Assessing the Potential of Distinctive Greek White Cultivars in Winemaking: Relationship Between Sensory Sorting Tasks and GC-MS Data

Abstract

This study explores the chemical and sensory differentiation of Greek white wines produced from five indigenous grape varieties—Savvatiano, Vidiano, Moschofilero, Assyrtiko, and Malagouzia—across diverse terroirs in Greece. A targeted analytical approach was employed to quantify 12 key volatile aroma compounds derived primarily from amino acid metabolism and lipid degradation, using GC-MS and GC-FID. The selected volatiles, including isoamyl alcohol, phenylethyl alcohol, tyrosol, and hexanoic acid ethyl ester, were chosen for their sensory relevance and their biosynthetic linkage to nitrogenous precursors. Principal Component Analysis (PCA) of wines from the 2019 and 2020 vintages revealed clear varietal clustering, under standardized winemaking conditions. Malagouzia wines were characterized by rich and diverse volatile profiles, particularly long-chain fatty acids and esters, while Vidiano exhibited a consistently restrained aromatic expression. Sensory analysis using sorting and ultra-flash profiling confirmed the chemical clustering, with Moschofilero, Vidiano and Malagouzia wines forming distinct sensory groups. The findings demonstrate that key amino acid-derived volatiles can serve as biochemical markers of varietal typicity and support the use of volatile profiling as a tool for terroir-driven wine classification and quality assessment in Greek white wines.

1. Introduction

Terroir, a term encompassing the environmental, viticultural, and enological factors that influence grape and wine composition, plays a defining role in the sensory and chemical attributes of wines [1]. Greece, with its diverse climatic conditions, distinct soil compositions, and indigenous grape varieties, presents a unique case for studying the impact of terroir on wine characteristics [2,3,4,5]. White wines from Greek varieties such as Savvatiano, Moschofilero, Assyrtiko, Malagouzia and Vidiano exhibit significant sensory diversity, a feature that can be linked, besides other compositional factors (terpenic composition, etc.), to their amino acid composition and the subsequent volatile aroma compounds formed during fermentation [4,6,7,8,9,10]. Understanding these relationships is crucial for both scientific and commercial applications, as they provide insights into regional wine typicity, quality markers, and potential consumer preferences [11].

The sensory character of wines resulting from those indigenous white cultivars was only recently investigated by employing rigid scientific sensory approaches [4,10]. Nevertheless, in both aforementioned studies, only the aspect of orthonasal aroma of the wines was studied through the method of frequency of attribute citation combined with Check All That Apply (CATA). Still, very important Greek grapes have gained global recognition and interest in the winemaking world mostly due to other unique sensory aspects. The most profound example is that of the Assyrtiko grape [7,8], mostly known among oenologists and sommeliers for its mouthfeel aspects [12], which have not yet been investigated with scientific sensory methods. Moreover, Vidiano, a relatively recently rediscovered grape that is indigenous to the island of Crete [9], has only been studied in terms of its sensory profile in one recent study [13], in which the objective was to understand the differentiation among the aroma profiles of Vidiano wines resulting from fermentations with different yeast species, rather than differentiation between Vidiano and other monovarietal white wines. In the above study, the sensory profile of the wines was only assessed in terms of orthonasal aroma using seven descriptors [13].

Amino acids play a crucial role in grape must composition, significantly shaping the sensory profile of wine through their impact on aroma and taste. They can account for as much as 40% of the total nitrogen content in grape juice and serve as the primary nitrogen source for yeast metabolism during fermentation [14]. The availability and composition of amino acids in grapes must influence yeast growth, fermentation kinetics, and the biosynthesis of key volatile compounds, including higher alcohols, esters, aldehydes, and sulfur-containing compounds [15]. Many of these volatile substances are responsible for fruity, floral, and spicy notes that enhance wine complexity [16,17]. The degradation of amino acids via the Ehrlich pathway results in the formation of higher alcohols and their corresponding esters, which have been identified as major contributors to wine aroma [18,19].

The amino acid composition of grapes is highly variable and depends on multiple factors, including grape variety, vineyard management practices, soil composition, and climatic conditions [20,21]. Several studies have demonstrated that amino acid profiles can be successfully employed to differentiate wines based on varietal origin, vintage, and terroir [22]. For instance, Soufleros and his coworkers et al. [22] were able to classify wines from Samos, Lemnos, and the Cyclades Islands according to their amino acid content. Similarly, ref. [21] differentiated Spanish red wines from Albariño, Godello, and Treixadura based on amino acid profiles, highlighting the strong influence of variety and fermentation conditions on final wine characteristics.

Beyond their role in aroma formation, amino acids also influence mouthfeel, balance, and aging potential in white wines. Wines rich in certain amino acids, such as glutamine and proline, have been associated with improved texture and body, whereas others, such as methionine and cysteine, contribute to the formation of sulfur-containing compounds that can affect wine aroma stability over time [17,23,24]. Understanding those interactions provides valuable insights for winemakers aiming to optimize fermentation strategies and enhance the expression of terroir through controlled nitrogen management [15].

This study aims to address the gap in current research by investigating how amino acids can contribute to the sensory differentiation of Greek white wines made from indigenous vine cultivars and originating from various terroirs. Namely, the Greek cultivars Moschofilero, Malagouzia, Assyrtiko, Vidiano and Savvatiano were used and vinified in the same way to make the respective monovarietal wines. Identification of key marker volatiles linked to amino acids was conducted using gas chromatography. Additionally, Principal Component Analysis (PCA) was employed to examine clustering patterns among wine varieties based on their volatile profiles.

On a second level we wanted to investigate the sensory profiles of a subset of the above wines by employing the method of free sorting task combined with Ultra Flash Profiling, considering not only aroma but all relevant flavor modalities (basic tastes, mouthfeel aspects, and aftertaste). Last but not least, we wanted to explore the full sensory fingerprint—aroma, taste and mouthfeel—of the Vidiano variety and how wines from that cultivar differ from other monovarietal Greek white wines.

2. Materials and Methods

2.1. Samples

Samples were obtained from the Agricultural University of Athens (AUA) from experimental vinifications conducted in 2019 and 2020 across thirty-nine vineyard sites throughout Greece. A standardized vinification protocol was applied across the selected varieties to minimize enological variability, with fermentation carried out using a common yeast to ensure that any observed differences in wine composition were primarily attributed to grape origin rather than yeast strain variation. Specifically, monovarietal wines of the Malagouzia, Moschofilero and Assyrtiko varieties were collected from the 2019 vintage, originating from twenty-four (24) distinct vineyard sites across twelve (12) geographical regions (

Table 1. Experimental white wines used in the research. Coding, vintage year, variety and regions of origin are presented. Also, membership of the samples in a PDO zone and participation in the sensory analysis tests is marked.

Year	Variety	Region	Vineyard Code	Zone of PDO	Code Name	Sensory
2019	Moschofilero	Ilia	AOLY		MSF_AOLY
		Arcadia	PART	✔	MSF_PART	✔
		Arcadia	AGIO	✔	MSF_AGIO
		Arcadia	LOUK	✔	MSF_LOUK
		Achaias	FTER		MSF_FTER
		Arcadia	RIZE	✔	MSF_RIZE	✔
		Korinthos	ZEYT	✔	MSF_ZEYΤ
		Korinthos	ZEYN	✔	MSF_ZEYΝ	✔
	Malagouzia	Attica	KANT		MLG_KANT
		Florina	AMYN		MLG_AMYN	✔
		Attica	KITH		MLG_KITH
		Achaias	OAIG		MLG_OAIG
		Korinthos	ANEM		MLG_ANEM	✔
		Argolida	OARG		MLG_OARG
		Ilia	AOLY		MLG_AOLY
		Korinthos	ASPR		MLG_ASPR
	Assyrtiko	Attica	SPAT		ASY_SPAT
		Korithos	KNEM		ASY_KNEM
		Evia	MEYM		ASY_MEYB
		Kozani	VELV		ASY_VELV	✔
		Drama	DRAM		ASY_DRAM
		Chalkidiki	OXALK		ASY_OXALK
		Karditsa	KARD		ASY_KARD	✔
		Thira	SANT	✔	ASY_SANT	✔
2020	Savvatiano	Attica	ERYT		SAB_ ERYT	✔
		Attica	SPAG		SAB_SPAG
		Attica	MEGR		SAB_ MEGR
		Attica	PAIN		SAB_PAIN	✔
		Attica	STAM		SAB_STAM	✔
		Attica	KERT		SAB_KERT	✔
		Attica	MARK		SAB_MARK	✔
		Viotia	ASKR		SAB_ASKR
		Viotia	SPAT		SAB_SPAT
		Attica	KATZ		SAB_KATZ	✔
	Vidiano	Heraklion	DAFN		BID_DAFN A	✔
		Heraklion	METX		BID_METX A	✔
		Heraklion	ASIT		BID_ASIT A	✔
		Kavala	KABL		BID_KABL A	✔
		Drama	DRAM		BID_DRAM A

). Similarly, monovarietal wines of the Savvatiano and Vidiano varieties were collected from the 2020 vintage, representing fifteen (15) vineyard sites across four (4) regions (Table 1). Each variety was thus evaluated in only one vintage. All experimental vinifications were conducted at the Laboratory of Oenology and Alcoholic Beverages (EOAP) of the Agricultural University of Athens, following a rigorously standardized white microvinification protocol. Grapes were destemmed and crushed, followed by the addition of sulfur dioxide (50 mg/L) and pressing. Pectinolytic enzymes were added, and static settling was performed at 4 °C. The clarified must was racked and inoculated with a commercial S. cerevisiae yeast strain (Safoeno GV s107) (200 mg/L), and alcoholic fermentation was carried out at 20 °C. After fermentation, the wine was separated from the lees and transferred to stainless steel containers, where malolactic fermentation was allowed to proceed. All vinifications were performed at least in duplicate under identical conditions. All wine samples were bottled and sealed with traditional corks. Upon receipt, they were stored at 4 °C in the dark. Separate bottles from the same batch were used for chemical and sensory analyses. While some terroirs were represented by a single vineyard, this was a deliberate choice to ensure broader regional coverage and capture potential site-specific aromatic signatures. The study’s primary aim was to explore varietal expression across diverse environments rather than to statistically generalize terroir effects.

2.2. Sample Preparation for Gas Chromatography

To determine the key volatile compounds, gas chromatography coupled with mass spectrometry (GC-MS) was employed. Volatile compound quantification was conducted using gas chromatography with flame ionization detection (GC-FID), a well-established method for identifying and quantifying key aromatic compounds in wines [25,26]. Standard compounds were used for the identification of the analytes in GC-FID. Following the guidelines of the Metabolomics Standards Initiative [27], compounds identified without reference standards are classified as “putative identifications.” Volatile concentrations were calculated using an internal standard (IS) to ensure consistency and accuracy.

The extraction protocol for volatile compounds was based on the methodology described by Tsapou et al. [28], with slight modifications. Each wine sample was diluted with deionized water to achieve an alcohol content of 12% v/v. From this solution, 50 mL were extracted twice using an equal volume (50 mL) of an diethyl ether (95%)/pentane (95%) mixture (Chem. Lab, Athens, Greece) in a 1:1 ratio (v/v). The combined organic phases were dried over anhydrous sodium sulfate (Sigma–Aldrich, St. Louis, MO, USA) and filtered through a paper filter. Subsequently, 25 µL of a 3-octanol solution (2.5 g/L) (Sigma–Aldrich, St. Louis, MO, USA) was added as an IS. Solvent removal was carried out by gentle heating in a water bath (50–60 °C) using a pear-shaped flask equipped with a 20–25 cm Vigreux column. The evaporation process yielded approximately 200 mg of residue, which was further concentrated under a nitrogen stream until the final sample weight reached 100 mg.

2.3. GC-MS Analysis

The analytical setup consisted of an Agilent 6890 Series Gas Chromatography (GC) system (Agilent Technologies, Santa Clara, CA, USA), coupled with a 5975C VL MSD detector and equipped with an Agilent HP-5MS fused silica capillary column, 30 m × 0.32 mm i.d. × 0.25 µm film thickness. Moreover, 1.0 µL of the sample was injected in a split ratio of 100:1. The injector temperature was set at 250 °C, the carrier gas was helium with a flow rate of 1 mL/min, and the oven temperature program was 50 °C for 2.5 min, increased to 180 °C at 2.5 °C/min, to 230 °C at 2 °C/min, to 250 °C at 6 °C/min, and then the temperature was kept at 250 °C for 5 min, increased to 270 °C at 5 °C/min, which was kept for 2 min (at 270 °C). The transfer line temperature was maintained at 280 °C. The mass spectrometer operated in electron ionization (EI) mode with an ionization energy of 70 eV, scanning a mass range from 40 to 550 amu, and with the manifold temperature set at 270 °C. Data acquisition was performed using Turbomass 5.0 ChemStation software. Quantification of the extracted volatile compounds was based on the recovery of the internal standard (3-octanol) and expressed in mg/L. Each sample was analyzed in duplicate.

2.4. GC-FID Analysis

Volatile compound analysis was performed using a Shimadzu Nexis GC-2030 gas chromatograph ( Agilent Technologies, Santa Clara, CA, USA), equipped with a flame ionization detector (FID). Separation was achieved on a BGB MG-HT-1-025-025-30 capillary column (30 m × 0.25 mm i.d. × 0.25 μm film thickness). A 2.0 μL aliquot of each extract was injected in split mode (5:1), with the injector temperature set at 200 °C. Helium served as both the carrier gas (1.33 mL/min) and the makeup gas (24 mL/min). The airflow and hydrogen flow rates were adjusted to 200 mL/min and 32 mL/min, respectively. The oven temperature program began with an initial hold at 40 °C for 2.5 min, followed by a ramp to 180 °C at 2.5 °C/min. The temperature then increased to 230 °C at 2 °C/min, followed by a rise to 250 °C at 6 °C/min with a 5 min hold. Finally, the temperature was raised to 270 °C at 5 °C/min and held for 2 min. The transfer line was maintained at 280 °C. Chromatographic data were processed using standard solutions for compound identification, and quantification was carried out in mg/L using an internal standard calibration approach. Each sample was analyzed in duplicate.

2.5. Samples for Sensory Analysis

In order to assess the results of the chemical analyses, we also conducted an organoleptic evaluation of the wines. Specifically, the sorting method was employed, an emerging sensory evaluation technique that has gained attention in wine assessment [29,30]. A total of 21 wines were evaluated in the organoleptic evaluation experiments from the 2019 and 2020 vintages collectively, including 18 of the experimental wines produced under controlled vinification conditions and 3 commercially available monovarietal wines, which were included to support the sensory assessment of varietal typicity. It should be noted that the vinification conditions of the commercial wines were not controlled or documented, and their inclusion is discussed further in the discussion section. The only reason why not all 38 experimental wines were assessed for their sensory character is because we did not have enough quantity left from all wines (GC-MS and GC-FID analyses were conducted, concluded and reported before starting the sensory work). All details of the samples assessed in sensory evaluation are shown in Table 1 and

Table 2. Commercial white wines that were used for sensory analysis. Coding, vintage year, variety and regions of origin are presented. Also, membership of the samples in a PDO zone is indicated.

Year	Variety	Region	Zone of PDO	Code Name
2019	Assyrtiko	Thira	✔	ASY SN 19
2019	Moschofilero	Korinthos		MSF K 19
2020	Savvatiano	Korinthos		SAB K 20

.

2.6. Assessors

Ten assessors, comprising five females and five males, aged 27–65, who were staff members and/or postgraduate students at the Department of Wine, Vine, and Beverage Sciences at the University of West Attica, evaluated all products over two sessions in total. All assessors had been screened and selected for becoming members of a wine panel as described in Tsapou et al. [27] and had gone through a training of 34 h on aromas, tastes and mouthfeel aspects relevant to wines according to ISO standards ISO8586:2023 [31] and ISO 5496:2006 [32]. Participants provided informed consent by confirming, “I am aware that my responses are confidential, and I agree to participate in this survey,” with the option to withdraw at any time without explanation. All products tested were safe for consumption, and participants confirmed their informed consent to participate.

2.7. Sensory Methodology

Sensory evaluation of the samples was conducted in two sessions. In the first session, each taster assessed 10 wines from the 2019 vintage, while in the second, they evaluated 11 wines from the 2020 vintage. All sensory evaluations were conducted in individual, off-white booths under controlled conditions, combining natural and artificial lighting, in an odor-free, air-conditioned room maintained at 20 ± 2 °C, in accordance with ISO 8589:2007 guidelines. Samples were presented in a randomized sequence for each panelist using a partially balanced design and were coded with unique three-digit identifiers to ensure anonymity. Assessments were carried out excluding visual criteria, and each assessor participated in an individual session for each set of wines (representing the 2019 and 2020 vintages). During both sessions, participants were instructed to initially smell and then taste each sample once, following the predetermined left-to-right order. They were asked to describe the sensory attributes that best captured the similarities and differences among the samples [33]. After this initial evaluation, panelists were free to revisit the samples as many times as they wished and in any sequence. They could then freely classify the wines into distinct groups [33], grouping them as many times as they deemed necessary into as many categories as they found appropriate. The objective was to detect sensory similarities and differences among the samples by describing them with their own vocabulary and classifying them as many times as needed, even assigning a single product to its own category if necessary. After making the groups, each assessor was asked to give a description to each group using a group of 3–5 words [28].

2.8. Statistical Analysis

All statistical analyses were conducted using XLSTAT Version 2024.3.0 (Addinsoft, (New York, NY, USA). Principal Component Analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS-DA) were employed for the projection of all GC results. A global similarity matrix was built for the sensory sorting data of the wines of each vintage separately, and a Multidimensional Scaling (MDS) analysis was applied to each matrix, representing the grouping of wines from the sensory results. All sensory descriptors given to each product (subtracted from all groups of products constructed from all assessors)—for each vintage separately—were collected, and the number of times each descriptor was assigned to each product was recorded in a product attribute matrix. Following that, Correspondence Analysis (CA) and Agglomerative Hierarchical Clustering (AHC) were run on those data in order to observe valid associations between products and terms assigned to them by the assessors.

3. Results and Discussion

3.1. Selection of Key Volatile Compounds from Gas Chromatography Analyses

The concentration and composition of nitrogen in grape must is highly variable and affects yeast metabolism and consequently the aroma of wine [34]. There are many studies in which the different amino acids used by the yeast of the genus Saccharomyces, mainly the species cerevisiae, have been classified according to their ability to support yeast growth, measured as production time, when they are the sole nitrogen source [17]. During fermentation, yeast metabolizes the available amino acids and other nutrients to support growth and biomass production. In this process, a series of volatile aromatic compounds are produced, including esters, higher alcohols, volatile fatty acids, carbonyls, and sulfur compounds. The production of many of these aroma-active compounds is directly dependent on the nitrogen sources available on the substrate. Extensive studies have been conducted on the effect of complex amino acid mixtures in real and synthetic grape musts [35]. In many findings among the studies, whether concerning amino acid mixtures on synthetic or real musts, a connection has been found in the formation of various aromatic compounds, such as higher alcohols and higher acids, with the degradation of branched-chain and aromatic amino acids via the Ehrlich pathway.

In this study, a targeted selection of key volatile compounds was used to evaluate their contribution to the chemical and sensory differentiation among Greek white wine varieties. Initial analyses with the gas chromatograph with mass spectrometry detector (Agilent Technologies, Santa Clara, CA, USA), yielded a total of 88 identified compounds, including a sum of 25 acids, 16 alcohols, 42 esters (ethyl, hexyl, acetate, phenyl, and methyl) and 5 aldehydes (Tables S1 and S2). The selected volatile compounds were chosen based on their consistent detectability across all samples and their suitability for quantification using FID. This strategy aimed to support the development of a reproducible and practical method for varietal discrimination. Identification and quantification (

Table 3. Key volatiles for the research after GC-MS analysis. Details about amino acids that could be linked through the Ehrlich metabolic pathway, other metabolic pathways and odor descriptions.

“Key” Volatiles	Amino Acids	Metabolic Pathway	Odor Description
Isoamyl alcohol	Leucine	Ehrlich pathway (from leucine)	Fermented/fusel 1
Isoamyl acetate	Leucine	Esterification of isoamyl alcohol	Fruity 1
1-Hexanol	Valine/Isoleukine	Lipoxygenase pathway (linoleic acid)	Herbal/green 1
Hexanoic acid	Valine/Isoleukine	Fatty acid degradation	Fatty/cheesy 1
Hexanoic acid ethyl ester	Valine/Isoleukine	Esterification of hexanoic acid	Fruity 1
Phenylethyl alcohol	Phenylalanine	Ehrlich pathway (from phenylalanine)	Floral 1
Diethyl succinate	Leucine	Citric acid cycle (succinate esterification)	Fruity 1
Tyrosol	Tyrosine	Tyrosine metabolism	Floral 1
Tryptophol	Tryptophan	Tryptophan metabolism	Mild floral, fusel-like 1
Hexadecanoic acid		Fatty acid synthesis	Waxy 1
2-Propenoic acid, 3-(phenylthio)-ethyl ester		Amino acid catabolism + esterification
Octadecanoic acid		Fatty acid synthesis	Waxy 1

¹ the good scents company [37].

; Tables S3 and S4) of specific volatile “key compounds” were carried out in all samples, based on two criteria: First, volatiles derived from the metabolism of various amino acids were selected, and the second criterion was the significance of the volatiles based on their concentration. Those compounds included isoamyl alcohol, isoamyl acetate, 1-hexanol, phenylethyl alcohol, diethyl succinate, tyrosol, tryptophol, hexanoic acid, hexanoic acid ethyl ester, hexadecanoic acid and octadecanoic acid. Although direct quantification of amino acids in the grape musts was not performed, the observed volatile profiles provide indirect yet meaningful evidence of varietal-specific amino acid utilization. This interpretation is consistent with previous studies linking amino acid composition to aroma expression in wines and supports the hypothesis that nitrogen metabolism plays a key role in shaping varietal aromatic identity [18,19]. Differentiation of Greek white wine varieties based on their volatile compound profiles can demonstrate the significance of specific metabolic pathways in shaping varietal aroma and typicity. The selection of 12 key volatile compounds in our case provided a focused yet informative perspective, capturing the essential biochemical and sensory markers across multiple vintages (Table 3). These volatiles arise predominantly through amino acid catabolism via the Ehrlich pathway and lipid metabolism, with their concentrations influenced by nitrogen availability, yeast strain activity, and must composition [34,36]. Those compounds were ultimately selected for their dual role as biochemical markers and contributors to varietal aroma typicity in Greek white wines.

3.2. Principal Component Analysis on Wine Volatiles from the 2019 Vintage

Principal Component Analysis (PCA) was conducted on the mean values derived from duplicate measurements of 24 wine samples and 12 volatile markers. The first two principal components (PC1 and PC2) accounted for 50.48% of the total variance (PC1: 27.55%, PC2: 22.92%) and are illustrated in Figure 1a,b. To provide a more comprehensive view of the data structure, the third and fourth components (PC3: 13.51%, PC4: 10.51%) were also examined and are presented in the Supplementary Files (Figure S1a,b). PCA of the 2019 vintage revealed distinct patterns of varietal separation based on the distribution of volatile compounds, particularly among Malagousia, Assyrtiko, and Moschofilero grape varieties, which were driven by differences in ester, alcohol, and fatty acid profiles. Along with the first principal component (F1), which accounted for a significant proportion of the variance, the compounds contributing most strongly to the observed differentiation were hexanoic acid ethyl ester, hexanoic acid, and diethyl succinate. Those volatiles played a critical role in the discrimination of the varieties, with Assyrtiko and Moschofilero clearly separated and clustered toward the left quadrants of the biplot, indicating their association with lower levels of these compounds. In contrast, the second principal component (F2) was primarily influenced by isoamyl alcohol, phenylethyl alcohol, and tyrosol. These compounds significantly affected the positioning of Malagousia samples, which were distributed higher along this axis, while Assyrtiko showed a moderate association and Moschofilero was relatively unaffected. Compounds such as isoamyl acetate and 1-hexanol exhibited the least influence on the overall variance explained by the model, suggesting a limited role in differentiating the 2019 wine samples based on their volatile profiles (Figure 1a,b).

According to the findings of Nanou et al. [4], the volatiles that most significantly contributed to the differentiation of Moschofilero were terpenes such as linalool, cis-rose oxide, and geraniol, all of which are characterized by an intense floral and less fruity aromatic profile. However, these specific volatiles were not utilized in the classification of the present study’s results. Notwithstanding, the primary volatiles contributing to the grouping of samples associated with Moschofilero grapes, as illustrated in Figure 1, were diethyl succinate and phenyl ethyl alcohol, both of which exhibit a floral and fruity aromatic profile (Table 3). Regarding the samples derived from Assyrtiko and Malagouzia grapes, elevated concentrations of 1-hexanol and hexanoic acid ethyl ester were observed in comparison to those from Moschofilero. The aromatic profiles of the varieties are consistent with the findings of Nanou et al. [10], exhibiting predominantly fruity (citrus), green, and herbal characteristics. Along the F2 axis, a distinct intra-varietal separation was also evident among Malagouzia samples, primarily influenced by the presence of tyrosol, phenylethyl alcohol, and isoamyl alcohol—volatiles that contributed more significantly to samples originating from the Attica and Corinth regions. In the case of Assyrtiko wines, a strong intra-varietal differentiation was observed in both F1 and F2 dimensions (Figure 1a). Differentiation in the F1 axis was predominantly driven by the presence of succinic acid diethyl ester on the one side and hexanoic acid together with ethyl hexanoate on the other side, effectively distinguishing samples from Santorini and Evia from those of Attica and northern regions. In parallel, those wines were also differentiated in F2 based on their tyrosol, phenyl ethyl and isoamyl alcohols on the one side vs. tryptophol on the other side (Figure 1). In a recent work with Assyrtiko wines, where the objective was to study the effect of different yeast species on the fermentation behavior and aroma compounds of the products, isoamyl alcohol, tryptophol and hexanoic acid were also among the markers that significantly differentiated the products, even though wines in that study were made with non-Saccharomyces yeasts, which cannot relate directly to the Assyrtiko typicity [38].

3.3. Principal Component Analysis on Wine Volatiles from the 2020 Vintage

Principal Component Analysis (PCA) was applied to the mean values obtained from duplicate measurements of 15 wine samples and 12 selected volatile markers, all derived from the 2020 vintage. The first two principal components accounted for 63.17% of the total variance in the dataset, with PC1 explaining 54.21% and PC2 contributing 17.96% (Figure 2a,b). The PCA mapping from the 2020 vintage consistently supported varietal discrimination through aromatic profiling, with an even more pronounced separation of the varietal clusters, as follows. First, the Vidiano grape maintained its separation in the higher left quadrant, again strongly linked with tryptophol, notably consistent across both years, confirming the metabolic stability of this marker across vintages. Secondly, the Savvatiano grape retained differentiation along PC2, particularly through phenylethyl alcohol and tyrosol, reinforcing the floral typicity for this variety. These results complement previous studies regarding the aromatic profile of Savvatiano. Specifically, the floral character of Savvatiano wines is enhanced, in line with the work of Lola et al. [39]. Vidiano is an indigenous Greek grape originating from the island of Crete, and the only scientific work related to the quality of wines it can produce was recently published [9]. Some of the main volatiles associated with that variety in that research were 1 Hexanol and phenyl ethyl alcohol. This is consistent with our study, as we utilized both to perform the projection of the wines from that grape and compare them to other monovarietal wines.

3.4. Principal Component Analysis of the Combined Wine Volatiles of 2019 and 2020 Vintages

In order to better observe the variation between different cultivars as well as intra-varietal variation, a Principal Component Analysis (PCA) was performed on the averaged values of the duplicate measurements of the wines from both vintages (39 products) and 11 aromatic markers, and 52.09% of the total variance in the data was explained in two principal components (F1: 27.54%, F2: 24.55%) (Figure 3). The integrated PCA of the 2019 and 2020 vintages could provide a comprehensive view of the consistency and variability in the varietal expression of volatile compounds over time. The combined plot reveals robust, clear clusters of samples according to their volatile profiles by variety. A limitation of the present study is that each grape variety was evaluated in only one vintage year. The first two principal components accounted for over 50% of the total variance, ensuring a reliable two-dimensional space for interpretation (Figure 3). In that figure, wines made from the Vidiano grape variety (e.g., BID-DAFN, BID-KABL) consistently appeared in the lower-left quadrant, indicating a relatively homogeneous volatile profile characterized by low concentrations of most of the active compounds. This grouping suggests that Vidiano wines tend to be chemically less intense in terms of the specific aroma-active volatiles considered in this analysis. On the other hand, wines made from the Savvatiano grape were positioned in the upper-left quadrant and were primarily characterized by isoamyl acetate, phenylethyl alcohol, and tyrosol. This consistency underscores the genetic and biochemical uniqueness of those varieties. In contrast, Malagouzia samples were primarily distributed along the right side of the plot, with notable clustering in the upper and lower right quadrants. These samples remained strongly associated with long-chain fatty acids such as octadecanoic and hexadecanoic acids, as well as esters and alcohols. This positioning indicates a richer and more complex volatile profile for Malagouzia, likely contributing to its characteristic aromatic intensity. Nevertheless, both Assyrtiko and Moschofilero exhibited some vintage-specific variation while retaining overall coherence in their volatile profiles (Figure 3). The combined PCA demonstrates that varietal identity plays a significant role in shaping the volatile compound profile of the wines made from Malagouzia and Savatiano grapes, exhibiting complex and diverse aroma characteristics, while Vidiano samples clustered more tightly in the PCA space, indicating a more homogeneous volatile compound profile compared to the other varieties. This result also underlines the minimal impact of vintage on the core aroma-defining compounds when the viticultural and winemaking practices are controlled—an important consideration for terroir studies and Protected Designation of Origin (PDO)/Protected Geographical Indication (PGI) classification.

Previous studies have reported distinct volatile profiles among Greek grape varieties. In the research of Lola et al. [39], Savvatiano wines were reported to also exhibit variability in ester concentrations depending on terroir, with compounds like isoamyl acetate and ethyl decanoate influencing the fruity character of the wines. In another research [9], using a non-targeted GC-MS approach, a total of 89 free and 103 bound volatile compounds were identified across the studied grape varieties. Among them, Malagousia exhibited the highest terpene concentrations, particularly within the bound fraction, underscoring its pronounced floral aromatic profile, while Savvatiano was marked by fatty aromas. As for the comparison of Assyrtiko, Malagousia and Moschofilero, the results of sensory analysis of Nanou and colleagues [10] revealed that Malagousia wines are characterized by lemon, grapefruit, and citrus blossom aromas, while Moschofilero wines exhibit floral profiles such as rose and jasmine. Assyrtiko wines are noted for earthy, mushroom, and nutty odors, with some samples also displaying lemon and honey notes. The results of the present study are in agreement with these findings, particularly in the clustering of Savatiano samples in the low-variance region and the diverse aromatic signatures of Malagouzia and Moschofilero.

The classification model of PLS-DA demonstrated a high degree of accuracy in assigning wines to their correct varietal groups, with 94.87% overall accuracy (37 out of 39 samples correctly classified). Notably, Vidiano, Moschofilero, and Savvatiano achieved 100% correct classification, while Assyrtiko and Malagouzia were classified with 87.5% accuracy, each with only one misclassified sample. These results highlight the strong varietal identity expressed in the volatile profiles of Greek white wines. Despite the moderate predictive ability of the model (Q² cum = 0.265), the explained variance was substantial, with R²X = 0.684 and R²Y = 0.544 across the first four components. This indicates that the model effectively captured the underlying structure of the data, allowing for robust varietal discrimination even within an exploratory multivariate framework. The high classification performance reinforces the hypothesis that volatile composition can serve as a reliable marker of varietal typicity under standardized winemaking conditions.

The selected 12 volatile compounds could cover multiple metabolic pathways relevant to wine aroma (Table 3). For instance, phenylethyl alcohol and tryptophol derive from amino acid metabolism (phenylalanine and tryptophan, respectively), while hexanoic acid and 1-hexanol are products of fatty acid degradation. Isoamyl acetate and other esters are formed during yeast-mediated esterification. The inclusion of compounds from these distinct pathways ensures a representative and pathway-diverse dataset, enhancing their utility as potential varietal markers. Previous studies have also linked those compounds to variety-specific profiles [40,41], supporting their selection.

3.5. Sensory Mapping and Description of Wines from the 2019 Vintage

Results from the free sorting task of the wines from the 2019 vintage were initially sorted in a similarity matrix and a Multidimensional Scaling (MDS) analysis was run on them, resulting in a two-dimensional product space with clear clusters and separation of the products (Figure 4a). The inclusion of commercially available wines in the sensory dataset aimed to provide reference points for varietal typicity. However, due to the lack of information regarding their vinification protocols, these samples may introduce uncontrolled variability. Their results were interpreted with caution and primarily served to contextualize the sensory profiles of the experimental wines. In Figure 4, a clear separation of all assessed wines made from the Moschofilero grape was apparent, with all those products clustering among them and separating from the rest of the wines with positive to very high scores in the second dimension. More specifically, the three experimental Moschofilero wines MSF-ZEUN, MSF-RIZE, and MSF-PART formed a clear cluster on the highest scores of the second dimension and separated further from the fourth commercial Moschofilero wine from Corinth (MSF-K19). Among those three wines, the last two (MSF-RIZE, MSF-PART), originating from the Mantineia PDO zone, were clustered even closer together, possibly reflecting the unique sensory expression of the Moschofilero PDO region (Figure 4a). The collected Ultra Flash Profiling (UFP) [27] results for the same products, when subjected to Correspondence Analysis (CA), demonstrated that Moschofilero wines were indeed clustered, also based on the choice of terms used, and associated with notes ranging from rose and Turkish delight rose (typical of the commercial Moschofilero wine from Corinth) to sweet, floral, fruity, delicate, citrus fruit, cherry, perfumy, and citrus liqueur, but also medicinal and coffee, even though the last two descriptors were only mentioned once, for the MSF-PART and MSF ZEYT products, respectively (Figure 5a). Those results are consistent with the works of Nanou et al. [4,10] in which Moschofilero varietal wines have been highly associated with predominately floral—such as jasmine, rose and citrus blossoms—and citrus fruit-like aromas.

Back on the MDS space of products from the free sorting task, the two representatives of the Malagousia grape variety were also clustered together and separated from the other wines with high scores on the first dimension (Figure 4a). The same products were also clustered on the Correspondence Analysis run on the attribute frequency used to describe them, being central in that map (Figure 5a). Indeed, in Figure 5a, products MLG-AMYN and MLG-ANEM appear centrally in the space and are linked to the characters: wood, burnt wood, smoke, chemical, vegetative, bell pepper, canned/cooked, nuts, minerality, astringent, average aftertaste and sweet aromas/bonbon (Figure 5a). In the work of Nanou et al. [10], in which four commercial Malagousia products were profiled using the frequency of attribute citation combined with the CATA methodology, the following words are mentioned as the most typical of the cluster that contained three out of four Malagousia products: “earthy”, “lemon”, “mushroom”, “nuts” and “grapefruit”. Additionally, in the work of Nanou et al. [4], again most of the Malagousia wines were related to citrus and earthy odors. From comparison to those works with respect to the Malagousia grape, we could only find similarity to our findings as far as the nutty character was concerned but not so much on the earthy and citrus side. Nevertheless, we chose to profile the wines primarily by the sorting task, followed by Ultra Flash Profiling, rather than using the CATA method. Moreover, both above studies were aiming to characterize those unique Greek grape varieties based on their orthonasal aromas only, whereas in our case we let the assessors choose their descriptors based on what they found to be the most important in clustering and/or separating a product group from another when including all sensory flavor modalities in the assessment (namely ortho- and retronasal aromas, basic tastes, mouthfeel and aftertaste aspects).

Finally, the remaining four (4) wines from that experiment were from the Assyrtiko variety. In the case of those wines, no clear clustering was observed from the free sorting task exercise. In fact, all three experimental wines were spread in the MDS space, with Assyrtiko from Karditsa (ASY-KARD) being close to the Moschofilero wines from Corinth, the experimental Assyrtiko from Santorini (ASY-SANT) appearing close to the Malagousia group and clearly separated from the other three Assyrtiko products, and only the Assyrtiko from Kozani (ASY-VELV) being clustered to the commercial Assyrtiko product from Santorini (ASY SN 19) (Figure 4a). Looking at the Correspondence Analysis on the UFP data, again the Assyrtiko wines used in that experiment were not clustered based on the terms given to them (Figure 5a). Seen from the attribute choice point of view, the Assyrtiko from Karditsa (ASY-KARD) was closer to the Malagousia wines, characterized by similar words, and complimented by the words: “spicy”, “thyme”, “solvent”, “full body” and “high sourness” (Figure 5a). As far as that grouping was concerned, in the work of Nanou and co-workers (2020), their wines made from the Assyrtiko grape were also in the same cluster with the Malagousia ones, even though they were characterized mostly by earthy, nutty, lemon and honey attributes. The Assyrtiko from Kozani (ASY-VELV) was linked to yellow and sour fruits, as well as botanic and ethyl acetate flavors, and was relatively close to the experimental Assyrtiko from Santorini on the scores of the second dimension of the map, while they were both associated with high ethanol flavor (Figure 5a). The experimental Assyrtiko wine from Santorini (ASY-SANT) was characterized by the highest frequencies in ripe fruits, honey, and oxidation, as well as plastic and dust at times, and stood out from all other wines with the highest scores in the first dimension (Figure 5a). Finally, the commercial Assyrtiko wine from Santorini was standing separate from other wines, positioned with high scores of both dimensions and linked to white flowers, glue, yeastiness, and high complexity but a short aftertaste (Figure 5a).

It is worth noting that, as far as the Ultra Flash Profile (UFP) carried out by the ten assessors on the ten 2019 wines was concerned, the 53 terms that were collectively used appeared to have a significant relationship with the samples (p = 0.045), and the Agglomerative Hierarchical Clustering carried out on the Correspondence Analysis data gave automatically four (4) distinct groups of samples (Figure 5b). Those were (1) a group with all experimental Moschofilero wines and one experimental Assyrtiko wine from Karditsa (ASY-KARD), (2) a group with the experimental wines from the Malagousia grape, (3) a group with the experimental Assyrtiko wines from Santorini (ASY-SANT) and Kozani (ASY-VELV), and (4) a group with the two commercial wines used in the set, one from the Assyrtiko and one from the Moschofilero grapes (Figure 5b). The above automatic classification, based on the terms given to the samples by all assessors, demonstrates clear clustering of the varieties Malagousia and Moschofilero (excluding the commercial Moschofilero product, which was made through different oenological practices from the experimental wines) based on their global sensory profiles. The two commercial products included in the experiment also clustered between them, despite coming from different cultivars, which is also not unexpected considering the different oenological practices being used in a winery in comparison with those in the university (experimental wines). Finally, the experimental products from Assyrtiko demonstrated the largest variation among them, a fact that was also observed in the data of their aromatic profiles (Figure 1, Figure 3 and Figure 5b).

3.6. Sensory Mapping and Description of the Samples from the 2020 Vintage

Similarly to the 2019 data, results from the free sorting task of the wines from the 2020 vintage were initially sorted in a similarity matrix and again subjected to Multidimensional Scaling (MDS) analysis. The resulting two-dimensional product space gave two clear clusters of products: (1) A big cluster containing most wines made from Savvatiano grapes from Attica, and (2) a small but tight cluster of two wines made from Vidiano grapes originating from Heraclion, Crete (Figure 4b). The remaining four (4) wines, two from Vidiano and two from Savvatiano grapes, were more dispersed in the product space, even though they were all positioned on the positive score side of the 1st dimension (Figure 4b). Specifically, Vidiano from Kavala was the closest to the small Vidiano cluster of BID-METX and BID-DAF products, while the third Vidiano product from Heraclion, Crete, (BID-ASIT), even though close to the other Vidiano products in the 1st dimension, was separated from them in the 2nd dimension and appeared closer to the two Savvatiano products that were standing out of the big Savvatiano cluster (SAB-KERT, SAB-K20, Figure 4b).

In terms of the attributes given to the wines by the panel during the UFP exercise, clustering of three out of the four Vidiano wines was apparent also there, with products BID-METX, BID-DAFN and BID-KABL appearing close together in the product space and linked to a higher number of associations to alcohol, full body, petroleum, floral honey, spicy, roasted nut and burnt wood–smoked attributes (Figure 6a). The fourth Vidiano (BID-ASIT)—appearing close to the above cluster—was also clustered with three Savvatiano products from Attica, all linked to mentions of minerality—saltiness, acidity, vegetative, dried herbs, and yellow fruits—but also some chemical–pharmaceutical and sulfury flavors (Figure 6a). On the other side of the product space, we had four Savvatiano products from Attica, with a high number of mentions of overall fruitiness, citrus fruits and floral notes, but some also linked to some oxidized and moldy mushroom and caramelized attributes (Figure 6a). Specifically, predominantly the SAB-MARK wine was linked to the moldy mushroom and oxidized notes, followed by the commercial product SAB-K 20. On the other side of that group, SAB-PAIN and SAB-KATZ were with higher mentions of overall fruitiness, citrus fruits, lemon and a few caramelized mentions (Figure 6a).

The Savvatiano grape has previously been assessed in terms of the sensory characters it gives to still wines in the work of Lola and co-workers [39]. In that study, the aim was the characterization of the Protected Geographical Indication (PGI) of Savvatiano wines from different regions of Attica. There, they employed a vocabulary of ten aroma terms in a five-point intensity scale and were able to successfully group the wines in three distinct groups. The results of our study are in accordance with that work, where the terroir perspective was addressed mainly on different soil types with resulting variation in the intensity of fruity attributes and the presence of minerality and herbaceous aromas related to the vegetative, dried herb terms used by our panel ([39]; Figure 6a).

3.7. Relationship Between Volatile Markers and Sensory Data

The Moschofilero wines consistently clustered together in both the free sorting task and UFP-derived Correspondence Analysis, reflecting a shared aromatic fingerprint characterized by floral and fruity notes. This can be linked to the presence of volatile compounds such as isoamyl acetate and diethyl succinate imparting fruity aromas, alongside phenylethyl alcohol and tyrosol, contributing to floral notes. The clustering of the Mantineia PDO wines further highlights the PDO-specific expression of Moschofilero’s aromatic potential. Malagousia wines, similarly well-clustered, were characterized by smoky, woody, and vegetative aromas. These notes align with the presence of 1-hexanol, contributing herbal green aromas, and fatty acids such as hexanoic acid and its ethyl ester, known to impart cheesy and fruity nuances, respectively. These findings reinforce the strong varietal expression of Malagousia and suggest that volatile fatty acid derivatives play a pivotal role in defining its sensory character. Assyrtiko wines, in contrast, displayed significant variation in both sensory and volatile profiles. Their diverse clustering in the MDS and CA spaces reflects variability in aromatic markers. For instance, ASY-KARD’s proximity to Moschofilero and Malagousia wines, and its characterization by spicy and herbal notes, may be due to elevated levels of 1-hexanol and possibly higher alcohols like isoamyl alcohol (fusel, fermented notes). The separation of ASY-SANT, linked to ripe fruit, honey, and oxidative aromas, reflects differences in the abundance of esters such as hexanoic acid ethyl ester and oxidative markers like tryptophol. Overall, the integration of sensory data with key volatile markers underlines the varietal specificity in shaping wine character. The robust clustering of Moschofilero and Malagousia wines highlights their aromatic typicity, driven by specific volatile precursors, while the more variable Assyrtiko wines suggest a broader aromatic potential.

As for the varieties Vidiano and Savvatiano, Vidiano samples were more chemically and sensorially uniform, dominated by specific volatile markers such as tryptophol and hexanol. On the other side, Savvatiano samples showed more diversity, both in volatile composition and sensory perception. A closer inspection of the PCA and MDS plots for the 2020 vintage revealed possible terroir-driven influences on the volatile and sensory profiles of Savvatiano and Vidiano wines. Among the Savvatiano samples, wines originating from the Attica region—including SAB-KATZ, SAB-ERYT, SAB-STAM, and SAB-SPAG—formed a relatively tight cluster in both the PCA and MDS spaces. In contrast, Savvatiano samples from Viotia (SAB-ASKR) and Korinthos (SAB-K20) were positioned further apart from the core Attica cluster, displaying distinct volatile profiles and broader sensory divergence. Notably, wines like SAB-KERT and SAB-MARK, although also from Attica, appeared as outliers in the PCA and MDS plots, indicating significant intra-regional variability potentially attributable to microclimatic or vineyard-level effects. These patterns suggest that Savvatiano may exhibit higher sensitivity to terroir, with its volatile expression being modulated by site-specific environmental conditions. This is in accordance with the findings of Lola and co-workers [39]. In contrast, Vidiano wines showed greater chemical and sensory homogeneity, especially those from Heraklion, Crete (BID-DAFN, BID-METX, BID-ASIT). These samples were closely clustered in both PCA and MDS plots. However, Vidiano samples from northern regions such as Kavala (BID-KABL) and Drama (BID-DRAM) were located slightly further from the core group, suggesting that terroir may influence volatile expression in Vidiano, even though its overall aromatic profile remains more varietally consistent than that of Savvatiano. These findings support the hypothesis that Savvatiano is more terroir-expressive, whereas Vidiano demonstrates stronger genetic determinism in its volatile composition under standardized winemaking conditions.

4. Conclusions

The main objective of this study was to assess the aromatic and overall sensory potential of a set of white wines made from distinctive Greek grapes and various regions of origin, vinified in a standardized way. In parallel, our hypothesis was that for both aromatic and overall sensory aspects concerned, we could observe sample differentiation by identifying and measuring quantitatively several key marker volatiles linked to the original amino acid composition in the grapes and its metabolism. Last, this is the first scientific attempt to sensorially profile and characterize the indigenous Cretan variety Vidiano, compared to other white cultivars. In terms of both aromatic profiles—determined by a choice of amino acid-related markers—and overall characters, the varieties that gave clear clustering and demonstrated a distinct fingerprint were Moschofilero, Malagouzia and Vidiano. Savvatiano products also gave clear clusters even though there was a clear terroir effect observed. The Assyrtiko wines used in the study did not cluster well either from their GC or sensory profiles, revealing the pivotal effect of terroir on the aromatic expression of that variety. By combining chemical analysis with organoleptic data, this study provides a scientific framework for differentiating Greek white wines based on terroir, helping to clarify the relationship between amino acid composition, volatile and overall sensory profile of those wines. This integrated approach contributes to a better understanding of Greek white wine typicity and lays the groundwork for future research into regional classification, varietal authenticity, and potential enological interventions that could enhance aromatic complexity.