The balance between youthful fruitiness and tertiary aromas of fine aged wines within a complex, harmonious ensemble is particularly sought-after by wine professionals (Robinson and Harding, 2006; Jackson, 2009).

A significant mystery still surrounds the development of positive wine aging characteristics, usually called aging bouquet. Its chemistry and the factors responsible are not fully understood or well-characterized. Accordingly, deciphering the aging bouquet typicality of red wines remains one of the main current challenges in wine science. Recently, major advances have been made toward elucidating this complex olfactory concept, both from sensory and molecular perspectives. Specific aromatic properties (Picard et al., 2015a), as well as the contribution of several chemical markers (Picard et al., 2015b, 2016a), have been identified.

Moreover, it is well-established that the concepts of typicality and wine quality are closely linked with the notion of terroir (Jackson and Lombard, 1993; Sauvageot, 1994; Salette, 1997; Verdù-Jover et al., 2004; Casabianca et al., 2005; van Leeuwen and Seguin, 2006; Cadot et al., 2010, 2012). Nevertheless, even if the importance of terroir has been clearly identified in the conceptualization of the aging bouquet by wine professionals (Picard et al., 2015a), as well as in the perceived difference in mint aroma intensity (Picard et al., 2016b), no in-depth study had previously investigated how certain viticultural parameters conditioned the genesis of an aging bouquet after several years of bottle storage.

Among the various parameters of terroir that influence grape quality, the key role of vine water uptake conditions has repeatedly been emphasized (van Leeuwen and Seguin, 1994; van Leeuwen et al., 2004; Peyrot des Gachons et al., 2005; De Royer Duprè et al., 2014). More precisely, it has been shown in Bordeaux vineyards that a limitation in water supply (known as “water deficit”) is a key factor in explaining the vines' physiological and biochemical responses, leading to enhanced winemaking potential in the grapes (van Leeuwen et al., 2009). Water deficit occurs when vine transpiration and evaporation from the soil surface exceed incoming rainfall and irrigation (Lebon et al., 2003). In Bordeaux, all vineyards are dry farmed, so climatic factors are predominant in this study. In this process, soil water acts like a buffer: it compensates for climatic water deficits up to a certain threshold. When approximately 60% of the plant-available water reserves of the soil have been depleted, vines experience water deficit and trigger mechanisms, like stomatal closure, to save water (Pellegrino et al., 2006). The training system also influences the water balance by impacting light interception by the vines, which provides the energy for transpiration (van Leeuwen et al., 2010). Hence, vine water status is dependent on climate (variable from one vintage to another), soil (variable from one estate to another and even between plots on a given estate), and training system (variable from one estate to another; van Leeuwen et al., 2009). Training systems are rather homogeneous in high-quality Bordeaux wine producing vineyards (high-density, vertical-shoot positioned system) and are, therefore, unlikely to have a major influence on results in this study.

Due to the effect of carbon isotope discrimination during the photo-assimilation of CO₂ (Farquhar et al., 1989; Gaudillère et al., 1999), the ¹³C/¹²C ratio measured in grape sugar at ripeness (usually expressed as δ¹³C) is considered an integrated physiological indicator for assessing the average vine water status during grape ripening (van Leeuwen et al., 2001; Gaudillère et al., 2002). Moreover, it has been shown that δ¹³C, measured in grape sugar at harvest, is well-correlated with the ethanol δ¹³C in wine produced from these grapes (Roßmann et al., 1996).

Recently, an easy-to-implement method based on isotope ratio measured by mass spectrometry coupled to an elemental analyzer (irm-EA/MS) was developed to evaluate wine carbon ¹³C isotope composition and validated using a set of wine samples produced from well-characterized vineyard blocks (Guyon et al., 2015). This analytical method highlighted a clear correlation between wine δ¹³C and the water status of the vines on which the grapes had grown. These results provided a relevant scientific basis for tracing the water status history of the vines from which aged red wines originated. Consequently, this strategy offered an appropriate tool for evaluating the impact of water vine uptake conditions on aging bouquet expression.

In this study, our aim was to investigate the link between vine water status and the sensory expression of aging bouquet typicality in mature red Bordeaux wines. It focused particularly on undergrowth, truffle, and mint notes, since their chemical composition had already been partly deciphered. Indeed, the direct contribution of dimethyl sulfide (DMS) to truffle and undergrowth attributes, as well as the overall complexity and balance of wine has been widely reported (Du Plessis and Loubser, 1974; Spedding and Raut, 1982; Segurel et al., 2004; Picard et al., 2015b). Moreover, it has already been shown that the potential of DMS (PDMS), susceptible to be chemically released by S-methyl methionine degradation (Segurel et al., 2004; Loscos et al., 2008) is dependent upon numerous terroir factors, including vine water status, grape maturity at harvest, and the yeast assimilable nitrogen content of must (Bell and Henschke, 2005; Dagan, 2006; Dagan and Schneider, 2012; De Royer Duprè et al., 2014). Concerning the mint aroma, a varietal effect on a wine's piperitone profile has recently been reported (Picard et al., 2016b) and the presence of p-menthane lactones, resulting from the same monoterpene limonene secondary biotransformation pathway as piperitone, was recently reported in wine (Picard et al., 2017), suggesting a common varietal origin for all these minty-terpenoid aromatic compounds.

After a sensory characterization of the aging bouquet typicality of some mature red wines, the bulk wine δ¹³C ratio was measured to estimate the vine water status during grape ripening. This parameter was then used to evaluate whether the sensory perception of aging bouquet typicality was related to the water status of the vines from which the wines originated. Finally, the impact of vine water status on the expression of “mint,” “undergrowth,” and “truffle” descriptors was also investigated.

Previous published data had not detected strict correlations between aging and bouquet typicality scores and wine age, although a vintage effect was highlighted (Picard et al., 2015b). As vintage quality is closely determined by the climatic conditions during vine development and grape ripening, particularly vine water status (van Leeuwen et al., 2009; van Leeuwen and Darriet, 2016), we hypothesized a potential impact of vine water deficit on aging bouquet development during bottle storage.

Principal Component Analysis (data not shown) was firstly used to validate the consensus among wine professionals on aging bouquet typicality. According to ANOVA, the effect of wine was significant for the typicality score (F = 6.426; p < 0.001), mint (F = 8.099; p < 0.005), truffle (F = 10.913; p < 0.001), and undergrowth (F = 5.861; p < 0.005) notes, indicating that these attributes were useful in characterizing aromatic differences among the wines.

The 42 wines studied during the two tasting sessions, in 2015 and 2017, were then clustered into three main groups according to the data obtained from the typicality task. An aging bouquet typicality gradient emerged, ranging from wines with the lowest (19 wines; typicality scores from 1.5 to 4.3) to wines with the highest average typicality scores (13 wines; typicality scores from 4.6 to 6.7). Between these two clusters, a group consisting of 10 wines was identified as “intermediates,” as their typicality scores varied drastically among wine professionals.

The next step was to investigate the existence of a correlation between aging bouquet expression in the wines studied and vine water status.

The bulk wine ¹³C/¹²C ratio was measured in all wines and the corresponding δ¹³C-values computed. Results ranged from −27.5 to −24.2%. Previous experiments conducted on grape must had identified a significant correlation between δ¹³C measured on grape sugar and water deficit (Gaudillère et al., 2002; van Leeuwen et al., 2009). Moreover, a study of the bioconversion of sugar to ethanol during the fermentation process revealed that δ¹³C is 1.7% lower in wine ethanol than grape sugar (Roßmann et al., 1996). Consequently, δ¹³C measured in bulk wine was a good indicator for evaluating vine water status during grape ripening. Accordingly, and relying on previously published data, three levels of vine water deficit were defined: slight water deficit when the δ¹³C-values were lower than −26.2%, moderate water deficit when δ¹³C was between −26.2 and −24.7%, and severe water deficit when δ¹³C exceeded 24.7% (Figure 1). All wines tasted were classified according to their δ¹³C ratio and typicality score (Figure 2). Overall, the results showed a link between aging bouquet expression and the corresponding vine water status. Most of the wines with a marked aging bouquet were produced in vintages where the vines were subjected to a moderate to severe water deficit (i.e., 1995, 1998, 2000, 2005, 2006, and 2009; δ¹³C > −25.6%; Figure 2). On the contrary, most of the wines in the group with the lowest typicality scores were produced in “wet” vintages with a weak to moderate water deficit (i.e., 1996, 1997, 1999, 2002, and 2008).

The statistical Spearman test showed that δ¹³C differentiated significantly between the most- and least-typical wines (p < 0.001, Table 2). Moreover, a significant positive correlation was observed between aging bouquet typicality scores and vine water deficit status, with significantly higher δ¹³C-values measured in wines with a high typicality score (Table 3). However, for some vintages, a wide dispersion was observed concerning the water status indicator. Even in the same vintage, different vine water regimes may exist in the four appellations studied, due to spatial variations in rainfall and/or soil water-retention capacity. This was the case, for example, in 1998: wines in the Pomerol series (P1 and P2) were associated with a severe water deficit (δ¹³C = −24.3 and −24.8%, respectively) whereas the Saint-Emilion wine was produced from vines facing a slight water deficit (δ¹³C = −26.11%). The same tendency was observed in 1999, when wines from Saint-Emilion (δ¹³C = −27%) were produced under slight water deficit conditions, while those from Saint-Estèphe and Pomerol had faced moderate water deficits (δ¹³C = −25.5 and −25.6%, respectively). The 2003 vintage was considered atypical, as high temperatures were associated with regular rainfall. Under these conditions, the vines were subjected to only moderate water deficit, as observed in the P2 Pomerol series (δ¹³C = −26.2%), thus confirming the results of a previous study (van Leeuwen et al., 2009).

However, within the Saint-Emilion series, the wine produced in 2003 had a low ¹²C/¹³C ratio (δ¹³C = −26.9%), corresponding to a slight vine water deficit, instead of the expected moderate value. Differences in δ¹³C within the same vintage are, at least partly, explained by differences in Soil Water Holding Capacity (SWHC). Average SWHC was higher in the Saint-Emilion vineyard (SEM) than the Pomerol, Margaux, or Saint-Estèphe vineyards, accounting for the less severe water deficits in most vintages in SEM. SWHC also varies from one plot to another. Hence, δ¹³C-values reported for the wines in this study reflected the average vine water deficit of the plots where the grapes in the final blend of the estate wine were harvested. This is one of the limitations of this study. Greater precision would have been possible if the wines had been produced from a single plot, which is rarely the case in the Bordeaux area.

Furthermore, the typicality scores and sensory intensities of mint, undergrowth, and truffle notes, as well as the ethanol ¹³C/¹²C ratio in the wines, were statistically processed using PCA, considering the first two components (Figure 3). The first principal component explained 53.7% of total variance and was positively correlated with all studied data. This result emphasized the good correlation between the typicality score and δ¹³C-values. Moreover, the loading plots for mint and truffle intensity ratings were projected on the same positive side of the first axis and were both close to the typicality score and δ¹³C loading plots. These two aromatic notes were thus well-correlated with the typicality of the wine aging bouquet and, more importantly, with the ¹³C isotope distribution in the wines. Although the undergrowth note was also projected on the positive side of the first axis, its position was close to the correlation circle, indicating that the impact of δ¹³C on its expression was less clear.

These results were confirmed using the Spearman correlation test: δ¹³C correlated significantly and positively with truffle and mint notes, while it only marginally correlated with the undergrowth note (Table 3).

Vintage quality in Bordeaux is known to be related to vine Water Deficit Status Index (WDSI), calculated between veraison and ripeness. The winemaking potential of the grape will suffer both in the case of no water deficit as in the situation of excessive water stress (van Leeuwen et al., 2009). The ideal water uptake conditions for producing high quality grapes for the production of red wine corresponds to a moderate water deficit, preferably starting early in the season (i.e., before veraison). Such an early water deficit has a greater effect on the reduction of berry size than late water deficit (Ojeda et al., 2002). However, the impact of vine water status on the development of the aging bouquet in red Bordeaux wines, as well as on several of its key aromatic notes, had not previously been described. Interestingly, our results revealed that the vine water deficit level significantly impacted not only overall aging bouquet typicality, but also mint, truffle, and, to a lesser extent, undergrowth aroma perception. These observations confirm several previous studies, which, although not dealing with aroma expression in wines, also showed a positive impact of a moderate water deficit on grapes quality potential, through an increase in phenolic compounds (in particular anthocyanins) and grape sugar concentrations (Duteau et al., 1981; Matthews and Anderson, 1988; van Leeuwen and Seguin, 1994; Ojeda et al., 2002). Peyrot des Gachons et al. (2005) reported that mild water deficit increases the concentration of volatile thiol precursors in Sauvignon blanc, while severe water deficit has a negative impact. Moreover, in a study comparing soil and climate effect on grape ripening, it was shown that grape quality potential was highest on clayey soils, because these induced early but moderate water deficits in most vintages (van Leeuwen et al., 2004).

Overall, these new advances support the idea that the terroir—and particularly vine water status—plays a key role in the corresponding molecular marker levels in wine and thus, influences the subsequent development of the related aromatic notes during positive wine aging.

This study provides new insights into the influence of vine water status on aging bouquet expression in fine red Bordeaux wines. A recently-developed method, based on carbon-isotope discrimination measured in wine ethanol, applied to a large set of aged red wines from five different Bordeaux vineyards, confirmed the impact of vine water deficit on the development of the aging bouquet and, more precisely, the relevant impact of a moderate-to-severe water deficit on mint and truffle aroma expression.

Further work is still required to obtain a more comprehensive understanding of the links between water deficit and the aging bouquet on a molecular level. Following the recent identification of several molecular markers responsible for truffle and mint notes in wines with an aging bouquet, it would be relevant to perform a quantitative analysis of these specific aroma compounds, as well as their precursors, in monovarietal wines from various vintages and different soils, in order to investigate the relationship with vine water status. This next step is necessary to unravel the mechanisms that explain the biochemical pathways involved in this relation.

Our current work is only based on physicochemical analyses of commercial wines samples. It does not concern any sample issued from animal or human biological sources and, accordingly, does not require any ethics review process or written informed consents. Concerning the wine tastings, all wine professionals were consented and informed about the context of the study.

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.