Scientific Basis of Sherry Wine Food Pairing: Integrating Consumer Perception, Sensory Analysis, and Volatile Profiling

Abstract

Sherry wines from the Jerez–Xérès–Sherry PDO are globally renowned for their unique sensory complexity, yet their food pairing guidelines have traditionally remained empirical. This study aims to establish a scientific foundation for these harmonies by integrating consumer perception, descriptive sensory analysis, and volatile compound profiling. Three Sherry styles (Fino, Oloroso, and Pedro Ximénez) were evaluated against fifteen food matrices using consumer tastings (Likert and Linking tests via the SensoGraph application), an expert consensus group, and GC–MS techniques (SBSE and HSSE). Results indicate that Fino exhibits a strong affinity for saline and umami profiles, while Oloroso harmonizes best with umami taste and substantial body. Conversely, Pedro Ximénez is driven by sweetness and persistence, with saltiness acting as a primary antagonist. At the molecular level, the research suggested the existence of potential chemical correspondences, such as the furan family in the PX–chocolate pairing. However, findings suggest that pairing success depends more on the perceptual relevance and odor potency of shared molecules than on the absolute count of common compounds. This study provides certain evidence that pairing quality is determined by specific sensory and chemical interactions, transitioning Sherry wine gastronomy from intuitive tradition toward a systematic, indicative model.

1. Introduction

Wines covered under the Jerez–Xérès–Sherry PDO represent one of the most unique and complex oenological typologies in the world. However, their initial consumption without prior knowledge usually becomes an unexpected and, at times, disconcerting experience for the consumer, due to the high sensory intensity that characterizes these wines. As Saldaña points out [1], the surprise lies in the dichotomy of their profiles: a wine labeled as “seco” (dry) can exhibit extreme austerity, while sweet wines reach densities and sugar concentrations far exceeding traditional standards. This aromatic and structural complexity often generates confusion, since attributes such as aging, oxidative nuances, or alcoholic potency could be interpreted as defects if evaluated under conventional wine criteria. The determining factor of this identity is the production process and, fundamentally, the type of aging applied. Wines such as Fino start from the Palomino grape and can either be fortified up to 15% vol., or, after the latest modification of the product specification, they can also reach this alcoholic strength after fermentation naturally due to the climate change, without fortification being technically necessary [2]. This alcoholic content allows for the development of the velo de flor. This natural layer of yeasts (Saccharomyces) protects the wine from oxidation together with its active metabolic contributions (ethanol oxidation to acetaldehyde, glycerol consumption, synthesis of specific volatiles), and therefore grants it unique sensory characteristics: fresh, floral, and saline aromas, with notes of bakery, green apple, and toasted nuts like almond [3]. Conversely, Oloroso undergoes an exclusively oxidative aging by being fortified up to 17–18% vol., which prevents the survival of the velo de flor and results in a dark, full-bodied wine with intense notes of wood, spices, and walnut. On the extreme side of sweetness, Pedro Ximénez (PX) is produced by sun-drying (“asoleo”) the grapes to concentrate sugars (reaching levels above 212 g/L), resulting in unctuous textures with hints of raisins, figs, and coffee [4]. This heterogeneity makes Sherry wines an exceptional field of study for the science of pairing.

Pairing is defined as the gastronomic union of wine and food in the mouth, with the objective of producing a synergistic effect, provoking the emergence of new sensations. Etymologically derived from the French mariage (marriage), its application was dominated for decades by rigid and traditional rules that imposed strict norms, such as the prohibition of pairing red wine with fish. Nevertheless, authors such as Cepeda et al. [5] advocate for a more open vision, where even sparkling wines can accompany a meal from start to finish or full-bodied wines can be served chilled to balance complex dishes. Currently, two fundamental strategies are differentiated: affinity pairing and contrast pairing. The former seeks harmony through similarity, aiming for the wine and food to share common sensations and aromatic descriptors to create an experience of continuity. On the contrary, contrast pairing seeks balance between opposite poles, being crucial to compensate for dominant flavors that could saturate the palate, providing cleanliness and surprise. Harrington [6] introduced a mathematical and sensory hierarchy based on the levels of affinity between the components of the dish and the beverage. His theory classifies elements into basic flavor components (such as acidity, sweetness, and saltiness), textures (the fat of the food versus the tannins or alcohol of the wine), and aromas, creating an analytical map that precisely predicts whether a combination will result in a contrast, complementary, or synergistic pairing. Ultimately, the best pairing is the one that most pleases the consumer, prioritizing their personal tastes and preferences over the cost or prestige of the product.

The contemporary study of pairing has transcended the empirical thanks to François Chartier’s Molecular Gastronomy Theory [7], which bases gastronomic affinity on the existence of shared aromatic molecules between wine and food. According to this model, when two products share key compounds, potential chemical correspondences are established, which the consumer perceives as a natural harmony. For example, although cocoa aroma comes from the interplay of Strecker aldehydes, esters, short-chain fatty acids, furans and pyrazines, the latter are very important compounds in the roasted profile of chocolate [8]. Meanwhile, methyl ketones such as 2-heptanone are markers for blue cheese, although they are part of a broader balance involving other methyl ketones (C5–C9), secondary alcohols, and free fatty acids [9]. Therefore, these foods could pair better with wines that exhibit these types of molecules in their volatile fraction. However, results from recent research suggest that pairing quality does not depend solely on the gross number of molecular matches, but rather on their actual sensory impact, which is conditioned by detection thresholds. As demonstrated by Francis and Newton [10], compounds with extremely low thresholds for hydroalcoholic matrix (10% v/v), such as ethyl octanoate (0.005 mg/L), have a much higher perceptual relevance than others with high thresholds, such as diethyl succinate (200 mg/L) [11]. Therefore, pairings with numerous common compounds may fail if those elements are perceptually irrelevant or if flavor attributes (sweetness vs. saltiness) generate insurmountable dissonances. A traditional example of success can be the Oloroso wine-blue cheese binomial, where the coincidence in fatty acids and aromatic alcohols creates a robust profile that matches the intensity of both [4,12].

Sherry wines present specific sensory profiles that complicate the establishment of standardized pairing guidelines used for red, white, or sparkling wines. Some of the volatile compounds that act as fingerprints in this type of wine can be acetaldehyde, which is the marker for the biological aging of Fino, whereas furfural and benzaldehyde reflect the oxidative and almond-like character of Oloroso and PX [4,13]. Traditionally, harmonization proposals for these wines have been based on empirical criteria or customs, with a lack of research addressing pairing in a scientific and systematic manner. This lack of technical support limits the understanding of sensory interactions and the predictability of gastronomic results. The present work is proposed to systematize these relationships through a technical approach that analyses consumer preference, the sensory evaluation of an expert consensus group, and the identification of volatile compounds using SBSE/HSSE–GC–MS. Through the analysis of shared molecules and the sensory profiles of wine and dish, the goal is to replace intuitive criteria with an objective theoretical basis that explains the harmony between the different styles of Sherry wine and the food matrices analyzed. To address this, we hypothesize that the pairing success between Sherry wines and diverse food matrices is governed by a dual-level interaction, where the perceptual relevance of shared volatile compounds, rather than their absolute count, and the balance of core taste and structural attributes determine sensory harmony and consumer acceptance.

2. Materials and Methods

2.1. Samples

Three Sherry wines were studied: Fino, Oloroso, and Pedro Ximénez (PX), all of them provided by the Consejo Regulador de la Denominación de Origen Jerez–Xérès–Sherry.

Table 1. Technical parameters of the wines studied (Fino, Oloroso and PX).

Parameter	Fino	Oloroso	PX
Alcohol by volume (ABV)	15%	18%	16%
Residual sugar	<1 g/L	<1 g/L	350 g/L
Total acidity (Tartaric acid)	4 g/L	5 g/L	4 g/L
Volatile acidity (Acetic acid)	<0.25 g/L	<0.8 g/L	<0.8 g/L
Average aging period (years)	3 years	6 years	4 years

shows several technical parameters of the wines studied.

The 15 dishes for the preliminary consumer study, selected for their diversity in sensory profiles, were: boiled prawns, steak tartar, 70% dark chocolate, intense blue cheese, papas aliñás (seasoned potatoes), stewed red meat, pickled gherkins, lemon sorbet with ginger, vanilla custard, foie gras toasts with mango chutney, smoked salmon, salmorejo (a cold, thick Spanish soup made from puréed tomatoes, bread, garlic, and olive oil), grilled red meat, cured Manchego cheese, and Manzanilla olives. All of them were purchased at a local market.

The papas aliñás were prepared in-house according to the following formulation: per 1 kg of boiled potatoes, half an onion, 100 g of canned tuna in olive oil, 100 mL of extra virgin olive oil, 50 mL of white wine vinegar, 10 g of salt, and 5 g of chopped parsley were used. The ingredient list of the remaining processed food samples, ordered from highest to lowest proportion in accordance with current regulations, is detailed as follows: salmorejo (tomato, wheat bread, 5% extra virgin olive oil, sunflower oil, garlic, salt, and vinegar); foie toast (duck foie gras, water, brandy, salt, sugar, white pepper, antioxidant E-301, and preservative E-250) and mango chutney (sugar, 41% mango, salt, acetic acid, spices, cumin seeds, ground dried chili, and paprika extract); and lemon sorbet with ginger (non-caloric sweeteners, 9.5% lemon juice, date juice concentrate, 1% ginger, and stabilizers).

2.2. Consumer Tasting

To establish the pairings, a balanced incomplete block design (BIBD) was used [14]. For each wine, each consumer evaluated three dishes (k = 3) out of the 15 included in the study (t = 15). The design generated 105 blocks (b = 105), in which each pairing was replicated 21 times (r = 21) and each pair of pairings was evaluated together three times (l = 3).

The tastings were carried out over 2 days, involving a total of 105 consumers. An exclusion questionnaire was applied to identify and filter out individuals with extreme aversions or marked preferences/neophilias toward any of the evaluated foods or wines, thereby ensuring a homogeneous panel free of pre-existing individual preference biases. Consumers had to evaluate nine pairings (three wines × three dishes). They were asked to taste each wine and dish combination simultaneously and to perform two tests. Using the Likert test, they marked their hedonic impression after each pairing on a category scale [15] from 1 (Dislike extremely) to 7 (Like extremely). On the other hand, in the Linking test they had to draw a line connecting each wine–dish pair that, in their opinion, paired correctly. This was done using the webapp application “SensoGraph” [16], which collects and processes qualitative dichotomous data (yes/no pairing).

Wines were served at a temperature of 18 °C. Room temperature was selected as the serving temperature for most food items, which was maintained at 23 °C in the tasting room. The lemon sorbet with ginger was served immediately upon removal from the freezer (−4 °C), whereas the grilled meat and the stewed meat were reheated in a microwave oven to 60 ± 5 °C.

Portions of 30 mL of each wine were served in standard tasting glasses, which were intended to be paired with the three assigned dishes. Serving sizes for each food item were standardized as follows: an amount equivalent to three bites for steak tartar, 70% dark chocolate, intense blue cheese, papas aliñás, stewed red meat, smoked salmon, grilled red meat, and cured Manchego cheese; three units for boiled prawns, pickled gherkins, foie gras mini-toasts with mango chutney, and Manzanilla olives; and 50 mL for semi-solid samples, including lemon sorbet with ginger, vanilla custard, and salmorejo.

Food and wine samples were labeled with their respective names. Plates were presented following the randomized balanced incomplete block design (BIBD) sequence. Each dish was evaluated sequentially: first paired with Fino wine, followed by Oloroso, and finally with Pedro Ximénez (PX). Participants were instructed to rinse their mouths with a sip of mineral water and rest for two minutes between consecutive pairing evaluations.

2.3. Descriptive Sensory Analysis

For each wine, the dishes with the highest and lowest pairing scores in the consumer tasting were considered of interest. In total, eight dishes were selected: Manzanilla olive, grilled meat, dark chocolate, boiled prawn, pickled gherkin, blue cheese, lemon sorbet with ginger, and foie gras toast with mango chutney.

In the first session, the experts evaluated by duplicate each wine and each dish separately across the following descriptors: sweetness, acidity, bitterness, saltiness, and umami as basic flavors; aromatic intensity, persistence, and spicy notes in the nose; regarding texture, the fatty character and body of the dishes, and alcohol and body in the wines were rated; and finally, they provided an estimate of the complexity of both wines and dishes. The descriptors used for the wines were an adaptation of those proposed by Harrington [6] to the specific characteristics of Sherry wines. Regarding the dishes, Harrington’s sensory characterization sheet was likewise followed [6]; as it includes generally familiar terms, it was easily adopted by the judges without requiring specific training. In this regard, it should be clarified that the descriptor complexity was previously defined as the quantity of perceived stimuli. Interval scales ranging from 0 to 10 points were used.

In a second session, they rated the pairing of each wine with each dish by duplicate on a scale from 0 to 10.

The expert consensus group consisted of a select group of evaluators with extensive experience in wine and agrifood product tasting. The tastings were conducted individually in a standardized room [17] which, due to its characteristics, minimizes the influence of external factors on the concentration and judgment of the tasters.

2.4. Analysis of Volatile Compounds

The volatile compound profiles of the eight dishes selected after the consumer tasting, as well as the profiles of the Fino, Oloroso, and PX wines, were extracted in duplicate and later analyzed using GC–MS. For the extraction of volatiles from the wines, the stir bar sorptive extraction technique in immersion mode (SBSE–GC–MS) was used, employing 25 mL of sample under stirring at 1250 rpm for 2 h [18], whereas for the gastronomic products, the device involved headspace extraction (HSSE–GC–MS) using 4 g of the product at 40 °C for 1 h. Analytes extracted onto the bars were thermally desorbed and analyzed using an Agilent 6890 GC coupled with a 5973N MS system (Agilent Technologies, Santa Clara, CA, USA). Separation was performed on a DB-Wax capillary column (60 m × 0.25 mm i.d. × 0.25 μm film thickness; J&W Scientific, Folsom, CA, USA) [19]. Compounds were identified by matching their mass spectra (>90% similarity) against a reference library (Wiley Registry of Mass Spectral Data, 7th Edition). When available, retention times of authentic standards were also used for confirmation. All volatile compounds were semi-quantified by calculating the base peak area ratio of each compound’s quantifier ion relative to that of the internal standard.

2.5. Statistical Data Treatment

The balanced incomplete block design (BIBD) was generated using R software version 4.5.2 (R Core Team, Vienna, Austria, 2024).

Both the collection and processing of the Linking test data were carried out using the “SensoGraph” software [16]. The procedure by Lahne et al. [20] was followed to obtain a dissimilarity matrix from the graph of each participant, where the dissimilarity between samples i, j is defined as 0 if i = j or they are directly linked and is otherwise defined as 1 − 1/d(i,j), where d(i,j) is the graph distance between those two samples.

Sensory evaluation data (Likert scale) were analyzed using a linear mixed model (LMM) fitted by restricted maximum likelihood (RML), accounting for the balanced incomplete block design structure. To independently assess the impact of the main factors, models were specified with dish type and wine type as fixed effects. Likewise, the pairing (wine × dish interaction) was included as a fixed factor. In all model runs, consumer was defined as a random factor to isolate variance arising from individual differences and to account for dependencies among measurements. The statistical procedure and the bar chart were performed using Statistica 7.0 software (StatSoft, Inc., Tulsa, OK, USA).

Statistica 7.0 software was also used to perform analysis of variance of each volatile compound as a function of pairing quality with each wine.

PLS regression was performed using XLStat Standard software (Lumivero, version 2026.1.0), employing leave-one-out cross-validation to evaluate model performance.

3. Results and Discussion

3.1. Consumer Study

Of the 105 consumers, 60.2% were male and 39.8% were female. Participant ages ranged from 20 to 78 years, with 27.4% being under 30 years old, 15.0% over 55 years old, and 57.5% aged between 30 and 55 years. To participate in the study, consumers were required to be at least occasional Sherry wine drinkers, a criterion met by 55.4% of the participants. Among the remaining participants, 24.6% reported consuming Sherry wine occasionally during the month, 13.8% several days a week, and 6.2% daily. Prior to the sensory sessions, participants completed a screening questionnaire designed to collect information on specific food aversions or strong individual preferences (neophilias), ensuring that candidates with extreme pre-existing biases that could alter, skew, or confound the pairing evaluations were excluded. Additionally, candidates reporting food allergies, medical conditions incompatible with alcohol consumption, or pregnancy were excluded from participation.

The purpose of this tasting session was to evaluate consumer perception through two distinct methodological approaches: subjective hedonic liking (evaluated via a Likert scale) and conceptual pairing appropriateness or degree of fit (evaluated via the Linking method). Identifying which combinations generated the best and worst evaluations across both dimensions was crucial, as these specific pairings would then be selected for further investigation with the expert consensus group to uncover the underlying drivers of pairing success.

By utilizing the “SensoGraph” application, the results of the Linking test (focusing on perceived pairing appropriateness and harmony) were obtained and processed (Figure 1). The preferred pairings were: PX wine with chocolate, vanilla custard, blue cheese, foie gras toast with mango chutney, and lemon sorbet; Oloroso with blue cheese, grilled meat, foie gras toast with mango chutney, boiled prawns, and stewed meat; and Fino with boiled prawns, seasoned potatoes, grilled meat, salmorejo, and smoked salmon. Conversely, the lowest-rated pairings were: PX with olives, salmon, grilled meat, and boiled prawns; Oloroso with lemon sorbet, pickled gherkins, and olives; and Fino with pickled gherkins, chocolate, and vanilla custard.

Regarding the Likert scale test (measuring individual hedonic acceptance), the comprehensive dataset is displayed in Figure 2. While acceptance patterns for Fino and Oloroso wines were highly similar, the pattern for PX was markedly distinct. Model analysis confirmed that the random effect of the consumer is highly significant (p < 0.001), fully justifying the adoption of the mixed model. The wine factor (p = 0.696) was not significant. Regarding the dish factor (p < 0.001), pairings with pickled gherkins received the lowest ratings, while those with foie gras toast with mango chutney were the most appreciated; a significant difference was found between these two, but not when compared to most of the remaining dishes. As expected, the wine x dish interaction (the pairing itself) was confirmed to significantly influence (p < 0.001) the consumers’ hedonic evaluation. Therefore, the highest-rated pairings in the Likert test were: PX wine with chocolate, vanilla custard, foie gras toast, Manchego cheese, and blue cheese; Oloroso with grilled meat, foie gras toast, boiled prawns, and blue cheese; and Fino with grilled meat, boiled prawns, seasoned potatoes, and olives. The lowest-rated pairings included: PX wine with olives, vanilla custard, pickled gherkins, and lemon sorbet; Oloroso with pickled gherkins, lemon sorbet, vanilla custard, and chocolate; and Fino with chocolate, vanilla custard, pickled gherkins, and lemon sorbet.

As can be observed, the outcomes from both methodologies (hedonic acceptance and conceptual appropriateness) are highly consistent. However, the Linking method proved simpler to implement and required less data processing, resulting in a faster response time.

Based on these findings, our objective was to maximize the signal-to-noise ratio in order to clearly identify the key chemical markers (volatiles) and sensory descriptors that drive strong harmony or disharmony in Sherry wine pairings, avoiding the confounding noise typically associated with intermediate or neutral samples. To achieve this, the dishes representing both the highest- and lowest-rated pairings (across both hedonic and appropriateness criteria) for each wine type were selected for the subsequent phase of the study: Manzanilla olives, grilled meat, dark chocolate, foie gras toast with mango chutney, boiled prawns, pickled gherkins, blue cheese, and lemon sorbet. These selected foods were then analyzed via gas chromatography and subjected to sensory analysis by the expert consensus group.

3.2. Analytical Sensory Evaluation by an Expert Consensus Group

Table 2. Sensory scores assigned by the expert consensus group to the eight evaluated dishes using 0–10 scales. Data are presented as mean ± standard deviation.

Dish	Basic Tastes					Texture		Aroma			Complexity
	Sweetness	Acidity	Bitterness	Saltiness	Umami	Fatness/Oiliness	Body	Intensity	Persistence	Spicy
Manzanilla olives	0.0 ± 0.0	2.6 ± 0.6	6.7 ± 1.1	7.6 ± 0.4	1.4 ± 2.1	5.1 ± 0.5	5.9 ± 1.2	6.5 ± 0.6	5.4 ± 1.8	1.5 ± 0.6	5.0 ± 0.6
Grilled meat	1.3 ± 0.9	1.1 ± 1.1	0.8 ± 1.1	3.6 ± 0.6	5.8 ± 2.3	2.8 ± 1.4	6.0 ± 1.0	6.9 ± 1.1	5.8 ± 1.4	3.1 ± 1.4	6.4 ± 1.4
Chocolate	6.3 ± 1.7	1.3 ± 0.9	5.1 ± 1.3	1.1 ± 1.1	0.9 ± 1.4	5.0 ± 0.6	7.8 ± 1.6	7.1 ± 1.3	8.7 ± 0.9	3.6 ± 1.1	5.6 ± 1.6
Foie gras–chutney	5.3 ± 1.7	5.1 ± 1.3	1.1 ± 0.9	4.7 ± 1.1	4.7 ± 2.0	8.1 ± 1.1	8.1 ± 0.6	8.0 ± 0.9	8.4 ± 0.6	8.3 ± 0.6	8.2 ± 0.6
Boiled prawns	3.6 ± 2.1	1.6 ± 1.3	3.1 ± 1.4	7.6 ± 0.9	4.0 ± 2.6	1.1 ± 0.9	3.1 ± 1.1	6.2 ± 0.6	6.9 ± 0.9	0.0 ± 0.0	4.9 ± 1.0
Pickled gherkins	3.0 ± 0.6	7.2 ± 1.4	3.3 ± 1.3	6.9 ± 1.1	1.7 ± 3.3	2.2 ± 0.0	3.3 ± 1.3	6.4 ± 1.9	6.1 ± 0.6	3.1 ± 0.6	4.2 ± 1.1
Blue cheese	1.1 ± 1.1	6.2 ± 2.9	1.9 ± 1.1	6.9 ± 0.6	4.4 ± 4.1	9.1 ± 0.9	8.7 ± 1.2	9.1 ± 0.9	10.0 ± 0.0	4.4 ± 2.8	8.0 ± 0.9
Lemon sorbet	5.3 ± 1.4	6.0 ± 1.5	0.6 ± 0.6	0.4 ± 0.6	1.1 ± 1.6	1.1 ± 0.9	2.0 ± 0.9	5.8 ± 1.6	5.1 ± 1.0	0.9 ± 1.2	4.9 ± 1.3

summarizes the group’s evaluations of the dishes selected in the previous phase of the study across a series of taste attributes (sweetness, acidity, bitterness, saltiness, umami), aromatic attributes (intensity, persistence, and spiciness), and texture attributes (fatty character and body). As a measure of reproducibility, standard deviations were calculated for both intra-taster repeatability (repeated measurements) and inter-taster reproducibility (panel). In both cases, the deviations were generally low, demonstrating an acceptable level of agreement.

Similarly, the group evaluated the three wines (Fino, Oloroso, PX) based on a set of sensory attributes ranging from basic tastes (sweetness, acidity, bitterness, saltiness, umami) to more complex dimensions related to their structure (alcohol, body) and aromatic expressiveness (intensity, persistence, and spiciness), as shown in

Table 3. Sensory scores assigned by the expert consensus group to the three evaluated wines using 0–10 scales. Data are presented as mean ± standard deviation.

Wine	Basic Tastes					Texture		Aroma			Complexity
	Sweetness	Acidity	Bitterness	Saltiness	Umami	Alcohol	Body	Intensity	Persistence	Spicy
Fino	0.0 ± 0.0	5.6 ± 0.9	5.6 ± 1.4	6.4 ± 1.2	3.3 ± 2.5	7.1 ± 0.6	3.6 ± 0.9	5.8 ± 0.9	5.8 ± 1.2	0.8 ± 0.6	6.7 ± 1.8
Oloroso	0.2 ± 0.5	4.7 ± 0.9	2.7 ± 2.2	3.8 ± 3.0	1.6 ± 2.9	8.0 ± 0.9	5.8 ± 2.0	8.9 ± 0.0	8.4 ± 1.3	5.6 ± 2.4	8.9 ± 0.8
PX	9.8 ± 0.5	4.4 ± 1.8	3.3 ± 2.6	2.0 ± 1.2	2.0 ± 2.1	7.1 ± 2.0	9.6 ± 1.0	8.2 ± 1.3	8.7 ± 0.9	5.3 ± 2.9	7.8 ± 1.8

. These data systematically characterize the sensory profile of each food and wine, serving as the foundation for the subsequent pairing analysis.

Furthermore, the tasters provided an overall score for each of the pairings (

Table 4. Overall scores for the 24 pairings (eight dishes × three wines) evaluated by the expert consensus group on a 0–10 scale. Data are presented as mean ± standard deviation. Scores are classified as good, medium or bad.

Wine	Dish	Pairing Score
Fino	Boiled prawns	9.6 ± 0.8
	Grilled meat	9.3 ± 0.9
	Foie gras with mango chutney	8.9 ± 0.8
	Manzanilla olives	8.1 ± 1.5
	Pickled gherkins	5.8 ± 0.8
	Blue cheese	4.2 ± 0.7
	Lemon sorbet with ginger	3.9 ± 1.5
	Chocolate	0.9 ± 0.7
Oloroso	Boiled prawns	8.6 ± 0.8
	Grilled meat	8.3 ± 1.5
	Foie gras with mango chutney	7.8 ± 1.2
	Blue cheese	7.6 ± 0.7
	Chocolate	6.4 ± 1.1
	Manzanilla olives	6.3 ± 0.5
	Pickled gherkins	3.7 ± 1.2
	Lemon sorbet with ginger	2.2 ± 1.0
PX	Chocolate	9.8 ± 0.4
	Foie gras with mango chutney	8.3 ± 0.4
	Lemon sorbet with ginger	8.1 ± 0.8
	Blue cheese	7.5 ± 1.1
	Grilled meat	1.8 ± 2.2
	Boiled prawns	1.1 ± 1.1
	Manzanilla olives	0.3 ± 0.4
	Pickled gherkins	0.0 ± 0.0

). A Partial Least Squares (PLS) regression analysis was applied for each wine, using the sensory descriptors of the dishes as predictor variables. The data for the fitted models are presented in

Table 5. PLS regression results between the pairings for each wine and the sensory profiles of the dishes. VIP (Variable Importance in Projection) values > 1 indicate significant variables.

Attribute	Fino Wine			Oloroso Wine			PX Wine
	Model Coefficients	VIP Scores	X-Y Correlation	Model Coefficients	VIP Scores	X-Y Correlation	Model Coefficients	VIP Scores	X-Y Correlation
Intercept	11.826			1.130			−4.927
Sweetness	0.102	1.346	−0.435	0.107	0.748	−0.279	0.471	1.346	0.655
Acidity	−0.356	0.629	−0.224	−0.508	1.557	−0.593	0.085	0.315	0.155
Bitterness	0.014	0.714	−0.167	0.077	0.359	0.024	−0.306	0.659	−0.302
Saltiness	0.668	1.572	0.557	0.277	0.854	0.401	−0.729	1.441	−0.673
Umami	0.857	1.762	0.634	0.543	1.352	0.753	−0.592	0.774	−0.100
Fatness/Oiliness	−0.002	0.550	−0.129	−0.031	0.944	0.403	0.304	0.933	0.473
Body	−0.194	0.537	−0.153	0.073	1.022	0.575	0.113	0.913	0.460
Intensity	−1.018	0.604	−0.124	0.043	0.997	0.495	0.649	0.924	0.452
Persistence	−0.857	1.022	−0.345	0.286	0.876	0.469	0.617	1.076	0.596
Spicy	0.541	0.678	0.008	−0.002	0.795	0.272	−0.448	1.094	0.479
Complexity	0.387	0.498	0.093	0.184	1.002	0.550	0.746	1.032	0.536
R2	0.965			0.997			0.980
Standard deviation	3.134			2.286			4.155
RMSE	0.551			0.113			0.547

.

The results of the fitted models reflect an excellent descriptive capacity for all three wine types, with remarkably high R² values (ranging from 0.965 to 0.997). This indicates that the sensory descriptors of the dishes explain nearly the entirety of the pairing quality. Likewise, the Root Mean Square Error (RMSE) values, which are lower than the standard deviations of the dependent variable “Overall Pairing Score,” confirm the precision of the model in describing the observations. However, rather than the quantitative models themselves, the value of these results lies in interpreting the role of the distinct dish descriptors in their pairing with each wine.

For Fino wine, its minimal residual sugar content (<1 g/L) and clean organic acid profile (volatile acidity < 0.25 g/L) explain its strong positive association with saltiness (VIP = 1.572, r = 0.557) and umami (VIP = 1.762, r = 0.634). This reflects a classic taste enhancement mechanism, where the dry, saline character of biological aging synergizes with umami matrices. Conversely, the complete absence of residual sugar illustrates the severe taste contrast observed with sweet dishes (VIP = 1.346, r = −0.435), which strip the wine of its delicate fruitiness. Furthermore, excessive persistence of the dish could unbalance the combination (perhaps due to masking the Fino wine) as indicated by its negative correlation.

In the case of Oloroso wine, the driver for pairing success is heavily rooted in umami (VIP = 1.352, r = 0.753) and body (VIP = 1.022, r = 0.575). This is physically driven by texture–alcohol lipid cleavage: the elevated ethanol content (18% ABV, the highest among the studied wines) and extended oxidative aging (6 years) effectively dissolve the lipid coating of protein-rich foods (e.g., grilled meat or foie gras), clearing the palate while its glyceric body balances heavy textures. On the other hand, the penalty imposed by dish acidity (VIP = 1.557, r = −0.593) highlights a taste imbalance, as low-acid oxidative wines fail to buffer highly acidic ingredients (e.g., pickled gherkins), resulting in organoleptic discordance.

Finally, Pedro Ximénez (PX) pairings showcase both trigeminal and cognitive pairing mechanisms. The positive influence of sweetness (VIP = 1.346, r = 0.655) and spicy character (VIP = 1.094, r = 0.479) points to trigeminal suppression, where the high viscosity and sugar concentration of PX coat the oral mucosa, dampening the pungency of spices while providing a harmonious mouthfeel. Furthermore, the strong acceptance of PX with desserts or dark chocolate likely incorporates cultural and cognitive expectations of sweet-with-sweet pairings, whereas its antagonism toward saltiness (VIP = 1.441, r = −0.673) underscores the physical limits of high-viscosity sweet wines when contrasting with sharp saline matrices.

3.3. Characterization of Volatile Compounds in Food and Wine by GC–MS

The Supplementary Material (Tables S1 and S2) shows the mean relative areas of the main compounds identified in the food samples and analyzed wines. It is important to note that, as different extraction techniques involving distinct partition coefficients and extraction dynamics are used for food and wine samples, the relative areas obtained in both cases are not strictly comparable between each other. However, they are fully comparable within all food samples on the one hand and wine samples on the other. Regardless, they can provide an indication of the most abundant components that best characterize each food and wine sample.

3.3.1. Analysis of Food Matrices and Culinary Preparations

Chocolate: The chocolate profile was characterized by its alignment with classic roasting markers. Fundamental pyrazines, such as 2-ethyl-5-methylpyrazine and 2,3,5,6-tetramethylpyrazine, which are responsible for the roasted and cocoa aromatic notes, were identified. Their presence is consistent with previous findings reported by other authors [8,21]. The ester family (ethyl hexanoate, ethyl octanoate) contributed sweet and fruity notes. Regarding terpenes, dl-limonene was found instead of the commonly cited linalool, while still maintaining the characteristic citrus–floral contribution [21,22]. Notably, vanillin was absent, which could be attributed to the specific formulation of the chocolate or to instrumental limitations. Finally, the detection of furans, such as furfural, reflects Maillard reaction and caramelization products.

Foie gras toast with mango chutney: The chutney fraction contributed a rich variety of terpenes, such as limonene, α-pinene, and δ-3-carene, which impart fresh and resinous nuances [23]. The presence of furfural confirms the contribution of the Maillard reaction to the overall aroma. Although the literature identifies nonanal as a key contributor to the fatty notes of foie gras [24], cuminaldehyde was detected instead, providing spicy and anise-like complexity.

Boiled prawns: The analysis identified medium-chain aldehydes, such as hexanal and nonanal, which are markers for green and fatty notes in marine products [25]. The ester family (ethyl octanoate and ethyl hexanoate) contributed sweet notes, functionally replacing the methyl 2-methylbutanoate described in the literature. The absence of sulphur compounds (dimethyl disulphide) and high-impact aromatic heterocycles suggests a high-freshness sample or potential instrumental detection threshold limitations.

Pickled gherkins: The aromatic profile was dominated by acetic acid, a marker of lactic fermentation and pickling. Fatty acid ethyl esters (octanoate, decanoate) were identified, contributing fruity nuances. The detection of furfural and furfuryl alcohol is associated with carbohydrate degradation [26]. Likewise, the presence of limonene provides citrus notes common in plant matrices [23].

Blue cheese: As is characteristic of mold-ripened cheeses, some methyl ketones (2-hexanone, 2-octanone) were identified, and they are responsible for the moldy and buttery notes [9]. The detection of 2-heptanol confirms the enzymatic reduction of ketones. Furthermore, key volatile fatty acids, such as hexanoic acid and octanoic acid, were identified, contributing the typical rancid and pungent notes of its sensory profile.

Lemon sorbet with ginger: Limonene exhibited an exceptional abundance, consistent with the use of lemon peel [27]. The chutney’s contribution was reflected in terpenes such as β-myrcene, α-pinene, and linalool, aligning with ripe mango profiles [23]. Furans (furfural) derived from the thermal processing of sugars were detected, alongside acetic acid, which reinforces the overall acidic note [21].

Olives cv. Manzanilla: A high concentration of isodurene and furfural was identified. The presence of ethyl esters (octanoate and hexanoate) is common in fermented olives, although elevated levels may indicate prolonged storage [28]. The detection of 5-methylfurfural serves as a marker for thermal treatments, such as pasteurization [29].

Grilled meat: The analysis of the steak revealed hexanal, the primary marker of lipid oxidation in grilled meats [30]. Maillard reaction products were represented by furfural and furfuryl alcohol and terpenes (limonene) attributable to the use of seasonings or spices were also detected. A relevant finding was the identification of polycyclic aromatic hydrocarbons (PAHs), such as fluorene and naphthalene, derived from combustion during grilling.

3.3.2. Analysis of Sherry Wines

Fino Sherry wine: It presented a profile dominated by diethyl succinate and fruity esters such as ethyl hexanoate, but the detection of acetal confirms the impact of aging and oak maturation [31]. Additionally, TDN stood out as a compound to be monitored due to its potentially external origin [4].

Oloroso Sherry wine: Diethyl succinate was the major compound, indicating an advanced stage of chemical evolution [4]. Benzaldehyde contributes the almond-like nuances typical of oxidative aging [13]. As observed in the Fino wine, the presence of whiskey lactone is an unequivocal marker of contact with wood [32].

Pedro Ximénez Sherry wine: The profile was characterized by compounds derived from the raising and aging processes. Notable components included fatty acid esters (butanedioic, hexadecanoic) and acetic acid. The aromatic complexity of this sweet wine is underpinned by furans (furfural, 5-methylfurfural, and furfuryl alcohol), which are responsible for the caramel, roasted, and nutty notes [4].

3.4. Analysis of Pairings Based on Volatile Content

Table 6. Coincident volatile compounds by food product and wine type.

Food Product	Fino Wine	Oloroso Wine	PX Wine
Chocolate	Ethyl hexanoate; ethyl octanoate; furfural.	Ethyl hexanoate; ethyl octanoate; 5-methylfurfural; octanoic acid; decanoic acid; isoamyl acetate; hexadecanoic acid.	Ethyl hexanoate; ethyl octanoate; acetic acid; 5-methylfurfural; furfuryl alcohol; 3-methyl-1,2-cyclopentanedione; hexadecanoic acid; furfural; furfural; ethyl decanoate.
Foie gras with mango chutney	Ethyl hexanoate; furfural; ethyl pentanoate; ethyl octanoate; ethyl decanoate; octanoic acid; decanoic acid.	Ethyl hexanoate; ethyl octanoate; 5-methylfurfural; hexadecanoic acid.	Ethyl hexanoate; ethyl octanoate; acetic acid; furfural; ethyl decanoate; hexadecanoic acid.
Boiled prawns	Ethyl hexanoate; furfural; ethyl decanoate; furfuryl alcohol; isoamyl acetate; 2,6-dimethylheptan-4-one.	Ethyl hexanoate; ethyl octanoate; ethyl decanoate; isovaleric acid; ethyl dodecanoate; phenethyl alcohol; ethyl hexadecanoate; isoamyl acetate; benzaldehyde; hexadecanoic acid.	Ethyl hexanoate; ethyl octanoate; acetic acid; furfural; furfuryl alcohol; ethyl decanoate; ethyl dodecanoate; ethyl hexadecanoate; hexadecanoic acid.
Pickled gherkins	2,6-Dimethylheptan-4-one; ethyl hexanoate; ethyl octanoate; furfural; furfuryl alcohol; ethyl decanoate; octanoic acid; isoamyl alcohol; 2(5H)-furanone.	Ethyl hexanoate; nonanal; ethyl octanoate; ethyl decanoate; phenethyl alcohol; octanoic acid; 2-hexanol; isoamyl alcohol.	Ethyl hexanoate; ethyl octanoate; acetic acid; ethyl decanoate; furfuryl alcohol; furfural.
Blue cheese	Ethyl butyrate; pentanoic acid; isopropyl hexanoate; ethyl octanoate; ethyl decanoate; furfural.	Ethyl butyrate; isovaleric acid; ethyl octanoate; ethyl decanoate; phenethyl alcohol; octanoic acid.	Ethyl hexanoate; ethyl octanoate; ethyl decanoate; furfuryl alcohol; furfural; 3-methyl-1,2-cyclopentanedione.
Lemon sorbet with ginger	Ethyl octanoate; furfural; ethyl decanoate.	Ethyl octanoate; linalool; ethyl decanoate; hexadecanoic acid; ethyl 2-furoate.	Ethyl hexanoate; ethyl octanoate; ethyl decanoate; furfuryl alcohol; furfural; 3-methyl-1,2-cyclopentanedione.
Manzanilla olives	Ethyl hexanoate; ethyl octanoate; furfural; diethyl succinate.	Ethyl octanoate; 5-methylfurfural; diethyl succinate; octanoic acid; 2-methylnaphthalene.	Ethyl hexanoate; ethyl octanoate; 2(5H)-furanone; diethyl succinate; acetic acid; furfural; 5-methylfurfural.
Grilled meat	Ethyl octanoate; furfural.	Octanoic acid; hexadecanoic acid; furfuryl alcohol; 1,6,7-trimethylnaphthalene.	furfural; furfuryl alcohol; hexadecanoic acid; ethyl octanoate.

details the volatile compounds with the highest relative areas that matched within the analyzed wine and gastronomic product pairs. The results obtained from the hedonic test reveal that the quality of a pairing is not solely determined by the absolute number of shared molecules, but rather by the chemical nature and actual sensory impact of these compounds. Consequently, the application of Chartier’s Theory [7] is nuanced by its alignment with the sensory matrix of the dish and the relative intensity of the molecules.

The study identified cases where a high molecular overlap translated into excellent sensory evaluation, confirming the existence of potential chemical correspondences:

For example, Pedro Ximénez and Chocolate (score 9.8, Table 4) displays an exceptional affinity based on the furan family (furfural, 5-methylfurfural), which are linked to sweet and roasted notes [33,34]. The perceptual relevance of the pairing is ensured by compounds with low detection thresholds, such as ethyl octanoate (0.005 mg/L), and ethyl hexanoate (0.014 mg/L) [11]. The synergy between caramelized notes and fruity and fatty sensations accounts for its high acceptance. Another example is Fino wine and Boiled prawns (score 9.6, Table 4), where the shared presence of medium-chain esters, such as ethyl hexanoate, ethyl decanoate, and isoamyl acetate, establishes a fruity and fresh chemical correspondence that complements the crustacean’s lipid profile. Concurrently, the convergence of thermal and oxidative furanic compounds (furfural and furfuryl alcohol), arising from Maillard reactions during cooking and yeast autolysis during Fino’s biological aging, along with 2,6-dimethylheptan-4-one, generates a synergistic resonance of toasted and balsamic notes.

Despite the importance of molecular overlap, the analysis detected instances where an abundance of common volatiles did not prevent negative evaluations due to sensory conflicts. The Fino–Chocolate pairing illustrates how the wine’s fruity esters fail to harmonize with the roasted notes of cocoa, creating an unappealing contrast. Similarly, combinations with high molecular overlap, such as Oloroso–Chocolate or PX–Boiled prawns, were poorly rated by the expert consensus group, reinforcing that the specific aromatic character and its interaction with the food matrix are more critical than the mere quantity of shared compounds. In the case of Pedro Ximénez–Olives, the co-occurrence of furfural and acetic acid was insufficient to offset the opposition between the extreme sweetness of the wine and the saline and bitter profile of the olives. This negative contrast is driven by in-mouth gustatory perceptions rather than purely orthonasal aroma interactions; indeed, sensory analysis concluded that Fino pairs poorly with sweetness and is disadvantaged by high dish persistence, whereas PX displays a strong taste incompatibility with high saltiness.

An analysis of variance was performed on the relative areas of each of the 134 compounds identified in the eight dishes, taking the quality of the pairing with each wine (good, medium or bad) as the factor of variation and all the descriptors employed in the discussion were obtained from verified flavor databases [33,34].

In the pairings with Fino wine, only 12 compounds presented a p < 0.05, among which 3-methyl-1,2-cyclopentanedione (with toasted and sweet notes) and phenethyl alcohol (floral and sweet) confirmed a negative correlation; that is, the greater the amount of the compound in the dish, the worse its pairing with Fino was rated. On their part, compounds such as isoamyl alcohol (fermentation), 2,6-dimethylheptan-4-one (fruity and sweet), and α-terpinolene (citric) showed a positive but non-linear correlation, with a maximum rating at the intermediate concentration, as if an excess concentration penalized the evaluation. From these data, it could be deduced that Fino wine can accompany dishes of limited aromaticity well, provided that their most characteristic notes are not sweet and/or toasted.

In the pairings with Oloroso wine, there were 25 volatile compounds that presented some statistical significance between the categories (p < 0.05), all of them being negatively correlated with the quality of the pairing. Some of these compounds are characterized by toasted, sweet, woody, or spicy notes, but the most recurrent aromatic notes are balsamic, herbaceous, fruity, floral, and citric, which are characteristic of compounds such as 2-carene; geranyl acetate; 2-nonanone; α-terpinene; α-pinene; α-curcumene; β-bisabolene; β-myrcene; camphor; phenethyl acetate; camphene; cis-β-ocimene; fenchyl alcohol; limonene; γ-hexalactone; γ-elemene; linalool; and nerol. According to these data, it could be hypothesized that Oloroso wine does not accompany dishes with a broad and fresh nose well.

Regarding the pairings with PX wine, three compounds showed a positive correlation with the quality of the pairing (p < 0.05). The interaction of 3-methyl-1,2-cyclopentanedione (toasted, sweet, and present in the PX wine) and phenethyl alcohol (floral, sweet) in the pairing with PX is highly positive, precisely the opposite of what occurred with Fino wine, as previously explained. On the other hand, the compounds nonanal (citric), isoamyl alcohol (fermentation), and 2,6-dimethylheptan-4-one (fruity) presented a negative correlation with the quality of the pairing with PX (p < 0.05). From all of this, it could be deduced that a dish characterized by sweet and/or toasted notes, far removed from any sensation of freshness, is associated with a good pairing with this wine. And in this sense, it is interesting to specify that nine of the 28 compounds identified in the PX wine are defined by toasted, caramel, or sweet aromatic notes.

4. Limitations

Despite the significant findings regarding the sensory and molecular drivers of Sherry wine pairings, this study has several limitations that should be considered when applying the results to other contexts. First, instrumental limitations were identified during the volatile profiling of specific food matrices. In the analysis of dark chocolate, the absence of vanillin was noted, which may be attributed to the specific formulation of the local product or to the sensitivity limits of the equipment used. Similarly, the analysis of boiled prawns did not detect certain high-impact aromatic heterocycles or sulphur compounds (such as dimethyl disulphide), which could indicate either a very high level of sample freshness or limitations in the instrumental detection thresholds of the GC–MS methodology employed.

Second, the study utilized a specific set of samples (three types of Sherry wine and fifteen food matrices) purchased at a local market. While these are representative of common gastronomic products, the specific chemical and sensory profiles of ingredients can vary significantly based on brand, origin, and processing methods. Therefore, the potential chemical correspondences identified might differ when using international equivalents or different varieties of the same food products. Moreover, volatile compounds were identified and quantified tentatively without individual commercial standards, making these results semi-quantitative.

Regarding the sensory studies, the consumer study involved 105 participants. While this sample size provided statistically significant data for identifying general trends and preferences, the results may primarily reflect regional or cultural palate preferences associated with the location of the study. Further research involving a more diverse and global consumer base would be necessary to validate these pairing models across different cultural gastronomic traditions. An additional minor limitation of our study lies in the size of the expert consensus panel (n = 5). The small panel limits the generalizability and robustness of the descriptive sensory component; however, the extensive tasting experience of the selected experts, combined with duplicate measurements and acceptable standard deviations across replicates and panelists for most descriptors, helped support consistency within the study itself.

Selecting only the highest- and lowest-rated pairings for volatile and descriptive sensory profiling represents a methodological trade-off. While this contrastive approach maximized the signal-to-noise ratio to successfully identify key drivers of pairing success and failure, it may also inflate apparent correlations compared to a continuous sampling spectrum. Consequently, although highly effective for biomarker and descriptor discovery, the observed correlations should be interpreted within the context of this contrastive design. Furthermore, these relationships should be validated across intermediate pairing scores in future studies.

Finally, given that the PLS regressions were constructed using a limited number of food matrices (n = 8), the results should be interpreted with caution and from a fundamentally exploratory perspective. Although leave-one-out cross-validation suggests promising trends in the relationship between the 11 sensory descriptors and acceptance, the actual predictive capability of these models must be confirmed in future studies with a larger sample size.

5. Conclusions

This study contributes to establishing a scientific basis for the investigation of Sherry wine–food pairing by integrating consumer hedonic perception, descriptive sensory analysis, and volatile compound profiling. The results confirm that pairing quality is primarily driven by the specific interaction between the wine and the food matrix rather than the individual characteristics of either component. Key findings indicate that the studied Fino wine exhibits a strong affinity for saline and umami profiles, while the Oloroso tested harmonizes best with umami tastes and substantial body. In contrast, the Pedro Ximénez employed is driven by sweetness and persistence, showing a significant antagonism toward high saltiness. At a molecular level, the research identifies potential chemical correspondences through shared volatile compounds, such as the furan family in the PX–chocolate pairing. However, the effectiveness of these correspondences depends on the perceptual relevance and odor potency of the molecules, and also on other sensory attributes not related to the aromatic perception. Ultimately, these findings allow for the transition from empirical traditions toward a technical and indicative model for the systematic categorization of Sherry wine gastronomy.