A total of 43 white wines (vintages 2018–2023) and 33 red wines (vintages 2020–2023) underwent sensory characterisation. Samples from FRG (single-varietal or blended) and Vitis vinifera (single-varietal) were evaluated for both types of wine. The study included both experimental and commercial wines mainly from the Occitanie and Nouvelle-Aquitaine regions in the south of France. (Tables 1 and 2). Using experimental wines ensured 100 % varietal purity, whereas commercial wines, despite varietal labelling, may contain up to 15 % other varieties. The detailed composition of commercial wines was confirmed by the winery when possible. Wines were classified into three geographical-climatic areas, namely the Mediterranean (Type 8), the Southwest basin (Type 7) and the Altered oceanic (Type 4) areas, according to the classification provided by Joly et al. (2010). The altered oceanic climate is mild with moderate rainfall, the southwest basin climate is warmer with rare frosts, and the Mediterranean climate has hot, dry summers and wet winters.

The Vitis vinifera grape varieties were selected for their prevalence in the study region and because they are among the most widely planted varieties in 2022, used to produce white and red wines within the Protected Geographical Indication (PGI) Pays d’Oc, which is the main designation for varietal wines in southern France (https://www.paysdoc-wines.com). The reference wines were selected to represent the commercial offering of each variety in the study region. They were produced by major wine companies that sell several million bottles of these wines each year.

This study was conducted over two years to capture the full diversity of profiles, as seasonality and terroir are the main factors influencing the composition of wine, particularly with regard to aroma compounds and sensory characteristics (van Leeuwen et al., 2020).

The FRG wines used in this study were those produced in southern France identified in an earlier, comprehensive national study of the commercial FRG wine market in France (Paire et al., 2025b). In addition to geographical origin, the selection was guided by two main criteria: maximising sample diversity and including emerging varieties, such as ‘Floreal’, ‘Artaban’, and ‘Vidoc’, whose cultivated area is beginning to expand in southern France (Dolet et al., 2024). The selected wines were made from FRG varieties bred in several European countries such including France (e.g., ‘Floreal’, ‘Voltis’, ‘Artaban’, ‘Vidoc’, Bouquet G14, Bouquet 3176), Germany (‘Souvignier Gris’, ‘Muscaris’, ‘Cabernet blanc’, ‘Prior’, ‘Monarch’, ‘Cabernet Cortis’), Switzerland (‘Sauvignac’, ‘Pinotin’, ‘Cabernet noir’, ‘Cabernet Jura’), and Italy (‘Soreli’, ‘Sauvignon Kretos’, ‘Cabernet Volos’, ‘Merlot Khorus’).

The oenological parameters of the wines were measured just before the sensory analysis was performed, in accordance with the reference methods of the International Organisation of Vine and Wine (OIV). The analysis was carried out once without replication (Tables S1 and S2).

Sensory analysis was conducted at INRAE UMR SPO in Montpellier, France, over two consecutive years: 2023 (year 1) and 2024 (year 2). The panel comprised 19 panellists in year 1 and 17 in year 2. Panel members were selected based on their sensory performance and interest, and then underwent training in the descriptive sensory analysis of wines (norm ISO 8586-2, 2008).

The wines were analysed using the Check All That Apply (CATA) methodology (Meyners et al., 2013) and, more specifically, during the olfactory assessment with hierarchical CATA (Koening et al., 2023). The CATA principle is that panellists are given a list of predefined descriptors based on previous research and reviewed by project team members. They then selected all the descriptors that describe the sample to be tested. The set of descriptors used in this study is highly consistent with those identified by Garcia et al. 2022 in their previous research.

As for hierarchical CATA, this method allows for a more “natural” structure of attributes and reduces cognitive effort. It is a descriptive analysis method that enables the evaluation of a large number of samples. Prior to all tasting sessions each year, judges are trained to understand and consistently use the attributes, as well as to familiarise themself with the methodology. Standards were created by adding compounds to white or red wine to help the judges identify and remember the sensory attributes. A detailed list of the sensory attributes and related standards can be found in Table 3.

Wines were analysed over several sessions each year, with an average of 11 ± 1 white wines and 8 red wines tasted per session. The analyses were conducted in individual testing booths. Samples (30 ml) were served at room temperature (18 ± 1 °C) in black glasses to avoid potential visual influences during evaluation. The glasses were covered with a plastic cap and marked with random three-digit codes. The wines were presented in a monadic service according to William’s Latin square design to balance the order of presentation and the carry-over effect (Macfie et al., 1989). Fifty per cent of the wines were analysed in duplicate. For each sample, judges had to select the most relevant attributes from lists of 30 olfactory terms (classified hierarchically into 12 families) and 9 gustatory terms for white wines, and 29 olfactory terms (classified into 13 families) and 10 gustatory terms for red wines (Table 3).

Data acquisition was assisted by FIZZ software (FIZZ network, v.2.518; Biosystème, Courtenon, France).

HCATA data were collected in a product-by-attribute table for each subject, with scores of 0 (absent) or 1 (present).

To incorporate the hierarchical structure, we first aggregated the selected descriptors, carrying over those chosen at the lowest level to the higher levels (descriptor, then family), as described by Koenig et al. (2023): if a judge selected a descriptor, the corresponding family was assigned a checkmark. When multiple terms were evaluated within the same family, that family was assigned a single checkmark.

The obtained binary data were analysed using XLSTAT 2022 software (Lumivero, Paris, France). Before processing the results, we ensured the panel’s performance by analysing the average reproducibility index (Ri), which was calculated to assess each judge’s individual performance.

$R_i=\frac{1}{n}\times \sum 2\times \frac{{Des}_{rep}}{{Des}_{rep1}+{Des}_{Rep2}}$(Equation 1)

Des_rep is the number of terms used by the judge for each replicate.

Des_rep1 and Des_rep2 are the total numbers of terms used by the judge for the first and second replicates, respectively.

n is the number of samples duplicated.

This parameter, which ranges from 0 to 1, was used in previous studies (Campo et al., 2008).

To identify significant differences between wines for each attribute, a Cochran’s Q test was performed. Descriptors that were significant at the 5 % level in at least one of the two evaluation years were selected.

Multiple Factor Analysis (MFA) was performed on the contingency table by separating the significant olfactory and gustatory descriptors into two tables and adding the year of analysis as a supplementary variable to investigate its effect on the sensory results.

In this study, the reproducibility index (Ri) had a maximum value of 0.90, indicating 90 % consistency in common terms between two replicates for a given judge, and a minimum value of 0.42. The median Ri values were 0.69 and 0.68 for white wines in years 1 and 2, respectively, and 0.70 and 0.65 for red wines in the same periods. These results can be considered robust, as they are superior to those reported in the reference study (Campo et al., 2008). Overall, the jury’s performance exhibited notable homogeneity, with slightly higher reproducibility observed in year 1 and for red wines (Table S3).

Cochran’s Q tests revealed significant differences (p < 0.05) for 15–21 olfactory attributes in white wines and 9–14 in red wines (Table S4), depending on the year, representing 25 % to 50 % of the descriptors. Notably, 11 olfactory descriptors consistently distinguished white wines across both years, compared to just 6 for red wines. In contrast, gustatory attributes were more variable, with 60 % to 100 % of descriptors showing significant differences between wines.

The sensory analysis revealed a greater number of significant differences in the gustatory profile than in the olfactory profile. This is consistent with practice, since the gustatory dimension typically involves fewer descriptors. In contrast, olfactory evaluation is often subject to descriptor overlap whereby multiple attributes describe a similar sensation. However, specific panel training helps to reduce this effect, which is consistent with the findings of Seinforin et al. (2023). Their analysis of recognition rates indeed demonstrated higher overall performance for olfaction than for taste, supporting the conclusion that the sense of smell is more effective than taste for the identification of grape varieties. Furthermore, more significant differences were identified among white wines than red wines. This may be due to the larger sample size of the white wine set, which potentially introduced greater sensory diversity. The most significant descriptors enabled clear characterisation of the wines using common terminology such as fruity, floral, chemical, and defects such as reduced and animal. Interestingly, the mineral descriptor reached statistical significance, despite the poorly defined nature of this attribute (Parr et al., 2018). The sensory descriptors used to describe wines show a strong alignment with the terminology previously established by Koenig et al. (2023).

In both MFA analyses, the “year of analysis” was included in a supplementary table. The results revealed no significant effect of the year of analysis, as evidenced by the very low RV coefficients (0.12 for white wines and 0.06 for red wines). This confirms that the wines’ sensory profiles are consistent across the two years of analysis, enabling reliable comparisons to be made between them (Table S5).

Figure 1 shows the MFA results for the white wines, with the PC1 and PC2 axes accounting for 40.51 % of the total variance.

MFA reveals two primary sensory dimensions that distinguish the wines. Axis 1 captures the contrast between wines characterised by reduction (boiled vegetable, reduced, and sulphur) and acidity, which are positioned on the negative side, and those characterised by aromatic intensity and sweetness, which are located on the positive side. This axis reflects a key opposition in the sensory profiles, with wines distributed along it based on their balance of these contrasting attributes.

Axis 2 introduces a secondary differentiation, opposing wines characterised by alcohol intensity, pastry and empyreumatic descriptors (biscuit, caramel, empyreumatic) to wines described as aqueous, short and fruity descriptors (blackcurrant, exotic fruit). Together, these two axes provide a comprehensive framework for understanding the sensory diversity and complexity of the wines studied, which may be attributed to the grape variety or the winemaking process.

Figure 1 illustrates the variety of sensory profiles that can be produced from a single FRG cultivar. In this case, the ‘Floreal’ cultivar was used to produce 9 wines from various regions in the south of France. Some of these wines (Flo_8 and Flo_9) are associated with fruity and aqueous perceptions, while others (Flo_2 and Flo_3) are associated with sweetness perception. One wine (Flo_1) is associated with boiled vegetables and reduced aromas. It is worth noting that the nine ‘Floreal’ wines tested here had alcohol percentages ranging from 10.6 % to 14.8 % (Table S1), demonstrating the wide range of grape maturities used in these ‘Floreal’ wines.

Score plot (A) showing the positioning of individual wine samples and loadings plot (B) showing sensory descriptors. The analysis was based on significant CATA attributes from two consecutive vintages (Year 1: n = 19 panellists; Year 2: n = 17 panellists).

Samples made from Vitis vinifera grapes are highlighted in grey. The acronyms are explained in Table 1.

The impact of climate was particularly noticeable for ‘Floreal’, with Flo_7, Flo_8, and Flo_9 exhibiting similar sensory profiles due to being produced under the Southwest basin climate, as reflected by their close proximity in the MFA. Compared with the warmer Mediterranean climate, which accounted for the majority of the samples (34 out of 43 wines), wines produced under the cooler conditions of the Southwest basin were perceived as more aqueous by the panel. This effect may be due to familiarity, as participants recruited in Montpellier were probably more accustomed to the fuller-bodied wines typically produced locally in the Mediterranean region. The climate effect was less pronounced for Souv_5, which was produced in the Southwest basin and exhibited similar characteristics to Souv_2, Souv_3, and the Souv-Mus_1 and Souv-Mus_2 blend, all of which were produced under warmer Mediterranean conditions. The close proximity of Flo_2, Flo_3, and Flo_6, all of which were produced under the Mediterranean climate from different vintages, shows that the season had a more limited impact on sensory profiles than the varietal effect.

The sensory profiles of the eight blended wines revealed the distinct influence of their parent varietals. Blend Bq3159_Flo has a profile that is closely aligned with pure ‘Floreal’ wines of the Mediterranean type. Blends incorporating Muscat-scented varieties, such as G5 and ‘Muscaris’, are characterised by prominent Muscat notes (floral, rose, and white fruit), but these notes are less intense than in pure wines. Conversely, the ‘Sauvignac’ – ‘Muscaris’ blend is more strongly characterised by the ‘Sauvignac’ profile. Similarly, other blends containing ‘Sauvignac’ resemble pure wines made from this grape variety more closely. In contrast, ‘Souvignier gris’ appears to contribute no distinct sensory characteristics when used in blends, with wines containing this variety tending to align more closely with the aromatic and structural profiles of the other associated grape varieties.

For wines made from Vitis vinifera varieties, the central position of the sample on the MFA and the limited distance between samples of the same variety suggest less distinctive and pronounced profiles and lower diversity. However, this observation is based on a small sample size, and more wines are needed to draw firm conclusions. The samples analysed primarily serve as reference standards and may not fully represent the broader diversity of commercial wines currently available. As expected, the Chd_1 and Chd_2 samples were described as buttery, a term frequently used to characterise ‘Chardonnay’ wines (Bartowsky & Henschke, 2004). Surprisingly, ‘Sauvignon’ wines (i.e., Sau_1 and Sau_2) did not exhibit intense fruity or exotic fruit notes or aromatic character, despite this variety’s potential to produce wines with high levels of varietal thiols. This may be due to the warm and dry climatic conditions experienced in their region of production (Mediterranean climate), which did not favour the biosynthesis of thiol precursors (Roland et al., 2011). Similarly, ‘Viognier’ wines, namely Vio_1 and Vio_2, were not perceived by the panel as having an intense white fruit aroma. This attribute, together with stone fruit, is often used to describe these wines, and its intensity is related to the concentrations of linalool, geraniol and nerol (Siebert et al., 2018). Nevertheless, a more detailed study of the aroma composition of the wines in question would be valuable to build on the present work.

Overall, the lack of clear clustering between Vitis vinifera and FRG across the extensive wine sample examined in our study does not corroborate the hypothesis of FRG distinctiveness. This finding is consistent with previous research (Duley et al., 2025a).

For the red wines’ MFA (Figure 2), the PC1 and PC2 axes accounted for 36.62 % of the total variance.

Score plot (A) showing the positioning of individual wine samples and loadings plot (B) showing sensory descriptors. The analysis was based on significant CATA attributes from two consecutive vintages (Year 1: n = 19 panellists; Year 2: n = 17 panellists).

Samples made from Vitis vinifera grapes are highlighted in grey. The acronyms are explained in Table 2.

Axis 1 clearly distinguishes between wines characterised by acidity and aqueous attributes and those defined by sweetness and aromatic descriptors (amyl, aromatic, floral, violet). In contrast, axis 2 is primarily explained by descriptors from chemical families (chemical, amyl), and reduced attributes (reduced, sulphur, boiled vegetable), which further refine the differentiation of the wines’ sensory profiles. This axis highlights the influence of chemical nuances and reduction-related characteristics in defining the complexity of red wines, a factor that can be attributed to either the grape variety or the winemaking process.

Similar to the observations made for the white wines, the FRG red wines showed considerable variability in their sensory profiles, with some exhibiting spicy (e.g., Vid_1), fruity, oxidised and acid (e.g., Merlk1, Vid_22, Art_2), reduced and bitter (e.g., Bq3176_1, CabJur_2, Bd3328), or amyl, aromatic and floral (e.g., CabCor, Bq3176_2) characteristics.

Within the FRG varieties represented by the largest number of replicates, ‘Artaban’ and ‘Vidoc’ are of particular interest. ‘Vidoc’ samples showed clear differences in sensory expression, as indicated by their scattered positions on the MFA plot. Specifically, Vid_1 is perceived as having a higher alcohol intensity than the other samples. However, it should be noted that the alcohol percentage does not differ between Vid_1 and Vid_3 (both 14.22 %). Vid_2 is characterised by higher acidity and lower sweetness, consistent with the analytical data, particularly the pH levels and total acidity measurements for this wine. Additionally, Vid_2 is less spicy. In contrast, Vid_3 differs in its aromatic profile, which is described as more amyl. In contrast, the ‘Artaban’ wines were more homogeneous despite originating from different vintages and climatic conditions. The latter wines were generally described as fruity and tended to show lower viscosity and persistence, which is consistent with their negative association with fat and long attributes. The oenological parameters of the ‘Artaban’ wines are relatively uniform (Table S2). However, the Art_2 sample showed slight variations, most notably in its acidic profile (pH and total acidity), indicating a marginal differentiation.

A comparable degree of homogeneity was also observed for ‘Pinotin’. Conversely, some FRG red wines exhibited greater variability than ‘Vidoc’, particularly the Bouquet varieties (e.g., Bq3176_1, Bq3176_2, BqG14_1, and BqG14_2). As these wines were produced under experimental conditions, it is possible that the vintage effect had a greater influence on the results than the varietal characteristics themselves.

Regarding red wines made from Vitis vinifera, all samples were distributed along PC2 and exhibited greater variability in sensory profiles compared to the white wines. This can be attributed to the greater diversity of grape varieties used, including those traditionally planted in Bordeaux (‘Merlot’ and ‘Cabernet-Sauvignon’), the Mediterranean (‘Grenache’ and ‘Syrah’), and Burgundy (‘Pinot noir’). As previously noted, this principal component primarily captures differences in olfactory characteristics rather than taste or mouthfeel perception. Notably, the ‘Syrah’ sample (Syr) did not exhibit the highest intensity of sulphur or reduced aromas despite this variety’s known tendency to develop such notes (Geffroy et al., 2020). Similarly, ‘Pinot noir’ wines exhibited only weak cherry notes, despite this being a descriptor frequently used to describe wines made from this variety (Aubry et al., 1999).

‘Cabernet Volos’, ‘Cabernet noir’, ‘Cabernet Jura’, and ‘Merlot Khorus’ exhibited sensory profiles that were notably different to those of ‘Cabernet-Sauvignon’ and ‘Merlot’, respectively (Figure 2). The breeders named these FRG to indicate their Vitis vinifera parentage. However, our results suggest that there is no sensory resemblance between wines made from the parental varieties and their FRG derivatives, which could potentially mislead consumers. This issue requires the attention of regulatory authorities, particularly given the ongoing discussions at the OIV concerning varietal nomenclature (Resolution OIV‑VITI 609‑2019). According to the current protocol for naming a new variety, proposed names must avoid the use of names that may lead to possible confusion with the names of other existing varieties, particularly those that are already commercially established. Nevertheless, these findings should be interpreted with relative caution, given that our experimental design included only one wine produced from each of ‘Merlot’ and ‘Cabernet-Sauvignon’.

Despite the absence of a year effect in the analysis and the robust performance of our panel, it should be noted that the MFA representation of the wines (Figures 1 and 2) explains a maximum of 40 % of the total variability in the data for both colour categories. This relatively low percentage is due to the large number of variables included in the model (35 for white wines and 26 for red wines). As only statistically significant descriptors were selected for these analyses, these results highlight a high degree of sensory variability in the perception of the wines studied.

Statistically significant correlations were observed between sensory attributes and oenological parameters for both white and red wines. Specifically, the perception of alcohol was positively correlated with its actual content (r = 0.62 and r = 0.54), while perceived acidity showed significant correlations with measured pH levels (r = –0.64 and r = –0.62) and total acidity (r = 0.79 and r = 0.42). These alignments between chemical analyses and sensory data demonstrate the trained panel’s ability to evaluate these wine attributes accurately.

The study reveals distinct sensory dynamics between white and red fungus-resistant grape (FRG) wines. White FRG wines exhibited high variability, which was significantly driven by vintage and climate. Certain varieties, such as ‘Floreal’, showed a wide range of profiles, from boiled vegetables to fruity notes, depending on their origin. In contrast, red FRG wines, such as ‘Artaban’, displayed more consistent sensory patterns, regardless of production context. Furthermore, the sensory panel identified a higher number of significant olfactory descriptors for white wines (11) than for red wines (6), which is likely due to the larger sample size and the greater aromatic diversity found among white varieties.

Furthermore, analyses of white and red wines revealed no significant overall sensory differences between new and traditional grape varieties. This finding is consistent with recent research (Duley et al., 2025b). This suggests that the aromatic and gustatory potential of the new varieties is comparable to that of traditional varieties and is mainly influenced by factors such as grape ripeness, winemaking practices, soil and climate conditions rather than genetic origin alone.

As a complementary study, it would be interesting to investigate the differences between FRG and Vitis vinifera wines, with respect to their content of key grape-derived aroma compounds such as terpenes (monoterpenols and rotundone), C13-norisoprenoids, varietal thiols, and alkylmethoxypyrazines (Robinson et al., 2014). Similarly, it would be valuable to evaluate the appreciation of the large set of wines studied by both experts and consumers (Borello et al., 2024), either with or without disclosing information about the fungus-resistant characteristics, in line with the approach adopted by Dolet et al. (2026).

Although this study involved a large sample of wines and constitutes, to our knowledge, the largest-scale research project on FRG sensory profiles to date, it is important to note that such profiling is still in its early years in France. There are currently limited quantities of FRG wines on the market as the cultivation of these varieties has only recently resumed (Dolet et al., 2024). These authors have demonstrated that Occitanie (in the south of France) leads the way in France in terms of areas planted with FRGs. For representativeness and logistical reasons, the majority of the wines tested in this study were sampled in this region. However, it would be interesting to extend this sensory characterisation to FRG wines from other French regions. Furthermore, including a broader range of vintages would account for interannual climate variability. The present study focused primarily on the 2022 and 2023 vintages for white wines and the 2021, 2022, and 2023 vintages for reds. Analysing additional years would provide a more comprehensive and diverse overview of the sensory potential of these cultivars.