Soil types as extrinsic cues differentially shape sensory perception of German Riesling wine
Abstract
Despite the lack of consensus regarding the effects of different soils on wine flavour, the simplification of terroir to soil type has become increasingly common in wine marketing, particularly for German Riesling. To examine the effects of this strategy on sensory perception, twenty German Rieslings from four commonly marketed soil types (Buntsandstein, Löß, Muschelkalk, and Schiefer) were evaluated for aroma, taste, and mouthfeel attributes (rate-all-that-apply) by a panel of twenty wine experts, first without and then with each wine’s soil type given as an extrinsic informational cue. While ratings from the uninformed tastings suggest baseline differences among Rieslings from the various soils, the provision of soil type cues resulted in more extensive discrimination, particularly in terms of attributes related to minerality (earthy, chalky, marine, flinty, and salty). Soil type-dependent changes in ratings suggest distinct a priori conceptions for each soil shaped the panel’s expectations and differentially altered their perception, though further work is needed to determine whether nonexpert consumers would be similarly biased by these cues. For wine soil typing to be a viable marketing strategy, a well-defined “image” of each soil would need to be communicated to consumers, to ensure their expectations are set to trigger perception of the desired sensory properties.
Introduction
Widely regarded as a fundamental antecedent to wine quality, terroir (from the French terre, meaning “land”), may be conceptualised as the collection of interactions between the grapevine and its physical environment (van Leeuwen, 2022). Vine development, grape ripening, and consequently the sensory properties of wine produced are influenced by many external factors including weather and climate, human intervention, and the prevailing microbiome (Seguin, 1986; van Leeuwen and Seguin, 2006; Bokulich et al., 2016), which altogether give wine a “sense of place.”
Regarding the geological aspects of terroir, the notion that minerals derived from various soil substrates can be taken up by the grapevine and confer distinct sensory properties to wine has largely been disproven (Maltman, 2013; Parr et al., 2018). The influence of soil is more likely indirect, controlling such parameters as root penetration depth, nutrient availability, and water-holding capacity (Huggett, 2006; Willwerth and Reynolds, 2020). However, there is no clear consensus regarding the differential effects of specific soil types: while some studies have found significant differences in the flavour profiles of wines originating from different soils (de Andres-de Prado et al., 2007; Löhnertz et al., 2008; Bauer et al., 2011), others have demonstrated that such effects, if any, are negligible in comparison to those from other aspects of terroir, particularly local winemaking practices and climatic conditions (Rankine et al., 1971; Noble, 1979; Wahl, 1988; Bader and Wahl, 1996; Fischer et al., 1999).
Nonetheless, it has become increasingly common in wine marketing to reduce terroir down to soil type. Information regarding the geological substrate found at vineyard sites, such as slate or limestone, is often provided to consumers, in some cases even replacing geographical indications on bottle labels, price lists, etc. (Bauer et al., 2011). This appears especially true for Riesling in Germany since techniques such as barrel ageing and malolactic fermentation are not widely used; by minimising the impact of winemaking, it is believed more genuine expression of the environment can be achieved (Robinson, 2015; Bauer et al., 2011). Naming specific soil types when marketing wine presumably affects how it is perceived, though any precise outcomes from this strategy have yet to be determined.
Extrinsic informational cues can give rise to expectations that a product will exhibit certain characteristics, consequently modifying perception during consumption (Deliza and MacFie, 1996; Brochet and Morrot, 1999). Consumers will perceive added value, particularly from information that solidifies the link between a product and its origin, such as explicit mentions of production size or sustainable practices (Lenglet, 2014; Capitello et al., 2021). However, for such cues to effectively trigger consumer associations and shape their expectations, there must exist a consistent a priori “image” of the place of origin (van Ittersum et al., 2003). That is to say, with regard to the soil typing of wine, terms such as “slate” or “limestone” must carry meaning, but given the lack of consensus regarding the effects of different soils on wine flavour, it is uncertain whether this assumption is valid.
We propose tasters’ preconceptions of soil types can be determined by controlling the information they receive during sensory evaluation, as has been done in previous studies to reveal biases and expectations (Siegrist and Cousin, 2009; Peterson, 2014; Boncinelli et al., 2016). For the study presented here, our sensory panel evaluated German Rieslings first without and then with cues informing them of the soil type each wine came from. Between the uninformed and informed tastings, sensory perception was expected to shift differentially according to any panel preconceptions, which would confirm each soil type has a distinct “image” and thus justify how these wines are being marketed.
Materials and methods
1. Wines
Twenty German Riesling wines from the 2020 and 2021 vintages were evaluated in this study. Wines were chosen from vineyards cultivated on one of four commonly marketed soil types: Buntsandstein (coloured sandstone), Löß (loess), Muschelkalk (shell-bearing limestone), and Schiefer (slate), as stated by each producer. To ensure variance, each soil type was represented by five wines from different producers throughout winegrowing regions where Riesling is the predominant white variety (Supplementary Table 1), though the influence of producer, region, and physicochemical composition was not the focus of our study. All wines were donated by or purchased directly from each producer.
Additionally, to evaluate the effects of each soil type cue under limited influence from other variables, a composite control wine was prepared by blending all twenty wines in equal proportions one day prior to the study. Bottles were filled under nitrogen, sealed with screwcap closures, and stored in the same conditions as the other wines until sensory evaluation.
2. Participants
Twenty participants (4 female, 16 male) with a mean age of 47 (range 25–67) comprised the panel of judges for this study, all considered wine experts according to the criteria proposed by Parr et al. (2002). Email invitations were sent to Weincampus alumni and other wine industry affiliates in Germany. The study was presented as a unique opportunity to taste vineyard-designated Riesling from throughout the country, as well as network with fellow industry members. Those who responded with interest were sent additional information regarding the study’s agenda and format before formally consenting to participate, in keeping with the ethical requirements of the Weincampus Neustadt, Dienstleistungszentrum Ländlicher Raum (DLR) Rheinpfalz. The study was conducted in accordance with the Declaration of Helsinki for studies involving human subjects.
3. Sensory evaluation
The study took place in the sensory laboratory at Weincampus Neustadt. Upon their arrival, the participants were asked to smell aroma standards to familiarise themselves with the attributes to be evaluated (Table 1). Taste and mouthfeel attributes were also evaluated (sweet, sour, salty, bitter, alcohol, body, and length), though no standards for these were provided. Attributes were selected by sensory science staff members at the Weincampus with prior research experience with Riesling. The list of attributes was compiled from those used in several previous studies on Riesling (Fischer et al., 1999; Bauer et al., 2011; Douglas et al., 2001; Schüttler et al., 2015), and additionally included terms often used to describe minerality (Ballester et al., 2013; Maltman, 2013, Heymann et al., 2014; Rodrigues et al., 2015). It should be noted this study took place in German, and the attributes here have been translated as closely as possible into English.
Aroma attribute | Standard ingredients |
apple/pear | 20 mL apple juice (Rewe, Cologne, North Rhine-Westphalia), 20 mL pear juice (Lösch's, Ramstein-Miesenbach, Rhineland-Palatinate) |
citrus | 25 mL freshly squeezed lemon juice, 20 mL freshly squeezed grapefruit juice |
peach/apricot | 40 mL peach/apricot juice blend (Eos, Weinstadt, Baden-Württemberg) |
tropical | 40 mL mango-passionfruit juice blend (Rewe) |
sweet/sugary | 15 g honey (Bienenwirtschaft Meißen, Meißen, Saxony), 10 g fig marmelade (Bonne Maman, Breuberg, Hessen) |
floral | 10 mL elderflower syrup (Rewe), 20 uL linalool (2 μg/μL in ethanol) |
fresh/grassy | 50 g diced cucumber, 2 g fresh grass |
vegetal | 10 g canned green beans (Bonduelle, Villeneuve d'Ascq, Hauts-de-France), 15 g diced bell pepper |
herbal | 0.5 g green tea (Costco Kirkland Signature, Issaquah, Washington), 0.5 g herbes de Provence (Ostmann, Dissen am Teutoburger Wald, Lower Saxony) |
spicy | 0.2 g ground cinnamon, 0.2 g ground nutmeg, 0.1 g ground white pepper (Ostmann) |
yeasty/lactic | 1 g yeast (Uniferm, Werne, North Rhine-Westphalia), 1 g butter (Rewe), 1 g yoghurt (Weihenstephan, Freising, Bavaria) |
reduced | 20 uL sodium sulfide (0.5 μg/μL in water) |
nutty/oxidised | 1 g roasted almond butter (Rewe), 5 mL Pedro Ximénez sherry (Lustau, Jerez de la Frontera, Andalusia) |
TDN | 10 uL 1,1,6-trimethyl-1,2-dihydronaphthalene (0.1 μg/μL in ethanol) |
chalky | 1 g powdered chalk, 5 uL geosmin (0.05 ng/μL in ethanol), in 100 mL of 25 % (v/v) base wine/water solution |
earthy | 2 g chopped portabello mushrooms, 10 μL geosmin (0.05 ng/μL in ethanol) |
flinty | 10 uL benzylmercaptan (1 ng/μL in ethanol) |
marine | 1 g roasted seaweed (Kreyenhop & Kluge Miyako, Oyten, Lower Saxony), 10 mL dimethyl sulfide (0.25 μg/μL in ethanol) |
metallic | 10 uL 1-octen-3-one (0.1 μg/μL in ethanol) |
Judges were seated at individual booths with computers where FIZZ Software v2.5 (Biosystèmes, Couternon, France) was used for sensory evaluation. The study consisted of two 90minute sessions, separated by a onehour break. Opening and pouring (30 mL) of the wines began in a separate room approximately one hour before the start of each session. Wines were served at ambient temperature (20 ºC) in clear DIN 10960 Sensus wineglasses (Zwiesel Kristallglas, Zwiesel, Germany), each labelled with a random three-digit code and covered with a plastic lid. The wines were served in a balanced random order for each judge according to a Williams Latin square design to control for presentation order and carryover effects.
During session 1, the judges knew only that the wines were German Rieslings from 2020 or 2021. The control wine was presented to each judge in quadruplicate (n = 4), while the 20 original wines were each presented once. For session 2, the judges were made aware of the soil type from which these wines originated, with each wine’s soil type appearing on FIZZ alongside its three-digit code. The 20 original wines were again each presented once, while the control was presented a total of 8 times: each of the 4 soil types (Buntsandstein, Löß, Muschelkalk, and Schiefer) was shown twice for the control, such that the effect of each cue on the perception of the control was evaluated in duplicate (n = 2 for each soil type). Accounting for replication, each judge conducted in total 24 evaluations during session 1, and 28 evaluations during session 2.
Rate-all-that-apply (RATA) was used for sensory evaluation (Ares et al., 2014). Aroma attributes were presented on a single page on FIZZ in a random order for each judge, followed by a second page with randomised taste and mouthfeel attributes. Judges rated the intensity of attributes they deemed perceptible by clicking on an unstructured line scale anchored by “not present” and “very intense” (valued from 0 to 100); unrated attributes were assumed an intensity of 0. Judges were instructed to pause in between wines as needed to avoid fatigue; water and flatbread were available for the judges throughout the study. They proceeded at their own pace, without consulting those around them, and the majority completed each session in less than 90 min.
4. Data analysis
All data was analysed in RStudio v2023.03.0+386 (Boston, Massachusetts, USA) running R v4.2.1, with the additional packages reshape2, lmerTest, FactoMineR, emmeans, multcomp, and factoextra. A confidence level of 95 % was used for all significance testing (α = 0.05).
To compare the wines and soil types from session 1, a two-way multivariate analysis of variance (MANOVA) using Wilk’s lambda was first conducted using replicate, wine, and soil as the fixed effects with the wine effect nested within soil. Given the control was evaluated in quadruplicate (n = 4) by each judge, the replicate effect was included in the MANOVA model to assess panel repeatability. The control was included as its own category, alongside Buntsandstein, Löß, Muschelkalk, and Schiefer, under the soil effect. To determine which attributes characterised each soil type, univariate analysis of variance (ANOVA) was conducted for each attribute using a two-way mixed model with soil and judge as fixed and random effects, respectively. Attributes for which soil was a significant source of variance were used for principal component analysis (PCA). The correlation matrix of mean ratings for these attributes, calculated across judges for each wine, was used to construct the wine product space. Marginal means for each attribute were calculated for each soil type, after which they were compared post hoc using Tukey’s honestly significant difference (HSD) test.
Data from session 2 was analysed in a similar manner as the data from session 1, though the controls were categorised according to the soil type with which they were labelled. Additionally, the soil*replicate interaction term was included in the MANOVA model, to check judge repeatability when evaluating each of the differently labelled controls (n = 2 for each soil type). ANOVA and PCA were conducted as described above, after which marginal means were also calculated and compared via Tukey’s HSD.
To examine more closely the effects of each soil type cue, the difference in ratings between session 2 and session 1 (Δ, “delta”) was calculated for each judge, attribute, and wine. A positive Δ would indicate a judge perceived an attribute more intensely upon knowing a wine’s soil type, whereas a negative Δ would indicate a decrease in perceived intensity. For the labelled controls in session 2, Δ values were calculated by subtracting the mean for the unlabelled control from session 1 (n = 4 for each judge). MANOVA and ANOVA were conducted as described above for session 2, but using Δ values rather than intensity ratings. PCA was done using mean Δ values for each soil type.
Results
1. Session 1: Uninformed tastings
The effects of soil and wine (nested within soil) were significant according to MANOVA, indicating significant differences among soil types and among wines within each soil type. The replicate effect was not significant, suggesting judges performed repeatably in their evaluations of the composite control wine (n = 4). According to ANOVA, soil was a significant source of variance for eight aroma attributes as well as four taste and mouthfeel attributes (Table 2); the remaining attributes were excluded from subsequent analysis of the data from session 1.
Buntsandstein | Löß | Muschelkalk | Schiefer | Control | |
tropical.A | 17.1 abc | 18.8 bc | 10.6 a | 23.2 c | 14.5 ab |
sweet/sugary.A | 12.9 a | 16.7 ab | 10.6 a | 23.5 b | 9.3 a |
floral.A | 22.4 ab | 22.8 ab | 18.4 a | 30.6 b | 15.1 a |
herbal.A | 20.9 ab | 17.2 ab | 23.1 b | 17.0 ab | 14.0 a |
spicy.A | 18.8 ab | 16.2 a | 26.6 b | 20.6 ab | 15.8 a |
yeasty/lactic.A | 6.6 ab | 7.9 ab | 9.4 ab | 12.3 b | 5.4 a |
nutty/oxidised.A | 9.7 a | 11.5 ab | 18.8 b | 10.1 a | 5.3 a |
TDN.A | 13.6 a | 10.3 a | 26.8 b | 10.1 a | 7.4 a |
sweet.T | 36.2 b | 40.5 bc | 25.5 a | 47.1 c | 33.2 ab |
sour.T | 45.0 b | 43.3 b | 53.0 c | 36.3 a | 44.8 b |
bitter.T | 24.5 ab | 23.0 a | 31.2 b | 21.8 a | 20.6 a |
body.M | 46.9 a | 53.1 b | 46.7 a | 53.0 b | 46.5 a |
The plots generated from PCA illustrate the perceived differences and similarities among wines (Figure 1a) based on these significant sensory properties (Figure 1b). Explaining 43.9 % of the variance, PC1 appears to be driven mainly by differences between wines from Muschelkalk and Schiefer. All Muschelkalk wines (M1, M2, M3, M4, and M5) are found in the positive direction of PC1, correlated primarily with TDN, bitter, and sour, and to a lesser extent also nutty/oxidised, spicy, and herbal. With the exception of wine S4, the Schiefer wines (S1, S2, S3, and S5) tend toward the negative direction of PC1, correlated with tropical, floral, sweet/sugary (aroma), and sweet (taste). Wines M4 and S5 were rated highly in yeasty/lactic, nutty/oxidised, and body, and lie separate from the other wines along PC2. Wines from Buntsandstein (B1, B2, B3, B4, and B5) and Löß (L1, L2, L3, L4, and L5) are generally closer to the centre of the score plot, along with the control (Ctrl). Despite the likelihood of chemical and sensory interactions upon blending twenty wines (e.g. enhancing or masking effects), the control is located just below the centre of the score plot along PC2, which explains only 20.5 % of the overall variance. The situation of the control in this region of the plot, where attribute loadings are generally absent, suggests it is a proper “average” uncharacterised by any particular attributes.
A comparison of the marginal means for each soil type suggests Schiefer and Muschelkalk yield the most distinct wines (Table 2). Muschelkalk wines were rated exceptionally high for TDN and sour. They were also the most nutty/oxidised and bitter, as well as the least sweet, though these were attributes shared with Löß, Buntsandstein, and the control, respectively. On the other hand, Schiefer wines were the least sour; they were also rated most highly along with Löß for sweet/sugary aroma and sweet taste. Wines from Buntsandstein and Löß, as well as the control, always shared groupings with other soil types.
2. Session 2: Informed tastings
Results from MANOVA on the data from session 2 indicate significant differences among soil types and among wines within each soil type. Neither replicate nor the soil*replicate interaction were significant sources of variance, suggesting judges evaluated each of the labelled controls repeatably (n = 2 for each soil type). According to ANOVA, several attributes not found to significantly differ among soils in session 1 were then significantly different in session 2: apple/pear, citrus, peach/apricot, fresh/grassy, vegetal, reduced, chalky, earthy, flinty, marine, salty, and alcohol (Table 3). On the other hand, soil was no longer a significant source of variance for spicy (p = 0.387), nutty/oxidised (0.145), and TDN (0.787).
Buntsandstein | Löß | Muschelkalk | Schiefer | |
apple/pear.A | 29.1 b | 28.0 b | 21.5 a | 27.8 b |
citrus.A | 34.8 b | 28.4 a | 29.8 ab | 35.3 b |
peach/apricot.A | 30.8 b | 25.7 ab | 24.5 a | 30.6 ab |
tropical.A | 19.5 b | 16.4 ab | 12.0 a | 19.0 b |
sweet/sugary.A | 10.2 ab | 13.2 b | 8.4 a | 14.2 b |
floral.A | 22.5 b | 18.4 ab | 15.5 a | 23.8 b |
fresh/grassy.A | 8.9 ab | 5.2 a | 10.9 b | 9.8 b |
vegetal.A | 7.3 a | 8.2 ab | 12.2 b | 6.9 a |
herbal.A | 15.5 ab | 13.8 a | 18.4 ab | 20.1 b |
yeasty/lactic.A | 5.8 ab | 4.1 a | 5.9 ab | 8.3 b |
reduced.A | 10.6 ab | 8.3 a | 10.7 ab | 12.1 b |
chalky.A | 4.3 a | 5.6 a | 11.5 b | 4.7 a |
earthy.A | 7.7 a | 11.5 b | 7.1 a | 6.7 a |
flinty.A | 11.9 ab | 9.1 a | 15.1 b | 21.7 c |
marine.A | 1.4 a | 2.2 ab | 4.5 b | 3.2 ab |
sweet.T | 41.0 b | 42.1 b | 32.6 a | 43.4 b |
sour.T | 47.2 ab | 42.9 a | 48.9 b | 46.4 ab |
salty.T | 23.0 a | 24.3 ab | 28.4 bc | 29.6 c |
bitter.T | 21.8 ab | 22.2 ab | 24.5 b | 18.6 a |
alcohol.M | 55.6 b | 56.2 b | 57.4 b | 48.2 a |
body.M | 52.6 b | 55.1 b | 54.0 b | 48.5 a |
In the PCA score plot (Figure 2a), wines from Muschelkalk can be found in the positive direction of PC1 (29.6 % of the variance), while most wines from the other soil types tend toward the negative direction. The positioning of the Muschelkalk wines indicates they were rated highly for chalky, vegetal, marine, and bitter, but low for various sweet, fruity flavours (Figure 2b). The Schiefer, Buntsandstein, and Löß wines are generally separated along PC2 (18.1 % of variance) from top to bottom. Their locations in the score plot suggest the Schiefer and Löß wines were perceived as opposites: Schiefer wines were characterised by more tropical, floral, and flinty aromas, whereas Löß wines were higher in earthy aroma, body, and alcohol. Furthermore, the Löß wines were generally lacking in several other attributes driving PC2: herbal, reduced, fresh/grassy, and salty. Wine B3 and the control labelled “Buntsandstein” (BCtrl) were perceived as being particularly sweet with apple/pear aroma, though the other Buntsandstein wines lie much closer to the centre of the plot. BCtrl and the other controls labelled “Löß,” “Muschelkalk,” and “Schiefer” (LCtrl, MCtrl, and SCtrl, respectively) are well separated from each other and are generally found in close proximity to the other wines from their respective soil types.
Wines from Löß, Muschelkalk, and Schiefer had at least one distinguishing attribute, while those from Buntsandstein always shared groupings with other soil types (Table 3). The Löß wines were rated as the most earthy, whereas Muschelkalk wines were highest in chalky aroma, and lowest in apple/pear aroma and sweet taste. Schiefer wines were clearly perceived as being the most flinty and having the least alcohol and body.
3. Changes between sessions: “Delta” values
According to MANOVA on Δ values, soil was a significant source of variance, while wine nested within soil was not, indicating the shifts in perception between sessions were soil-dependent, but not wine-dependent. Despite the wines differing significantly within each soil type when the sessions are considered separately, results here from Δ analysis suggest perception changed collectively, in a similar manner for all wines from each soil type. It is particularly worth noting that Δ values for the labelled controls did not differ significantly from those of the other wines from their respective soil types.
Most of the Δ values for which soil was a significant source of variance according to ANOVA belong to attributes that were significant in only one of the two sessions. As seen in the PCA biplot (Figure 3), PC1 (52.3 % of variance) is driven mainly by differences between Löß and Muschelkalk Δ values, with Buntsandstein in the middle. Perception of spicy, earthy, and nutty/oxidised tended to increase from session 1 to session 2 for the Löß wines, while fresh/grassy decreased. In the other direction of PC1, chalky increased for Muschelkalk wines, while TDN decreased. Differences between Schiefer wines and the other soil types appear to drive PC2 (41.4 % of variance). In the positive direction, ratings for flinty, sour, and salty increased for Schiefer wines while decreasing for the others, and in the negative direction, alcohol and body decreased for Schiefer wines while increasing for the others.
Differences in Δ values for these attributes among the labelled controls clearly demonstrate how the perception of even a single wine can be altered significantly by the cue provided during sensory evaluation (Figure 4). Perception of fresh/grassy decreased significantly from session 1 to session 2 for the LCtrl, chalky increased for the MCtrl, while sour increased for the SCtrl. Additionally, the SCtrl exhibited a distinct increase in flinty aroma, which decreased significantly for all three of the other controls.
Discussion
1. Baseline differences among soil types
The purpose of our study was to determine the influence that specifying soil types would have on sensory perception, though the results from the uninformed tastings in session 1 warrant discussion. Even without soil type cues, our judges identified what appeared to be soil type-dependent differences among the wines. In particular, wines from Muschelkalk were higher in TDN, nutty/oxidised, spicy, and herbal aromas, as well as sour and bitter taste, while wines from Schiefer were higher in tropical, floral, sweet/sugary aromas, and sweet taste (Table 2, Figure 1). Several previous studies have also observed sensory differences among Riesling wines from different soil types (Löhnertz et al., 2008; Douglas et al., 2001; Bauer et al., 2011), though the exact nature of these differences varies significantly from study to study. For example, in contrast to our findings, Bauer et al. (2011) found wines from slate and sandstone (presumably corresponding to our Schiefer and Buntsandstein, respectively) to be quite similar, exhibiting relatively high levels of minerality, box tree, and grassy/cucumber aromas, while wines from limestone (corresponding to Muschelkalk) were characterised by mango/passion fruit and peach/apricot aromas. Despite trends found within each study, differences among studies are apparent, not only due to different sensory evaluation methodologies but also due to variability in environmental parameters other than soil type. To the best of our knowledge, only one study to date has attempted to control for this variability, by moving Silvaner vines from seven different soil types in lysimeters to a single location and subsequently finding no significant differences in flavour among the wines produced (Wahl, 1988; Bader and Wahl, 1996).
No mechanisms have been established by which particular vineyard soil types could influence the formation of compounds responsible for differences in wine flavour. As described recently by van Leeuwen et al. (2020) in their extensive review of terroir and aroma, the appearance of these compounds in wine is instead largely dependent on other aspects of a physical environment to which soil is inextricably linked. All soil types cannot exist everywhere and are instead necessarily localised to certain places, each with their own climate, topography, etc. For example, water deficit conditions, particularly in vineyards on steep slopes with low water-holding capacity, can increase concentrations of volatile thiols and terpenes (Peyrot des Gaschons et al., 2005; Savoi et al., 2016; van Leeuwen et al., 2020), compounds responsible for the tropical, floral, and sweet/sugary aromas perceived here in the wines from Schiefer. Because this soil type is found almost exclusively in the Mosel and Nahe winegrowing regions of Germany, known for their steep, stony vineyards, it cannot be said whether these aromas should be attributed to the Schiefer soil type or the inherent drainage of these vineyards’ slopes.
Rather than focusing on differences between soil types, the variability among wines within each soil type should be considered, as well as the overlap between them (Figure 1). Having studied Riesling from several vineyard designations in the Rheingau region of Germany, each with homogenous soil, Fischer et al. (1999) observed that vintage and the individual estates producing wine from these vineyards had significantly greater effects on wine flavour than the sites themselves. These findings highlight the multifactorial complexity of terroir, and soil typing can provide only an incomplete understanding of how wine flavour is derived. Nonetheless, the precise nature and magnitude of soil’s contribution remains an open avenue for continued investigation. Future research could set aside soil type, perhaps as nothing more than a given name, and focus instead on measurable properties such as texture, structure, porosity, solum or rooting depth, heat and water retention, etc., which could influence grape development and the sensory properties of the resultant wine.
2. Soil-type cue-dependent changes in perception
More descriptors were used to differentiate among soil types in session 2 (Table 3) than in session 1 (Table 2), several of which are notably often used to express minerality: flinty, chalky, marine, earthy, and salty. On one hand, it is possible the soil type cues during session 2 heightened our judges’ sensitivity to certain flavours, prompting more focused sensory evaluation (Hughson and Boakes, 2001). On the other, however, given the duration of the study and the number of wines evaluated, fatigue might have led judges to become increasingly reliant on semantic or conceptual associations, resulting in “false positives” (Hughson and Boakes, 2002). During their investigation of minerality in Burgundian Chardonnay, Ballester et al. (2013) observed a similar phenomenon: during a free-sorting task, panellists tended not to use minerality-related descriptors, and only did so when explicitly asked to assess minerality. The notion of objectivity in sensory perception remains a topic of continued discussion (Nguyen and Durner, 2023), and investigations into both the chemical and psychosocial/cultural bases of minerality in wine are ongoing (Deneulin et al., 2016; Rodrigues et al., 2017).
Regardless of whether such flavours are a “chemical reality or cultural construct” (Parr et al., 2016), results from session 2 indicate that soil type, provided as an extrinsic informational cue, can influence the perception of German Riesling. Several previous studies on wine tasters’ biases and expectations have similarly compared the results from sensory evaluations with and without the provision of certain information, such as critic scores (Siegrist and Cousin, 2009), price and grape variety (Peterson, 2014), and blend composition (Boncinelli et al., 2016). Expectations arise from previous knowledge or experience with a product, after which extrinsic cues can lead tasters to believe the product will exhibit certain sensory properties (Deliza and MacFie, 1996; Brochet and Morrot, 1999). According to the assimilation-contrast theory, if the discrepancy between expectation and perception is not so large that contrast occurs, assimilation will shift perception to align it with expectations (Anderson, 1973).
Working “backwards” through this model, shifts in perception thus indicate underlying expectations and allow us to infer the knowledge or experience that shaped them. Differences in intensity ratings suggest our judges shared distinct concepts for Rieslings from Buntsandstein, Löß, Muschelkalk, and Schiefer (Table 3, Figure 2). This is further evidenced by the soil type-dependent Δ values calculated between sessions (Figure 3), particularly for the single control wine merely labelled with the different soil types (Figure 4).
Löß (loess) is a silty sediment formed by the accumulation of wind-blown dust (Frechen, 2011), and is considered a rich and fertile soil of great importance to agriculture (Catt, 2001). Knowing this might explain the relatively high ratings judges gave Löß wines for earthy, body, and alcohol, as well as the lower ratings for fresh/grassy and salty. The positive Δ values for spicy and nutty/oxidised might also be attributed to a conceptual understanding of Löß and its properties.
Muschelkalk (shell-bearing limestone) refers to layers of calcareous sediment often containing the fossilised remains of various ocean-dwelling invertebrates (Aigner, 1982; Sommaruga et al., 2017), which would explain the induced perception of certain attributes, particularly chalky and marine. It should be noted the German descriptor kreidig used for this study (translated here as “chalky”) refers to blackboard chalk and not directly to limestone (Kalk), to avoid such an obvious association, which apparently occurred anyway. Explaining the inter-session decrease in perceived TDN among the Muschelkalk wines is less straightforward. Often found in Riesling wine made from grapes exposed to high temperatures and excessive sunlight (Kwasniewski et al., 2010; Szmania et al., 2023), the kerosene- or petrol-like aroma of TDN (1,1,6-trimethyl-1,2-dihydronaphthalene) can be categorised as empyreumatic (Ferreira et al., 2022), and thus perhaps dissonant with a cold, oceanic mental representation of Muschelkalk.
The association of flinty aroma with Schiefer (slate) is rather interesting, given flint and slate are entirely different types of rock. It could be that slate is a particularly popular example of soil type brought up in discussions of minerality, for which flinty is often an associated sensory descriptor. More figurative terms such as “sharp” or “lean” are also commonly used to express minerality (Maltman et al., 2013), reflected here perhaps in the Schiefer wines’ positive Δ values for sour and salty, and negative Δ values for alcohol and body.
Conceptions of Löß, Muschelkalk, and Schiefer wines appear to be well-defined and differentiated, thus these soil types might be considered relatively valuable from a marketing perspective. On the other hand, should Buntsandstein (coloured sandstone) appear on bottle labels, etc., the sensory properties to be expected are less obvious. While results here suggest the lack of a commonly understood “image” for Buntsandstein, with fewer connotations and weaker associations, it is possible that other descriptors not included in our study, or the use of another methodology such as free-choice profiling, would have allowed for better definition of wines from this soil type, and perhaps the other soil types as well. Future studies could be expanded to include other soil types seen on bottles of German Riesling; volcanic soils in particular, such as basalt or rhyolite, would likely elicit interesting expectations. The influence of soil type cues on the perception of wines from other grape varieties and regions is also worth investigating.
3. Moving forward: Consumer expectations
It should be acknowledged that our panel was composed entirely of wine experts, and such distinct expectations and biases during sensory evaluation would likely not have been apparent with nonexpert consumers. In studies of wine typicity, the reference group used to determine the distinguishing characteristics of a given class/category of wines normally comprises key stakeholders, such as wine producers, merchants, and regulatory agents, as they can forgo time-consuming training (Ballester et al., 2008; Barton et al., 2020) and are best capable of reporting expectations (Leriche et al., 2020; Souza Gonzaga et al., 2021). Moreover, central to our investigation was a reliance on experts’ susceptibility to bias, given their cognitive approach to evaluating wine often allows the “top-down” application of prior knowledge to override “bottom-up” sensory perception (Spence and Wang, 2019; Spence, 2020). Nonetheless, it is possible that a wider pool of consumers, despite their lack of expertise, might have their own preconceptions of Rieslings from different soils (Deneulin et al., 2016), thus an additional investigation into the response of consumers to wine soil typing may be warranted.
Consumer expectations can also be managed, perhaps using the findings here as a starting point. It cannot be said whether the effects observed were those intended by the producers when they chose to provide soil type information for their wines, nor whether these changes to sensory perception should be considered positive or negative. Therefore, it remains up to the discretion of industry members how the complexity of terroir might be distilled, and the role of soil highlighted when communicating with consumers, to set the “correct” expectations and ensure the desired flavour profile is not only perceived but appreciated.
Conclusion
In their review of the many factors governing grape and wine quality, Jackson and Lombard (1993) once commented, “Despite its complexity, the concept of terroir is a valuable one. The reductionist might find difficulty in appreciating its value for scientific analysis, but… the effects are real.” There is clearly a tremendous range of flavours possible in wine, as we and many others have observed, even from a single grape variety, and research indeed continues to better understand the source of this variance, more clearly define terroir, and disentangle the effects of its many components. Yet soil remains a particularly potent aspect of terroir, one whose influence seems to reach beyond physical or chemical reality.
Our investigation has revealed that the mere mention of wine soil types as an informational cue during the evaluation of German Riesling can significantly influence sensory perception. Changes in ratings were soil type cue-dependent, suggesting our panel of industry professionals held distinct a priori conceptions for each soil, which differentially shaped their expectations and consequently primed them to perceive wines of each soil type differently. Even the perception of the control wine, a composite of all the other Rieslings in this study, changed significantly depending on which cue accompanied it. Such biasing is presumably contingent on some level of expertise or previous knowledge of the different soils, thus further work is needed to determine whether nonexpert consumers would be similarly influenced. Although perhaps an oversimplification of terroir, specifying soil types can nonetheless be a valuable strategy in wine marketing. For this to be effective, however, a clear “image” of each soil type must first be communicated to consumers to ensure their expectations yield the desired sensory experience.
Acknowledgements
This work was conducted as part of Project PINOT (grant number 28DK107A20), funded by the Bundesanstalt für Landwirtschaft und Ernährung (Federal Office for Agriculture and Food). We thank our expert judges, as well as Gabriele Görgens, Florian Schraut, and sensory staff members Martha Wicks-Müller, Xenia Petermann, and Sandra Klink, without whom this study would not have been possible. We also thank Pascal Wegmann-Herr and Ulrich Fischer for their input regarding the experimental design. We appreciate the wineries that donated or discounted their wines.
References
- Aigner, T. (1982). Calcareous tempestites: Storm-dominated stratification in upper Muschelkalk limestones (Middle Trias, SW-Germany). In G. Einsele, & A. Seilacher (Eds.), Cyclic and Event Stratification (pp. 180-198). Berlin: Springer. https://doi.org/10.1007/978-3-642-75829-4_13
- Anderson, R. E. (1973). Consumer dissatisfaction: The effect of disconfirmed expectancy on perceived product performance. Journal of Marketing Research, 10(1), 38-44. https://doi.org/10.2307/3149407
- Ares, G., Bruzzone, F., Vidal, L., Silva Cadena, R., Giménez, A., Pineau, B., Hunter, D.C., Paisley, A.G., & Jaeger, S.R. (2014). Evaluation of a rating-based variant of check-all-that-apply questions: Rate-all-that-apply (RATA). Food Quality and Preference, 36, 87-95. https://doi.org/10.1016/j.foodqual.2014.03.006
- Bader, W., & Wahl, K. (1996). Der Einfluss des Bodens ist minimal. Der Deutsche Weinbau, 18, 18-19.
- Ballester, J., Mihnea, M., Peyron, D., & Valentin, D. (2013). Exploring minerality of Burgundy Chardonnay wines: A sensory approach with wine experts and trained panellists. Australian Journal of Grape and Wine Research, 19(2), 140-152. https://doi.org/10.1111/ajgw.12024
- Ballester, J., Patris, B., Symoneaux, R., & Valentin, D. (2008). Conceptual vs. perceptual wine spaces: Does expertise matter? Food Quality and Preference, 19(3), 267-276. https://doi.org/10.1016/j.foodqual.2007.08.001
- Barton, A., Hayward, L., Richardson, C. D., & McSweeney, M. B. (2020). Use of different panellists (experienced, trained, consumers and experts) and the projective mapping task to evaluate white wine. Food Quality and Preference, 83, 103900. https://doi.org/10.1016/j.foodqual.2020.103900
- Bauer, A., Wolz, S., Schormann, A., & Fischer, U. (2011). Authentication of different terroirs of German Riesling applying sensory and flavor analysis. In S. E. Ebeler, G. R. Takeoka, & P. Winterhalter (Eds.), Progress in Authentication of Food and Wine (pp. 131-149). Washington, DC: American Chemical Society. https://doi.org/10.1021/bk-2011-1081.ch009
- Bokulich, N. A., Collins, T. S., Masarweh, C., Allen, G., Heymann, H., Ebeler, S. E., & Mills, D. A. (2016). Associations among wine grape microbiome, metabolome, and fermentation behavior suggest microbial contribution to regional wine characteristics. mBio, 7(3), e00631-16. https://doi.org/10.1128/mBio.00631-16
- Boncinelli, F., Casini, L., Contini, C., Gerini, F., & Scozzafava, G. (2016). The consumer loves typicality but prefers the international wine. Agriculture and Agricultural Sciences Procedia, 8, 236-242. https://doi.org/10.1016/j.aaspro.2016.02.098
- Brochet, F., & Morrot, G. (1999). Influence du contexte sur la perception du vin: Implications cognitives et méthodologiques. Journal International des Sciences de la Vigne et du Vin, 33(4), 187-192. https://doi.org/10.20870/oeno-one.1999.33.4.1017
- Capitello, R., Agnoli, L., Charters, S., & Begalli, D. (2021). Labelling environmental and terroir attributes: Young Italian consumers’ wine preferences. Journal of Cleaner Production, 304, 126991. https://doi.org/10.1016/j.jclepro.2021.126991
- Catt, J. A. (2001). The agricultural importance of loess. Earth Science Reviews, 54(1-3), 213-229. https://doi.org/10.1016/S0012-8252(01)00049-6
- de Andrés-de Prado, R., Yuste-Rojas, M., Sort, X., Andrés-Lacueva, C., Torres, M., & Lamuela-Raventós, R. M. (2007). Effect of soil type on wines produced from Vitis vinifera L. cv. Grenache in commercial vineyards. Journal of Agricultural and Food Chemistry, 55(3), 779-786. https://doi.org/10.1021/jf062446q
- Deliza, R., & MacFie, H. J. H. (1996). The generation of sensory expectation by external cues and its effect on sensory perception and hedonic ratings: A review. Journal of Sensory Studies, 11(2), 103-128. https://doi.org/10.1111/j.1745-459X.1996.tb00036.x
- Deneulin, P., le Fur, Y., & Bavaud, F. (2016). Study of the polysemic term of minerality in wine: Segmentation of consumers based on their textual responses to an open-ended survey. Food Research International, 90, 288-297. https://doi.org/10.1016/j.foodres.2016.11.004
- Douglas, D., Cliff, M. A., & Reynolds, A. G. (2001). Canadian terroir: Characterization of Riesling wines from the Niagara Peninsula. Food Research International, 34(7), 559-563. https://doi.org/10.1016/S0963-9969(01)00071-0
- Ferreira, V., de la Fuente, A., & Sáenz-Navajas, M-P. (2022). Wine aroma vectors and sensory attributes. In A. G. Reynolds (Ed.), Managing Wine Quality, Vol. 1: Viticulture and Wine Quality (2nd ed.) (pp. 3-39). Cambridge: Woodhouse Publishing. https://doi.org/10.1016/B978-0-08-102067-8.00008-7
- Fischer, U., Roth, D., & Christmann, M. (1999). The impact of geographic origin, vintage and wine estate on sensory properties of Vitis vinifera cv. Riesling wines. Food Quality and Preference, 10(4-5), 281-288. https://doi.org/10.1016/S0950-3293(99)00008-7
- Frechen, M. (2011). Loess in Europe: Guest editorial. E&G Quaternary Science Journal, 60(1), 3-5. https://doi.org/10.3285/eg.60.1.00
- Heymann, H., Hopfer, H., & Bershaw, D. (2014). An exploration of the perception of minerality in white wines by projective mapping and descriptive analysis. Journal of Sensory Studies, 29(1), 1-13. https://doi.org/10.1111/joss.12076
- Huggett, J. M. (2006). Geology and wine: A review. Proceedings of the Geologists’ Association, 117(2), 239-247. https://doi.org/10.1016/S0016-7878(06)80012-X
- Hughson, A. L., & Boakes, R. A. (2001). Perceptual and cognitive aspects of wine expertise. Australian Journal of Psychology, 53(2), 103-108. https://doi.org/10.1080/00049530108255130
- Hughson, A. L., & Boakes, R. A. (2002). The knowing nose: The role of knowledge in wine expertise. Food Quality and Preference, 13(7-8), 463-472. https://doi.org/10.1016/S0950-3293(02)00051-4
- Jackson, D. I., & Lombard, P. B. (1993). Environmental and management practices affecting grape composition and wine quality – A review. American Journal of Enology and Viticulture, 44(4), 409-430. https://doi.org/10.5344/ajev.1993.44.4.409
- Kwasniewski, M. T., Vanden Heuvel, J. E., Pan, B. S., & Sacks, G. L. (2010). Timing of cluster light environment manipulation during grape development affects C13 norisoprenoid and carotenoid concentrations in Riesling. Journal of Agricultural and Food Chemistry, 58(11), 6841-6849. https://doi.org/10.1021/jf904555p
- Lenglet, F. (2014). Influence of terroir products meaning on consumer’s expectations and likings. Food Quality and Preference, 32(C), 264-270. https://doi.org/10.1016/j.foodqual.2013.09.003
- Leriche, C., Molinier, C., Caillé, S., Razungles, A., Symoneaux, R., & Coulon-Leroy, C. (2020). Development of a methodology to study typicity of PDO wines with professionals of the wine sector. Journal of the Science of Food and Agriculture, 100(10), 3866-3877. https://doi.org/10.1002/jsfa.10428
- Löhnertz, O., Böhn, P, & Muskat, S. (2008). Terroir Hesse – Soil determines wine style. In Proceedings of the VIIth International Terroir Congress. Nyon: International Viticulture & Enology Society. https://ives-openscience.eu/9325/
- Maltman, A. (2013). Minerality in wine: A geological perspective. Journal of Wine Research, 24(3), 169-181. https://doi.org/10.1080/09571264.2013.793176
- Nguyen, T. H., & Durner, D. (2023). Sensory evaluation of wine aroma: Should color-driven descriptors be used? Food Quality and Preference, 107, 104844. https://doi.org/10.1016/j.foodqual.2023.104844
- Noble, A. C. (1979). Evaluation of Chardonnay wines obtained from sites with different soil compositions. American Journal of Enology and Viticulture, 30(3), 214-217. https://doi.org/10.5344/ajev.1979.30.3.214
- Parr, W. V., Maltman, A. J., Easton, S., & Ballester, J. (2018). Minerality in wine: Towards the reality behind the myths. Beverages, 4(4), 77. https://doi.org/10.3390/beverages4040077
- Parr, W. V., Valentin, D., Breitmeyer, J., Peyron, D., Darriet, P., Sherlock, R., Robinson, B., Grose, C., Ballester, J. (2016). Perceived minerality in Sauvignon blanc wine: Chemical reality or cultural construct? Food Research International, 87, 168-179. https://doi.org/10.1016/j.foodres.2016.06.026
- Parr, W. V., Heatherbell, D., White, K. G. (2002). Demystifying wine expertise: Olfactory threshold, perceptual skill and semantic memory in expert and novice wine judges. Chemical Senses, 27(8), 747-755. https://doi.org/10.1093/chemse/27.8.747
- Peterson, K. (2014). The snob effect of red wine: Estimating consumer bias in experimental blind wine tastings. The American Economist, 59(1), 76-89. https://doi.org/10.1177/056943451405900107
- Peyrot des Gachons, C., van Leeuwen, C., Tominaga, T., Soyer, J-P., Gaudillère, J-P., & Dubourdieu, D. (2005). Influence of water and nitrogen deficit on fruit ripening and aroma potential of Vitis vinifera L. cv. Sauvignon blanc in field conditions. Journal of the Science of Food and Agriculture, 85(1), 73-85. https://doi.org/10.1002/jsfa.1919
- Rankine, B. C., Fornachon, J. C. M., Boehm, E. W., & Cellier, K. M. (1971). Influence of grape variety, climate and soil on grape composition and on the composition and quality of table wines. Vitis, 10(1), 33-50. https://doi.org/10.5073/vitis.1971.10.33-50
- Robinson, J. (2015). The Oxford Companion to Wine (4th ed.). Oxford: Oxford University Press. https://doi.org/10.1093/acref/9780198705383.001.0001
- Rodrigues, H., Ballester, J., Sáenz-Navajas, M-P., & Valentin, D. (2015). Structural approach of social representation: Application to the concept of wine minerality in experts and consumers. Food Quality and Preference, 46, 166-172. https://doi.org/10.1016/j.foodqual.2015.07.019
- Rodrigues, H., Sáenz-Navajas, M-P., Franco-Luesma, E., Valentin, D., Fernández-Zurbano, P., Ferreira, V., de la Fuente Blanco, A., & Ballester, J. (2017). Sensory and chemical drivers of wine minerality aroma: An application to Chablis wines. Food Chemistry, 230, 553-562. https://doi.org/10.1016/j.foodchem.2017.03.036
- Savoi, S., Wong, D. C. J., Arapitsas, P., Miculan, M., Bucchetti, B., Peterlunger, E., Fait, A., Mattivi, F., & Castellarin, S. D. (2016). Transcriptome and metabolite profiling reveals that prolonged drought modulates the phenylpropanoid and terpenoid pathway in white grapes (Vitis vinifera L.) BMC Plant Biology, 16, 67. https://doi.org/10.1186/s12870-016-0760-1
- Schüttler, A., Friedel, M., Jung, R., Rauhut, D., & Darriet, P. (2015). Characterizing aromatic typicality of Riesling wines: Merging volatile compositional and sensory aspects. Food Research International, 69, 26-37. https://doi.org/10.1016/j.foodres.2014.12.010
- Seguin, G. (1986). ‘Terroirs’ and pedology of wine growing. Experientia, 42, 861-873. https://doi.org/10.1007/BF01941763
- Siegrist, M., & Cousin, M-E. (2009). Expectations influence sensory experience in a wine tasting. Appetite, 52(3), 762-765. https://doi.org/10.1016/j.appet.2009.02.002
- Sommaruga, A., Mosar, J., Schori, M., & Gruber, M. (2017). The role of the Triassic evaporites underneath the north Alpine foreland. In J. I. Soto, J. F. Flinch, & G. Tari (Eds.), Permo-Triassic Salt Provinces of Europe, North Africa and the Atlantic Margins: Tectonics and Hydrocarbon Potential (pp. 447-466). Amsterdam: Elsevier. https://doi.org/10.1016/B978-0-12-809417-4.00021-5
- Souza Gonzaga, L., Capone, D. L., Bastian, S. E. P., & Jeffery, D. W. (2021). Defining wine typicity: Sensory characterisation and consumer perspectives. Australian Journal of Grape and Wine Research, 27(2), 246-256. https://doi.org/10.1111/ajgw.12474
- Spence, C. (2020). Wine psychology: Basic and applied. Cognitive Research: Principles and Implications, 5, 22. https://doi.org/10.1186/s41235-020-00225-6
- Spence, C., & Wang, Q. J. (2019). Wine expertise: Perceptual learning in the chemical senses. Current Opinion in Food Science, 27, 49-56. https://doi.org/10.1016/j.cofs.2019.05.003
- Szmania, C., Waber, J., Bogs, J., & Fischer, U. (2023). Sensory and aroma impact of mitigation strategies against sunburn in Riesling. OENO One, 57(3), 127-140. https://doi.org/10.20870/oeno-one.2023.57.3.7287
- van Ittersum, K., Candel, M. J. J. M., & Meulenberg, M. T. G. (2003). The influence of the image of a product’s region of origin on product evaluation. Journal of Business Research, 56(3), 215-226. https://doi.org/10.1016/S0148-2963(01)00223-5
- van Leeuwen, C. (2022). Terroir: The effect of the physical environment on vine growth, grape ripening, and wine sensory attributes. In A. G. Reynolds (Ed.), Managing Wine Quality, Vol. 1: Viticulture and Wine Quality (2nd ed.) (pp. 341-393). Cambridge: Woodhouse Publishing. https://doi.org/10.1016/B978-0-08-102067-8.00005-1
- van Leeuwen, C., Barbe, J-C., Darriet, P., Geffroy, O., Gomès, E., Guillaumie, S., Helwi, P., Laboyrie, J., Lytra, G., Le Menn, N., Marchand, S., Picard, M., Pons, A., Schüttler, A., & Thibon, C. (2020). Recent advancements in understanding the terroir effect on aromas in grapes and wine. OENO One, 54(4), 985-1006. https://doi.org/10.20870/oeno-one.2020.54.4.3983
- van Leeuwen, C., & Seguin, G. (2006). The concept of terroir in viticulture. Journal of Wine Research, 17(1), 1-10. https://doi.org/10.1080/09571260600633135
- Wahl, K. (1988). Climate and soil effects on grapevine and wine: Situation on the northern border of viticulture – The example of Franconia. In R. E. Smart, R. J. Thornton, S. B. Rodriguez, & J. E. Young (Eds.), Proceedings of the 2nd International Symposium of Cool Climate Viticulture and Oenology (pp. 1-5). Auckland: New Zealand Society for Viticulture and Oenology.
- Willwerth, J. J., & Reynolds, A. G. (2020). Spatial variability in Ontario Riesling vineyards. II. Berry composition. Canadian Journal of Plant Science, 100(5), 504-527. https://doi.org/10.1139/cjps-2019-0291
Attachments
Supplementary data
Download