. 2009). Thus, the present study extends this result, showing the importance of the combined use of olfactory and gustatory clues for aging potential judgment. Vin de garde is a rather complex concept involving a wide variety of perceptual clues. The six Burgundy wines categorized by overall and orthonasal evaluation as wines with an aging potential were of a high-quality appellation (village or “1er cru”) and had oak barrel aromatic notes. Both aspects are in agreement with the image of prestige and quality of vin de garde and

are further supported by studies that identify the ability of wine to improve with age as one of the quality dimensions for wine (Charters and Pettigrew 2007, Lecocq and Visser 2006, Verdú Jover et al. 2004). The link between vin de garde and wine quality level would be worth investigating in more detail. Finally, for Burgundy wines, the a priori allocation of wines in terms of aging potential by providers globally matched the professional judgments by overall evaluation. Whereas all wines from other areas (Bordeaux, Côte du Rhône, and Spain) were categorized by overall evaluation as wines with an aging potential, which is in partial disagreement with our providers. This finding may indicate that the vin de garde concept would be linked to a wine style. A geographic origin such as Burgundy wines indicates a particular style associated with specific grape cultivars, grapegrowing conditions, and winemaking techniques, inducing specific sensory characteristics (Jackson 2002). As a great majority of wines in our sample set were from Burgundy, this may have implicitly induced a “Burgundy reference” for the wine professionals working in Burgundy. Thus, their judgments for wine originating from other areas could have been affected, as they did not judge them in the appropriate framework. This hypothesis, however, would have to be confirmed, as the number of non-Burgundy wines was marginal in this study.

Conclusion The results introduce new information for the understanding of the vin de garde concept. The odor of wine provides clues for professional judgment of aging potential. The roles of astringency and acidity were important perceptual parameters for vin de garde. Visual clues were also important in judging aging potential. Overall, young red wines judged with an aging potential tended to have saturated color, woody and toasted notes, high astringency, and low acidity. Thus, it is the combination of visual, gustatory, mouthfeel, and olfactory clues that allows professionals to judge the aging potential of wines. However, gustatory and mouthfeel clues seem to have more weight than odor in this judgment. Indeed, they are required either to confirm or refute initial judgment based on odor or to make a decision when odor is not reliable enough to judge the aging potential of wines. The vin de garde concept seems quite complex: it involves a wide variety of perceptual clues, appears to be related to wine quality level, and could be linked to wine style.

Literature Cited Araujo, J. 2005. An American flying wine marketer perspective on Europe by Jaime Araujo. Wine Evolution Review Fall:10 Oct 2005. Atanasova, B., T. Thomas-Danguin, C. Chabanet, D. Langlois, S. Nicklaus, and P. Etievant. 2005. Perceptual interactions in odour mixtures: Odour quality in binary mixtures of woody and fruity wine odorants. Chem. Senses 30:209-217. Aznar, M., R. Lopez, J. Cacho, and V. Ferreira. 2003. Prediction of aged red wine aroma properties from aroma chemical composition.

Am. J. Enol. Vitic. 61:1 (2010)

22 – Langlois et al. Partial least squares regression models. J. Agric. Food Chem. 51:2700-2707. Campo, E., J. Ballester, J. Langlois, C. Dacremont, and D. Valentin. 2010. Comparison of conventional descriptive analysis and a citation frequency-based descriptive method for odor profiling: An application to Burgundy Pinot noir wines. Food Qual. Pref. 21:44-55. Campo, E., B.V. Do, V. Ferreira, and D. Valentin. 2008. Aroma properties of young Spanish monovarietal white wines: A study using sorting task, list of terms and frequency of citation. Aust. J. Grape Wine Res. 14:104-115.

gas chromatography-olfactometry data using generalized procrustes analysis applied to six French Chardonnay wines. J. Agric. Food Chem. 51:443-452. Mateus, N., R. Pinto, P. Ruão, and V. de Freitas. 2004. Inf luence of the addition of grape seed procyanidins to port wines in the resulting reactivity with human salivary proteins. Food Chem. 84:195-200. McCloskey, L.P., M. Sylvan, and S.P. Arrhenius. 1996. Descriptive analysis for wine quality experts determining appellations by Chardonnay wine aroma. J. Sens. Stud. 11:49-67.

Charters, S., and S. Pettigrew. 2007. The dimensions of wine quality. Food Qual. Pref. 18:997-1007.

Moio, L., P. Schlich, S. Issanchou, P.X. Etievant, and M. Feuillat. 1993. Description dela typicité aromatique de vins de Bourgogne issus du cépage Chardonnay. J. Int. Sci. Vigne Vin 27:179-189.

Cristovam, E., A. Paterson, and J. R. Piggott. 2000. Differentiation of port wines by appearance using a sensory panel: Comparing free choice and conventional profiling. Eur. Food Res. Technol. 211:65-71.

Monagas, M., P.J. Martín-Álvarez, C. Gómez-Cordovés, and B. Bartolomé. 2007. Effect of the modifier (Graciano vs. Cabernet Sauvignon) on blends of Tempranillo wine during ageing in the bottle. II. Colour and overall appreciation. LWT–Food Sci. Technol. 40:107-115.

Dovaz, M. 2004. 2000 Mots du vin. Les Livrets du Vin. Hachette, Paris.

Morineau, A., L. Lebart, and M. Piron. 1995. Classification mixte et description statistique des classes. In Statistique exploratoire multidimensionnelle, pp. 177-184. Dunod, Paris.

Dupont, J. 2005. Comment la garde vient aux vins. Le point 1721:168197. Garde-Cerdán, T., and C. Ancín-Azpilicueta. 2006. Review of quality factors on wine ageing in oak barrels. Trends Food Sci. Technol. 17:438-447. Gibson, R. 2006. Quels types de vins pour quel consommateur. In Compte-rendu technique Mondiaviti: Le mondial de l’information et de la consommation viti-vinicole. T. Coulon (ed.), pp. 95-98. ITV France & Vinitech, Bordeaux. Guinard, J.X., and M. Cliff. 1987. Descriptive analysis of Pinot noir wines from Carneros, Napa, and Sonoma. Am. J. Enol. Vitic. 38:211-215. Heymann, H., and A.C. Noble. 1987. Descriptive analysis of commercial Cabernet Sauvignon wines from California. Am. J. Enol. Vitic. 38:41-44. Jackson, R.S. 2002. Wine Tasting: A Professional Handbook. Food Science and Technology, International Series. Elsevier Academic Press, San Diego. Jaffré, J., J. Langlois, D. Valentin, C. Dacremont, and D. Peyron. 2007. Vins de Bourgogne: Comment les experts s’expriment sur le potentiel de garde? In Oeno 2007, 8th International Enology Symposium. 2527 June 2007, Bordeaux, France. Jaffré, J., D. Valentin, C. Dacremont, and D. Peyron. 2009. Burgundy red wines: Representation of potential for aging. Food Qual. Pref. 20:505-513. Johnson, H., and F. Robinson. 2002. L’Atlas mondial du vin. Flammarion, Paris. Lawless, H.T. 1999. Descriptive analysis of complex odors: Reality, model or illusion? Food Qual. Pref. 10:325-332. Lecocq, S., and M. Visser. 2006. What determines wine prices: Objective vs. sensory characteristics. J. Wine Econ. 1:42-56. Le Fur, Y., V. Mercurio, L. Moio, J. Blanquet, and J.M. Meunier. 2003. A new approach to examine the relationship between sensory and

Noble, A.C., R.A. Arnold, J. Buechsenstein, E.J. Leach, J.O. Schmidt, and P.M. Stern. 1987. Modification of a standardized system of wine aroma terminology. Am. J. Enol. Vitic. 38:143-146. Noble, A.C., A.A. Williams, and S.P. Langron. 1984. Descriptive analysis and quality ratings of 1976 wines from Bordeaux communes. J. Sci. Food Agric. 35:88-98. Pérez-Magariño, S., and M.L. González-San José. 2006. Polyphenols and colour variability of red wines made from grapes harvested at different ripeness grade. Food Chem. 96:197-208. Perrin, L., R. Symoneaux, I. Maître, C. Asselin, F. Jourjon, and J. Pagès. 2007. Comparison of conventional profiling by a trained tasting panel and free profiling by wine professionals. Am. J. Enol. Vitic. 58:508-517. Peynaud, E., and J. Blouin. 1999. Découvrir le goût du vin. Dunod, Paris. Piombino, P., S. Nicklaus, Y. Le Fur, L. Moio, and J.L. Le Quéré. 2004. Selection of products presenting given f lavor characteristics: An application to wine. Am. J. Enol. Vitic. 55:27-34. Ribéreau-Gayon, P. 2000. Atlas Hachette des vins de France. Hachette, Paris. Spurr, R. 2006. Le marché, ses modes de consommation et la attentes des consommateurs. In Compte-rendu technique Mondiaviti: Le mondial de l’information et de la consommation viti-vinicole. T. Coulon (ed.), pp. 83-84. ITV France & Vinitech, Bordeaux. Verdú Jover, A.J., F.J. Lloréns Montes, and M.d.M. Fuentes Fuentes. 2004. Measuring perceptions of quality in food products: The case of red wine. Food Qual. Pref. 15:453-469. Vidal, S., L. Francis, A. Noble, M. Kwiatkowski, V. Cheynier, and E. Waters. 2004. Taste and mouth-feel properties of different types of tannin-like polyphenolic compounds and anthocyanins in wine. Anal. Chim. Acta 513:57-65.

Am. J. Enol. Vitic. 61:1 (2010)