Three traditional wineries recognised for the high quality of their products were selected from the “Moriles Alto” subzone, classified as a premium wine-growing area within the Montilla-Moriles PDO in Andalusia, southern Spain (Montilla-Moriles, 2025b). The wineries are located at the following GPS coordinates: Lagar Casablanca (LC): 37°26′31″ N, 4°36′41″ W (425 m above sea level); Bodegas El Monte (EM): 37°26′23″ N, 4°37′14″ W (425 m ASL), and Bodegas Doblas-Martos (DM): 37°26′09″ N, 4°36′44″ W (385 m ASL). Each winery maintains its own vineyards nearby, where only PX grapes are grown for traditional wine production. This grape variety is well adapted to the soil and climate conditions of this subzone. The dominant pruning system used in these vineyards is the vertical trellis, although the traditional head pruning is used for the most ancient vines (Palma-López et al., 2024).

The Moriles Alto subzone is characterised by its white, calcareous “albariza” soils, whose parent materials are marls and limestones, with high calcium carbonate content and low levels of organic matter, non-crystalline Fe oxides and available Fe content. The most representative soils of this subzone are those of Lagar Casablanca, classified as Calcixerepts according to the Soil Surveys Staff (USDA, 2022) by Palma-López et al. (2024).

The geographical region within which the wineries are located has a Csa climate type, according to the Köppen classification system, characterised by hot, dry summers and mild, wet winters. This climate type is identified by average monthly temperatures above 10 °C for at least four months, with the warmest month exceeding 22 °C. Data provided by the nearest state meteorological agency (AEMET, 2023) indicated that, for 2021 and 2022, the average maximum temperatures were 24.6 °C and 25.8 °C, respectively, with absolute maxima of 46.1 °C and 43.1 °C. The average minimum temperatures were 11.5 °C and 12.2 °C, with absolute minima of –3.5 °C and 0.3 °C. Total annual precipitation for these years was 330 and 427 mm, respectively.

PX grapes were harvested manually from two consecutive vintages (2021 and 2022) in September, when they reached a sugar content above 250 g L ^-1 in order to yield wines with an alcohol content of 14.5 % v/v or higher. Wines were produced using grapes from vineyards whose vines were between 20 and 25 years old. The grape yield was about 6300 kg/ha in the 2021 harvest and 5400 kg/ha in the 2022 harvest. These values are below the maximum yield of 7000 kg/ha established for this grape variety within the zone by the regulations of Montilla-Moriles (2025a), consistent with traditional practices aimed at maximising grape quality through yield restriction. Vinification was carried out on an industrial scale at each winery using traditional practices, also according to these PDO rules, which include the extraction of 70 L of must per 100 kg of grapes by gentle pneumatic pressing, followed by pH adjustment with tartaric acid to approximately 3.3 and a settling period of 24 to 48 hours. Alcoholic fermentation was conducted at a controlled temperature of 18 ± 2 °C in 50 – 80 hL stainless steel containers with indigenous yeasts, using the traditional pied de cuve method. This procedure allows the natural selection of yeasts present in the grape microbiome of each vineyard, and no active dry yeasts were used as a starter culture. The pied de cuve methodology ensures that suitable yeast strains linked to each vineyard are responsible for fermentation. Additional grape must fractions were incrementally added during the exponential growth phase of yeast activity in alcoholic fermentation. The advantages of this method include improved yeast growth, nutrient complementation, and maintenance of the fermentation activity until the containers reach full capacity. Typically, this method ensures the complete fermentation of sugars and the production of dry wines with a high alcohol content; however, the alcoholic fermentation process generally lasts about one month.

From each year, three biological replicates of young wines, taken from three independent tanks (4500 – 6000 L), were collected per winery and vintage following the natural cold stabilisation and clarification period (post-winter). Samples (750 mL per replicate) were transported to the laboratory in portable coolers with ice and stored at 3 ºC until analysis, which was conducted within one week. All analyses were performed in triplicate.

The primary characterisation of wines was based on the analysis of standard oenological parameters, including ethanol content, total and volatile acidity, pH, density, and residual sugars. These parameters were quantified according to the analytical methods recommended by the International Organisation of Vine and Wine (OIV, 2025b). Lactic and malic acid contents were determined using the Merck reflectometric method with a Reflectoquant™ system (Merck ^®, Darmstadt, Germany) according to the procedure provided with the equipment. Absorbances at 280 and 420 nm were measured in a Cary 60 UV-Vis spectrophotometer (Agilent Technologies, Santa Clara, CA, USA).

Major volatile compounds, present in wines at concentrations above 10 mg L ^–1 (10 ppm), include the alcohols methanol, 1-propanol, isobutanol, 2- and 3-methyl-1-butanol, and 2-phenylethanol; the carbonyl compounds acetaldehyde and acetoin; and the esters ethyl acetate, ethyl lactate, and diethyl succinate. In addition, the polyols 2,3-butanediol ( levo and meso isomers) and glycerol were also quantified. All these fourteen compounds were determined by direct injection of wine following previously published methodology (Peinado et al., 2004; Fuentes-Espinosa et al., 2025a; Fuentes-Espinosa et al., 2025b; Alcalá-Jiménez et al., 2025). Briefly, the sample preparation consisted of adding 1 mL of an internal standard solution (4-methyl-1-butanol at 1.018 g L ^–1) to 10 mL of wine and 0.2 g of pure calcium carbonate. After sonication and centrifugation, 0.7 µL of the supernatant was directly injected into the injector (in split mode) of an Agilent 6890 GC equipped with an FID and a CP-WAX 57 CB (60 m, 0.25 mm, 0.40 µm capillary column), both from Agilent (Santa Clara, CA, USA). Target compounds were quantified using calibration curves obtained by subjecting standard solutions of pure compounds whose concentrations vary from 0 to 5 times the average values described in the literature, and using the same procedure as the samples (Peinado et al., 2004). The chemicals were provided by Merck and Sigma-Aldrich.

The methodology used for identification and quantification of minor volatile compounds (which are those with a content of less than 10 mg L ^–1) was described in detail in a previous work (Palenzuela et al., 2023). The volatiles were extracted using a Polydimethylsiloxane (PDMS) coated stir bar (Twister ^® 10 mm, 0.5 mm film thickness from Gerstel, Mülheim an der Ruhr, Germany). The Twister ^® was placed in a 10 mL vial containing 1.0 mL wine, 0.1 mL of internal standard solution (0.4116 g L ^–1 of hexyl butyrate in absolute ethanol), and 8.9 mL of a pH 3.5 buffered solution prepared with 12 % v/v ethanol, 2.6 g L ^–1 tartaric acid and 2.2 g L ^–1 potassium bitartrate. After stirring for 120 min at 1200 rpm and 20 °C, using a Variomag multipoint magnetic stirrer (Thermo Fisher Scientific, Waltham, MA, USA), the Twister was rinsed with distilled water, dried, and transferred to a thermal desorption unit (TDU) coupled to the GC for desorption and to transfer the volatiles to an HP-5MS fused silica column (60 m, 0.25 mm, 0.25 µm) from Agilent Technologies (Santa Clara, CA, USA).

Gas chromatography was performed on an Agilent 7890A GC coupled to a 5975C MSD detector and a Gerstel MPS autosampler, controlled by ChemStation v.02.02.143 and the Maestro v.1.2 software, from ChemStation International, Inc., Dayton, OH, USA, and Gerstel Mülheim an der Ruhr, Germany, respectively. Compound identification was carried out by comparing mass spectra of each compound with those of the NIST08 and Wiley7 libraries and further confirmed using the mass spectra of pure compounds. Quantification was based on calibration curves obtained from standard solutions of pure compounds, with concentrations ranging from 0 to 5 times the average values described in the literature and detailed in a previous article by Palenzuela et al., 2023. The chemicals used as standards were supplied by Sigma-Aldrich and Merck.

Data matrices obtained for wine characterisation and volatile compounds (major and minor) were subjected to a two-way Analysis of Variance (ANOVA) to evaluate the effects of vintage, winery and their interaction. Fisher’s Least Significant Difference (LSD) test was used for mean comparison ( p ≤ 0.05) using IBM SPSS Statistics software v. 29.0.2.0 (IBM Corp., Armonk, NY, USA). Furthermore, a Multivariate Analysis of Variance (MANOVA) and a Principal Component Analysis (PCA) were performed using Statgraphics Centurion (v. 16.1.11) (Statgraphics Technologies, Inc., The Plains, Virginia, USA). For PCA, data were auto-scaled (normalised) to reduce dimensionality and identify key volatile compounds with discriminant power regarding winery and vintage influence.

Table 1 summarises the average values, standard deviations and HGs at the p ≤ 0.05 significance level for the wine samples from the three wineries and two vintages. All wine samples met the Montilla-Moriles PDO normative for Fino-type wines (Montilla-Moriles, 2025a). The vintage was the dominant factor influencing general parameters, particularly ethanol content (higher in 2021) and absorbance at 420 nm (higher in 2022). In contrast, the winery factor, likely reflecting winemaking practices, primarily affected total acidity, volatile acidity and absorbance at 280 nm, with the exception of LC winery, where this index remained stable across years

The high variability in malic and lactic acid concentrations is attributable to both initial grape composition and the elective occurrence of malolactic fermentation (MLF). Because the bacteria responsible for MLF are highly susceptible to pH and ethanol values, the process is significantly modulated by discretionary oenological practices. Two-way ANOVA (vintage, winery, and interaction) confirmed that most variables, except density, were significantly influenced by the interaction ( p ≤ 0.001) between vintage and winery (Table S1), aligning with previous studies in young PX wines from other production zones in the same PDO (Fuentes-Espinosa et al., 2025a; Fuentes-Espinosa et al., 2025b; Alcalá-Jiménez et al., 2025).

Table 2 lists the average concentrations and standard deviations of 11 volatile compounds and 3 polyols, as well as the HGs at p ≤ 0.05. Several of these compounds exhibit significant statistical differences among wineries and vintages.

The concentrations and standard deviations of 51 minor volatile compounds, grouped by chemical families, are listed in Table 3. All of them exhibit concentrations at ppb levels (µg L ^–1), with the exception of γ-butyrolactone. These volatiles are mainly formed by yeast metabolism during alcoholic fermentation through several biochemical pathways involving amino acid catabolism, fatty acid synthesis, and esterification. Due to their low odour detection thresholds and pleasant nuances, they have a significant influence on wine aroma (Palenzuela et al., 2023; Carpena et al., 2021; He et al., 2023).

The MVA fingerprints (Figures 3 and 4) provide a global visualisation of how yeast metabolism responded to the studied factors, revealing a clear separation between vintages.

The 2022 vintage was characterised by higher relative intensities of higher alcohols (isobutanol, 2- and 3-methyl-1-butanol), ethyl esters, acetates, and terpene derivatives, confirming a more vigorous fermentation and enhanced esterification pathways. In contrast, the 2021 wines showed a profile shifted toward higher relative values of acetaldehyde, ethyl lactate, glycerol, lactones, furanic compounds, and carbonyl derivatives, alongside a moderate increase in volatile phenols. This fingerprint suggests slower fermentation kinetics and stress-related pathways in yeast cells, as well as a partial shift towards oxidative metabolic routes potentially linked to variations in must clarification or air exposure.

1: acetaldehyde; 2: ethyl acetate; 3: methanol; 4: 1-propanol; 5: isobutanol; 6: 2-methyl-1-butanol; 7: 3-methyl-1-butanol; 8: acetoin; 9: ethyl lactate; 10: 2,3-butanediol (levo); 11: 2,3-butanediol (meso); 12: diethyl succinate; 13: 2-phenylethanol; 14: glycerol. Each sunray shows the value for each variable scaled from the mean minus three times its standard deviation at the origin to the mean plus three times its standard deviation at the extreme. DM: Doblas Martos, EM: El Monte, and LC: Lagar Casablanca.