1. Are there sensory signatures associated with specific PDOs?

To address this question, we performed a literature review using four scientific databases: Elsevier, Taylor and Francis, Wiley, and Springer and two scientific wine journals: OENOOne and American Journal of Enology and Viticulture, between the years 2000 and 2022. The keywords “wine PDO”, “origin”, and “typicity" were used. The titles and abstracts of the articles retrieved during the initial search were reviewed to keep only articles presenting sensory studies. A total of 30 studies were found with this criterion. Then, a second selection was applied to focus on studies investigating sensory differences between PDO wines with the same grape variety within the same country. Nine studies met these selection criteria (Table 1). These studies were classified into three groups based on the type of comparisons carried out. The first two groups involved comparisons of PDOs between vineyards either from the old (France and Spain) or the new (Australia) wine world. The third group included comparisons within a given vineyard, the Loire Valley in France.

2. Can we rely on experts’ mental representations to evaluate PDOs sensory signatures?

Urdapilleta and Manetta (2011) defined mental representations as “knowledge or beliefs that are well stabilised in the subject's memory, but which may change as a result of experience or instruction (p.71)”. Two studies relied on this idea to evaluate whether wine experts have a consensual representation of the sensory properties associated with PDOs. Jose-Coutinho et al. (2015) explored experts' mental representations of 12 Portuguese Protected Geographic Indications (PGI). Twenty wine experts (winemakers, wine-science researchers, Masters of Wine, wine judges, wine writers and wine retailers) received the following instruction: “How would you define a typical young commercial white wine from this particular PGI and score each sensory attribute accordingly?” The sensory scales going for 0 (no trace) to 10 (extremely intense) included colour, aroma, and taste attributes. Results showed that only the PGI Minho had a specific sensory signature anchored in the mind of experts. Other PGIs were clustered in two groups: a Southern group, composed of four PGIs, and a Central group, composed of seven PGIs. More recently, Leriche et al. (2020) used a focus group methodology to explore wine experts' representation of the sensory characteristics of six PDO wines from the Languedoc vineyard (France). For each PDO, experts (i.e., winemakers, producers, oenologists, merchants, wine merchants, sommeliers, and staff of unions) started by generating individually a list of sensory descriptors that they associated with the PDO. Then, a debate was engaged to obtain a consensual list of descriptors. Results highlighted that only a few descriptors were specific to one or two PDOs (e.g., woody, persistence, density, and floral), and several descriptors were common to all the PDOs studied (e.g., scrubland, tannins, spices, red fruits, black fruits, and ripe fruits).

Results of these two studies showed that contrary to what was previously observed for grape variety (Hughson and Boakes, 2002; Solomon, 1997), experts did not abstract clear sensory concepts from PDO wine tastings. Only some PDOs gave rise to clearly defined concepts associated with specific sensory attributes (e.g., PGI Minho, Jose-Coutinho et al., 2015). The majority of PDOs from the same wine region seem to be associated with overlapping sensory concepts with many common sensory attributes. These results could be due, in part, to a lack of agreement between experts on the specific sensory characteristics of PDO wines, which prevents the creation of a shared concept specific to each PDO. This lack of agreement may be because, in these studies, wine experts are considered as a unified group independent of the nature of their expertise. In a recent study, Otheguy et al. (2021) showed that three groups of experts (Makers, Sellers, and Critics) constructed different mental representations of Beaujolais wines based on their specific experiences and exposures to wines from this region.

3. PDOs hierarchy: an evolutive system

PDOs are social constructions and, as such, constitute a dynamic and evolutive system (Hinnewinkel, 2004). PDO evolutions are materialised both by the creation of new PDOs and by an evolution in the hierarchy and quality of the wines within a given appellation (Legouy and Dallot, 2019). In France, this can lead to the birth of Complementary Geographical Denominations (Dénomination Géographique Complémentaire, DGC). This complementary mention corresponds to a well-defined sub-space of an already existing PDO and is affixed after the name of the PDO. For example, the Beaujolais vineyard was originally structured into three main categories: Beaujolais, Beaujolais-Villages and Cru Beaujolais is undergoing some restructuration. In 2017, a DGC called “Pierres Dorées” was developed within the Beaujolais PDO, resulting in the “Beaujolais Pierres Dorées” PDO. With the apparition of this new DGC, the Beaujolais PDO is separated into three subcategories: Beaujolais, Beaujolais Nouveau and Beaujolais Pierres Dorées. The development of DGCs generally results in a move upmarket with higher quality and marketing differentiation from the general PDO. The new DGC Pierres Dorées is being developed to obtain recognition for the wines produced in this region. The objective is to achieve a move upmarket within the Beaujolais appellation. But does that imply the emergence of specific sensory profiles?

4. Objectives of the study

The main objective of the present study was to understand if the evolution of the PDO system was paralleled by an evolution in the wine sensory profiles. To meet this objective, we used the Beaujolais vineyard as a case study. Our strategy was to compare the sensory profiles of wines from the new Beaujolais Pierres Dorées PDO with more traditional PDOs from the Beaujolais vineyard. Based on previous work, we used two sensory approaches: one based on experts' mental representations (study 1) and the other on actual tasting (study 2). For the tasting, we used both blind and informed tasting with the idea of accessing and studying the mental representations of the experts in an indirect way. Our rationale was that the changes in wine descriptions between the two conditions provide insightful information on the contents of the sensory concepts developed by the experts. Because previous work showed that wine experts might not be a unified group (Otheguy et al., 2021), a second objective of the present study was to understand to what extent differences in experts' training and experience influence their representations and evaluations of the sensory characteristics of PDO wines. Three groups of experts participated in the study: Wine makers from the Beaujolais vineyard, wine sellers and wine critics.

5. Hypotheses

The hypotheses of the first study were:

H1) through their experience with wines labelled with the mention Beaujolais Pierres Dorées, wine experts have developed a specific representation of these wines.

H2) winemakers in the Beaujolais region, more familiar with the new Beaujolais Pierres Dorées PDO, may have a clearer picture of these wines than wine sellers or wine critics.

The hypotheses of the second study were:

H3) the development of the new Beaujolais Pierres Dorées PDO, resulting from the evolution of the vineyard, leads to specific sensory characteristics of the wines, different from the Beaujolais PDO wines already in place.

H4) the experience of experts with PDO Beaujolais Pierres Dorées wines could lead to expectation effects. Thus, we expected to observe greater differences between Beaujolais and Beaujolais Pierres Dorées PDO wines during the informed tasting compared to the blind tasting.

H5) winemakers in the Beaujolais region, more familiar with the new Beaujolais Pierres Dorées PDO, could better discriminate between Beaujolais and Beaujolais Pierres Dorées PDO wines during tasting than wine sellers and wine critics.

1. Participants

Fifty-two wine experts (38 men and 14 women) with an average age of 49 years old were recruited. Three panels were formed based on the experts’ main professional activity. The first panel, named Makers, was composed of 19 individuals who make, supervise, or advise on all the stages and oenological practices in the Beaujolais vineyard (e.g., winegrowers, cellar masters, oenologists, and vine advisors). The second panel, named Sellers, was composed of 15 individuals who participated in the direct trade of wine bottles with consumers in stores or restaurants in the Beaujolais region (e.g., sommeliers and wine merchants). The third panel, named Critics, was composed of 18 individuals who have written articles, published books, or blogs about Beaujolais wines (e.g., wine critics and journalists). They were preferably recruited at the regional level and then at the national and international levels, checking in advance that they were all French-speaking and had knowledge about the Beaujolais wines. A questionnaire was sent a posteriori to the participants to obtain information on their training and experiences with Beaujolais PDO and Beaujolais Pierres Dorées PDO wines. Only 83 % of the participants completed it. Experts’ characteristics are reported in black in Table 2.

2. Procedure

Imaginary sensory profiles of five PDOs from the Beaujolais vineyard were obtained using the Rate-All-That-Apply (RATA) method. RATA is a variant of the CATA (Check-All-That-Apply) method (Ares et al., 2014). With RATA, wine experts had to select pre-determined descriptors and rate their intensity to describe the wines. The list of descriptors was obtained based on the descriptions of the red wines of the Beaujolais PDO and Beaujolais Pierres Dorées PDO present in the last ten editions of the Hachette wine guide. A list of terms and their frequency of occurrence was first compiled. Next, two independent researchers reduced the list by regrouping terms into categories. For example, descriptors related to the floral aromas of violet and peony were categorised under the floral descriptor. A total of 20 descriptors covering odour/aroma (blackcurrant, cherry, concentrated, expressive, floral, raspberry, spicy, strawberry, vegetal, and woody), taste (acid, bitter, and sour), and mouthfeel (astringent, balanced, complex, fresh, long, round, and supple) sensations was selected. A pre-test was carried out to judge the relevance of the descriptors chosen to obtain the final questionnaire. The RATA questionnaire was created with the TopDegust software (IFV, France) and sent by email to each expert.

The instructions for the online questionnaire were the following:

"Imagine yourself tasting a red wine of the indicated appellation. I am going to ask you to describe this wine by checking off only the descriptors that, for you, best describe it. For each descriptor checked, I ask you to rate its intensity on a scale from 1 ("not very intense") to 5 ("very intense"). If you wish, you can also add other descriptors by checking the "Other descriptors" box. "

Four PDOs were included in the questionnaire: Beaujolais PDO, Beaujolais-Villages PDO, Morgon PDO and Moulin-à-Vent PDO, in addition to the Beaujolais Pierres Dorées PDO. As this study focused mainly on the Beaujolais Pierres Dorées PDO, the following presentation order was used for all experts: Beaujolais Pierres Dorées PDO, Beaujolais PDO, Beaujolais-Villages PDO, Morgon PDO and Moulin-à-Vent PDO.

3. Data analysis

Only descriptors with a frequency of more than 20 % of the panels of the three experts were kept. All descriptors, except for acid and bitter, met this criterion. The descriptors added by the wine experts to the RATA list were not considered because they were often used by only one wine expert. Next, the RATA data were analysed using analysis of variance (α = 5 %). A 3-way mixed design ANOVA with judge nested in the type of expert as a random factor and type of expert and PDO as fixed factors (J < type of expert > *PDO) was carried out using the PROC GLM procedure of SAS 9.4 (SAS Institute, Cary NC). As the data were unbalanced, the type III Sum of square error was considered. Newman–Keuls pairwise comparisons were carried out when a significant effect was observed. The ANOVA was carried out on two dependent variables: frequency of use of descriptors and intensity score of descriptors.

1. Frequency of use of descriptors

Results showed a significant effect of PDO in the use of descriptors (F (4,196) = 26.91, p < 0.0001). The Newman–Keuls pairwise comparisons showed that globally more descriptors were checked to describe the two “Crus Beaujolais” PDOs, Moulin-à-Vent PDO (6.8 descriptors on average) and Morgon PDO (6.4 descriptors on average) than the Beaujolais PDO (4.8 descriptors on average) and Beaujolais Pierres Dorées PDO (4.6 descriptors on average). The number of descriptors checked to describe Beaujolais-Villages PDO was not significantly different from other PDOs, with an average of 5.7 descriptors. A main effect of the type of experts was also observed (F (2,196) = 8.17, p < 0.001). The Newman–Keuls pairwise comparisons showed that the Makers used significantly more descriptors (7.4 on average) than the other two types of experts: the Critics used 5.8 descriptors and the Sellers 3.8.

2. Intensity scores of descriptors

The ANOVA showed a significant main effect of PDO for 12 descriptors (Figure 1) and a significant main effect of type of experts for eight descriptors (Figure 2). A type of expert*PDO significant interaction was observed for four descriptors (Figure 3). The detailed results are presented in additional material (Additional table 1).

2.1 Main effect of PDO

Pairwise comparisons showed a separation between the regional PDOs (Beaujolais PDO, Beaujolais Pierres Dorées PDO and Beaujolais-Villages PDO) and the “Crus Beaujolais” PDOs (Morgon PDO and Moulin-à-Vent PDO). The regional PDOs were described as sourer (p < 0.0001), fresher (p < 0.0001), with more strawberry (p < 0.0001) and raspberry (p < 0.0001) aromas and were suppler (p < 0.0001) than the “Crus Beaujolais” PDOs.

On the contrary, the “Cru Beaujolais” PDOs were seen as more astringent (p < 0.0001), woodier (p < 0.0001), more complex (p < 0.0001), more concentrated (p < 0.0001), spicier (p < 0.0001) and longer (p < 0.0001) than the regional PDOs.

The separation between regional PDOs and “Crus Beaujolais” PDOs was not clearly identified for the descriptor blackcurrant. Results showed that the intensity of this descriptor was lower for the Beaujolais Pierres Dorées PDO than the Beaujolais-Villages PDO and the two “Crus Beaujolais” PDOs (p < 0.005).

Globally, the sensory profile of Beaujolais Pierres Dorées PDO was closer to the profile of the Beaujolais PDO than that of the crus PDOs. The pairwise comparison highlighted only a significant difference between Beaujolais Pierres Dorées and Beaujolais PDO for the descriptor: strawberry (Figure 1): the Beaujolais PDO wines received higher intensity scores (mean = 1.48) than the Beaujolais Pierres Dorées PDO ones (mean = 1.15).

Figure 1. Results of the ANOVA for the 12 significant descriptors between all PDOs.

Save View full size Expand inline Collapse inline

2.2 Main effect of the type of experts and interaction between the type of experts and PDO

Figure 2 shows the main effect of the type of expert for the eight significant descriptors. Except for the descriptor astringent, pairwise comparisons showed lower intensity scores for the Sellers than for the two other groups of experts. The Makers gave higher intensity scores for astringency, complex, concentrated, and long than the Critics.

Figure 2. Results of the ANOVA for the eight significant descriptors between the three types of experts.

Save View full size Expand inline Collapse inline

For the other descriptors, no difference was observed between Makers and Critics. This pattern of results, however, is modulated by the presence of significant interactions (Figure 3). The differences observed between types of experts for the descriptors spicy, astringent, and long hold only for the two “Crus Beaujolais.” A significant interaction between the type of expert and PDO was also observed for the descriptor woody, which was higher for the Makers than for other experts but only for the Moulin-à-Vent Cru.

Figure 3. Results of the interaction type of experts*PDO.

Save View full size Expand inline Collapse inline

1. Participants

Seventy-three percent of the panel of study 1 have accepted to participate in study 2 (29 men and 9 women) with an average age of 51 years old. Among the 38 participants, 15 were wine makers, 12 were wine sellers, and 11 were wine writers. Experts’ characteristics are reported in red in Table 2.

2. Wines

Two different red PDOs were used for this study: Beaujolais PDO and Beaujolais Pierres Dorées PDO. For each PDO, five wines were selected. As mentioned in the introduction, the selection of wines is a crucial step in PDO studies. To optimise the generalisability of our results, we used a randomisation strategy. First, a total of 10 commercial wines in each PDO were drawn at random from the harvest declaration for the 2018 vintage. A tasting selection was conducted to judge the wines and to eliminate those with potential defects (e.g., oxidation, reduction, spoilage micro-organisms). A group of seven wine experts, composed of oenologists and wine technicians, from the Institut Français de la Vigne et du Vin (IFV) and SICAREX Beaujolais team in Villefranche-sur-Saône was recruited. During this tasting, four wines were eliminated from the selection.

Next, a second random draw was carried out on the remaining wines to constitute two sets of five wines: Set A composed of five Beaujolais PDO wines and Set B composed of five Beaujolais Pierres Dorées PDO wines. Then, each bottle was repackaged in an inert, oxygen-free atmosphere into transparent 5cL Vinottes® samples to avoid wasting wine and to facilitate the procedure.

3. Procedure

This study was conducted individually, with each expert supervised by the same experimenter. All wine samples were tasted by an oenologist to ensure that the reconditioning did not cause any defects before each tasting session. Each expert participated in tastings following two conditions: first, a blind tasting without any information about the wines, and second an informed tasting with information about the PDO of each wine (Beaujolais PDO or Beaujolais Pierres Dorées PDO). The same RATA sensory questionnaire as for the imaginary study was used for the two tastings.

The wine samples were coded with a three-digit code, and each bottle was covered with a black sock. The wines were served at ambient temperature in black ISO glasses to focus only on aromatic and gustative descriptors. For each tasting, the ten wine samples from the two sets were randomly presented according to a Williams Latin square.

4. Data analysis

The same two dependent variables as in study 1 were used. All descriptors were kept as their frequency was higher than the 20 % threshold. As in the first study, the descriptors added by the wine experts to the RATA list were not considered as a single wine expert often used them. Next, the RATA data were analysed using analysis of variance (α = 5 %). A 4-way mixed design ANOVA with judge nested in the type of expert and wine nested in PDO as random factors and type of expert, PDO and condition as fixed within-subject factors (J < type of expert >*Wine < PDO >*Condition) was conducted using the PROC MIXED procedure of SAS 9.4 (SAS Institute, Cary NC)). Newman–Keuls pairwise comparisons were conducted when significant effects were observed (alpha = 5 %). A Multiple Factor Analysis (MFA) was then carried out using the R statistical software version 4.0.3 for Windows (R Core Team, 2020) with the FactoMineR package (Lê et al., 2008) to illustrate the influence of the information on wine tasting.

1. Frequency of use of descriptors

Results showed no significant difference in the use of descriptors between the two PDOs (5.59 descriptors on average for the Beaujolais PDO and 5.81 descriptors on average for the Beaujolais Pierres Dorées PDO). No significant effect of the type of experts was observed. The Makers used 6.10 descriptors on average, the Sellers 5.37, and the Critics 5.64. No significant difference in the use of descriptors was observed between the blind tasting condition (5.72 descriptors on average) and the informed tasting condition (5.68 descriptors on average).

2. Intensity scores of descriptors

Significant main effects were observed at the 5 % level for the PDO and the condition factors. These main effects are modulated by several significant interactions (PDO*condition, type of expert*condition and type of expert*condition*PDO). No significant main effect of expert type and no significant PDO*expert type interaction was observed. The detailed ANOVA results are presented in additional material (Additional table 2).

With regards to the PDO factor, the Beaujolais PDO was perceived as significantly less bitter and less spicy than the Beaujolais Pierres Dorées PDO. Concerning the condition effect, a significant effect was observed for the descriptor bitter with a higher intensity in the blind tasting than in the informed one. The effect of PDOs was modulated by the effect of the condition for two descriptors: complex and vegetal (Figure 4). In the informed tasting condition, the Beaujolais Pierres Dorées PDO was rated as more complex than the Beaujolais PDO. In the blind tasting condition, the Beaujolais Pierres Dorées PDO was more vegetal, while in the informed tasting, it was less vegetal than the Beaujolais PDO.

A significant type of expert*condition interaction was observed for the descriptor fresh (Figure 5). In the blind tasting condition, the Critics rated the wines fresher than the Makers, but in the informed tasting, no significant difference appeared between the three types of experts.

Figure 4. Results of the interaction PDO*Condition.

Save View full size Expand inline Collapse inline

Figure 5. Results of the interaction type of experts*Condition for the descriptor Fresh.

Save View full size Expand inline Collapse inline

Finally, a type of expert*condition*PDO interaction was significant for the descriptors spicy and vegetal (Figure 6). The Critics judged the Beaujolais Pierres Dorées PDO spicier than the Beaujolais PDO in the blind tasting but not in the informed tasting. In addition, the Critics rated the Beaujolais Pierres Dorées spicier than the Sellers did in the blind condition. However, this difference disappeared in the informed tasting. In the blind tasting condition, the Sellers rated the Beaujolais PDO wines as less vegetal than the Critics, but in the informed tasting, this difference was found for the Beaujolais Pierres Dorées PDO.

Figure 6. Results of the interaction type of experts*condition*PDO.

Save View full size Expand inline Collapse inline

3. Multiple Factor Analysis of RATA data

Although the first two dimensions explain only 25 % of the total variance, they provide interesting information regarding the variability of the sensory descriptions of the wine samples within each PDO. The remaining dimensions highlighted the same phenomenon but are not shown here. Figure 7 shows that globally the three groups of experts were consensual on the first dimension but less so on the second dimension, with a greater difference between the Sellers and the Critics.

Figure 7. Projection of the three types of experts on the first two dimensions of the MFA.

Save View full size Expand inline Collapse inline

Figure 8 highlighted that the first dimension (14.22 % of the variance) opposed the less qualitative descriptor vegetal to the more qualitative descriptors complex, concentrated, expressive, and long. The second dimension (10.98 % of the variance) opposed the descriptors floral, fresh, and round to the descriptors acid, bitter, and woody. In the blind tasting condition, a large variability can be observed among the wines from the same PDO. Furthermore, the results showed that the two PDOs overlap. Providing information on the wine PDO decreased the difference between wines within each PDO and, thus, allowed a better separation between the two PDOs along the first dimension, with the Beaujolais Pierres Dorées PDO wines described as more complex, concentrated, expressive, and long, and the Beaujolais PDO wines described as more vegetal.

Figure 8. Results of the MFA: a) the blind tasting, b) the informed tasting and c) the correlation circle. In the individual factor map, Beaujolais PDO wines were represented by dotted purple lines and Beaujolais Pierres Dorées PDO wines by solid orange lines. The descriptors were added following the correlation circle analysis.

Save View full size Expand inline Collapse inline