Table

titre du tableau
Variable	Level	%	Sum (liking)	Mean (liking)	Mean drop	Penalty	P-value
Colour intensity	Not enough	36.6 %	228	4.65	0.22
	JAR	59.7 %	390	4.88		0.25	0.385
	Too much	3.7 %	22	4.40	0.48
Peppery notes	Not enough	23.1 %	149	4.81	0.54
	JAR	29.9 %	214	5.35		0.82	0.006
	Too much	47.0 %	277	4.40	0.95
Fruity notes	Not enough	74.6 %	454	4.54	1.16
	JAR	20.1 %	154	5.70		1.16	0.001
	Too much	5.2 %	32	4.57	1.13
Sourness	Not enough	6.0 %	31	3.88	1.50
	JAR	33.6 %	242	5.38		0.91	0.002
	Too much	60.4 %	367	4.53	0.85
Alcohol	Not enough	20.1 %	114	4.22	0.74
	JAR	66.4 %	442	4.97		0.57	0.052
	Too much	13.4 %	84	4.67	0.30
Astringency	Not enough	31.3 %	196	4.67	0.35
	JAR	48.5 %	326	5.02		0.46	0.092
	Too much	20.1 %	118	4.37	0.65
Bitterness	Not enough	22.4 %	143	4.77	0.31
	JAR	37.3 %	254	5.08		0.48	0.089
	Too much	40.3 %	243	4.50	0.58

These results highlighted that the greatest significant penalty (P = 0.001) with a value of 1.16 was due to the intensity of fruity notes, which were perceived as insufficient by almost 75 % of the young panellists. It can be mentioned that Spearman's correlation coefficient (rs) of 0.325 with a P < 0.05 was observed between the score obtained for this JAR variable and the liking score. If studies investigating the appreciation of fruity notes by young consumers are scarce notably in red wine, these findings are somehow consistent with previous research conducted in France, Canada, New Zealand and Spain showing that young consumers particularly enjoyed fruity white wines (Beaulieu et al., 2022; Lesschaeve et al., 2012; Lund et al., 2009), considered flavour and aroma as the most important aspects in red wines, with floral aromas eliciting positive emotions and liking (Mora et al., 2021).

The second strongest penalty concerned the sourness attribute, which was perceived as being too much for 60 % of the participants. This observation is consistent with previous work emphasising that acidic wines are less well accepted by newer wine consumers and, therefore, in most cases, by young consumers with less experience (Francis and Williamson, 2015).

The third and last significant penalty was related to the intensity of peppery notes. Surprisingly, no significant relationship was identified (P = 0.206) between the JAR scores for fruity and peppery notes (rs = –0.134). Indeed, rotundone has been previously described as an impact aroma compound that can break the aroma buffer caused by ethanol and other major volatiles (Ferreira, 2012). Consequently, a stronger and more significant negative correlation would have been expected through a perceptual interaction mechanism (Atanasova et al., 2004), with rotundone and peppery notes exhibiting a masking effect on the fruity character. Given the significant rotundone concentration in the studied wine, the proportion of subjects who perceived such character as too much did not exceed 50 %, which suggests that such aromatic components can be positive for some young consumers.

It cannot be excluded that some anosmic participants to rotundone, who were not able to smell peppery notes in the studied wine and were likely to perceive it as fruity as previously demonstrated (Geffroy et al., 2020b), induced a bias in the dataset. Indeed, such subjects might have contributed to underestimating the percentage of consumers who rated the peppery notes as too much while overestimating the percentages of JAR or/and too much for the fruity notes. However, such impact that would deserve to be investigated through the determination of individual ODT for each consumer is likely to be minor as the percentage of anosmic respondents to rotundone among young consumers was estimated in a previous study at 10 % (Geffroy et al., 2023), a much lower rate in comparison with the general population.

Although astringency has been previously described as a negative driver of acceptance by inexperienced wine consumers (Francis and Williamson, 2015), the fact that none of the other studied JAR variables induced a significant penalty despite P-values close to the α = 0.05 threshold (P = 0.092 for astringency, P = 0.089 for bitterness) might be related to the wine characteristics that did not show any pronounced bitterness or astringency, some features not underlined by the expert panel. Although close to the significance threshold, the ethanol content was not identified as having a significant negative impact, which is consistent with recent research highlighting that ethanol concentration does not necessarily drive consumer preferences (Geffroy et al., 2022). Given the importance of colour in overall wine liking (Dooley et al., 2012), surprisingly, this characteristic was not identified as a significant penalty. The absence of familiarisation towards this JAR variable during the preliminary session and the relatively low amount of wine provided for the test (40 mL) might have played a role.

More generally, the fact that consumers were familiarised with most of the JAR variables prior to the experiment, using solutions with low concentrations close to the recognition threshold of the given attributes for trigeminal and taste sensations, might have contributed to these being over-perceived in wine, and to rating the product based on cognitive processing. Nevertheless, this hypothesis is unlikely as some variables for which familiarisation was undertaken, such as astringency, are still considered insufficient by many consumers. This latter attribute could also be explained by the total phenolic compound of the studied wine, namely 32 ± 5, which was lower in comparison with other Duras wines produced during warmer and dryer seasons that exhibited values up to 74 ± 12 (Geffroy et al., 2016b).

It can be pointed out that the panel used for this study was a convenience sample composed of French students in agriculture sciences with similar social and cultural backgrounds. It cannot be excluded that other results might have been obtained with French consumers exhibiting other and more diversified characteristics.

It should also be noted that these results may not apply to consumers from other French wine regions, particularly the places where young consumers are less familiar with wines containing great concentrations of rotundone. Indeed, the panel used for this study is likely to be familiar with rotundone, which has been reported apart from Duras, in other cultivars grown in the Toulouse area, such as Fer, Malbec, Duras or Prunelard (Geffroy et al., 2020a). As familiarity is known to positively affect consumers' appreciation of a given aroma (Saliba et al., 2009), it is possible that lower liking scores and other penalty values would have been observed in other French cities.

The fact that a significant proportion of respondents (9.7 %) were infrequent consumers, drinking red wine less than once a month might have induced a bias, notably by lowering the panel liking score. However, when performing another data treatment after removing the 13 panellists concerned, no changes were observed in the panel liking score (4.70 ± 1.59), the significance and amplitude of penalties applied to the studied product (lack of fruity notes, 1.25; excessive sourness, 0.97 and excessive peppery notes, 0.81).

Although the studied wine was selected as being representative of a Duras wine produced during a cool and wet season, these results are only valid for a specific wine produced by a single estate. It cannot be excluded that other outcomes might have been obtained using another sample. However, it should be pointed out that though Vitis vinifera L. cv. Duras represents 800 ha under vines; the number of 100 % pure varietal commercial references available for this variety is still limited during a normal season, even if in strong increase, as it is mainly used for blending or less frequently to make rosé wines. In 2021, which was characterised by a low crop as a consequence of severe spring frost events, most of the winegrowers producing such pure varietal wines decided to blend it with at least 20 % of other varieties (e.g., Fer or Syrah), reducing drastically the number of samples available for this research.

Although penalty analysis can be conducted with one single product, and the present study illustrates this feasibility, it could have been relevant to include other wines and, notably, a control sample in the experimental design. This control could have been the same 2021 Duras red wine with similar taste and mouthfeel characteristics but without rotundone. This sample could have been obtained by using the procedure previously described by Geffroy et al. (2020b). In this work, wines were dearomatised using supercritical CO₂ before performing aroma reconstitution by mixing aroma compounds with Odour Activity Value (OAV) above five. The test sample would have been the full reconstitution, and the control sample the reconstitution without rotundone.

The results can also be useful for Fer, Malbec, Tardif, and Prunelard producers, some cultivars in which rotundone has been previously identified (Geffroy et al., 2020a), even though the mouthfeel and aroma characteristics of the wines made from these latter varieties might differ in comparison with Duras. Indeed, they are likely to exhibit a more marked astringency, which could become a strong penalty for such a target consumer segment (Francis and Williamson, 2015).