Twenty-seven Moscatel de Setúbal fortified wines, representing vintages 1985 and 1995–2020, from the Setúbal PDO (Figure 1), provided by the wine company Bacalhôa Vinhos de Portugal, S.A., were analysed.

Grapes from all vintages were harvested from late August to late September. of the corresponding year, from the Península de Setúbal PDO wine region/vineyards. Wines were produced following the legally authorised vinification technology as schematised in the literature (Granja-Soares et al., 2025). Briefly, for all vintages, grapes were crushed and fermented in stainless steel tanks, and fortification with 96 % (vol.) vinous alcohol was performed four days after their reception. The maceration period, with grape skins and seeds and fortified wine, lasted approximately six months, with pumping over once a day.

The free run and press wine were fined/clarified separately using 1 g/L bentonite, decanted, subsequently blended and transferred into partially filled used French oak barrels of 225 L from Quercus petreae L., previously used for the maturation of red wine during five/seven years. Those were stored for the first fifteen years in a metal-roofed room, exposing the wines to high air thermal amplitudes that can range between 5 and 40 ºC (Garcia, 2001), and after that moved to a place with low thermal amplitudes. The barrels were not topped during the ageing period. Table 1 provides information on the vineyards of grape origin and summarises the general physicochemical composition of the wines at the time of barrel filling.

Vintage	Alcoholic strength by volume at 20 °C (% vol.)	Volatile acidity (g of acetic acid/L)	Titratable acidity (g of tartaric acid/L)	pH	Reducing substances (g/L)	Density at 20 ºC (g/mL)
1985	19.2	0.87	7.10	3.40	212.0	1.0681
1995	19.1	0.90	6.00	3.48	190.0	1.0595
1996	19.0	0.96	6.20	3.46	175.0	1.0541
1997	19.2	0.90	7.70	3.29	181.0	1.0560
1998	18.7	0.78	6.30	3.50	180.0	1.0560
1999	19.1	0.72	6.80	3.49	186.0	1.0580
2000	20.3	0.48	6.90	3.36	213.0	1.0672
2001	18.1	0.72	6.90	3.44	218.0	1.0715
2002	18.8	0.75	7.80	3.21	180.0	1.0560
2003	20.3	0.90	6.80	3.39	203.0	1.0631
2004	18.3	0.66	6.60	3.33	195.0	1.0625
2005	17.6	0.72	4.90	3.46	146.0	1.0440
2006	17.9	0.99	5.70	3.39	136.0	1.0400
2007	18.4	0.51	5.60	3.29	144.0	1.0425
2008	18.3	0.54	5.60	3.46	164.0	1.0505
2009	17.1	0.18	5.10	3.34	158.0	1.0495
2010	18.4	0.24	4.90	3.40	160.0	1.0485
2011	17.8	0.66	6.40	3.53	176.0	1.0485
2012	18.0	0.36	5.60	3.36	148.0	1.0445
2013	18.5	0.51	5.80	3.41	148.0	1.0431
2014	19.0	0.48	5.60	3.54	150.0	1.0440
2015	19.6	0.54	5.60	3.41	178.0	1.0544
2016	20.2	0.36	6.40	3.37	209.0	1.0656
2017	17.2	0.33	5.00	3.51	158.0	1.0494
2018	17.5	0.30	4.00	3.72	152.0	1.0467
2019	18.1	0.26	5.60	3.60	156.0	1.0475
2020	17.1	0.24	6.00	3.55	166.0	1.0525

In May 2023, samples representative of each vintage included in this study were collected. Each sample consisted of a blend of wines obtained from the surface of seven randomly selected barrels, without stirring. The wines were stored at 20 ºC and protected from light before analysis.

A significant effect of vintage year on general physicochemical parameters of the wines was verified (Table 2), with the exception of the 2007 vintage due to practical constraints. The vintage year denotes the growing season corresponding to the grape harvest (IVV, 2025), which is climate-dependent (Fraga et al., 2023). The wines of the oldest harvests presented, in general, higher values of density, alcoholic strength by volume and glucose plus fructose content, thus having higher values of total dry extract and sugar-free extract, as well as higher values of volatile and fixed acidity, and consequently, titratable acidity. Conversely and congruently, these denote lower values of pH. It should be emphasised that this trend was already observed in the wines prior to ageing. However, the 2011 vintage combined the highest volatile acidity (1.57 g acetic acid/L) with one of the highest pH values (3.57). According to the Caderno de Especificações (IVV, 2026) wines must have between 16 and 22 % (vol.) of alcoholic strength by volume; volatile acidity ranging up to 1.5 mg or up to 1.8 mg of acetic acid/L, if the wine has until or more 10 years, respectively – since by chemical oxidation ethanol is successively transformed into ethanal and acetic acid (Ribéreau-Gayon et al., 2006) and reducing sugars between 280 and 340 g/L if the wine has until or more than 20 years. Franco and Singleton (1984) found similar results in their study, following the tendency for higher volatile acidity and sugars in the older vintages. Nevertheless, this should be regarded as a tendency rather than a rule, likely resulting from one or more phenomena, such as concentration by evaporation, extraction from the wooden barrel, or initial differences among wines from different vintages, as authors pointed out.

In a study conducted with Madeira wines (other Portuguese Generoso wine), it was observed that the levels of acetic acid and ethyl acetate increased during wood ageing and were strongly influenced by the degree of sweetness, with sweeter wines exhibiting higher concentrations of both compounds (Miranda et al., 2017). With respect to sulphur dioxide, for which the legal limit is 200 mg/L (IVV, 2026), higher values of total sulphur dioxide were observed in some of the oldest vintages; however, all values remained well below the legal limit. According to Franco and Singleton (1984), sulphur dioxide addition to grapes occurs prior to crushing and is dependent on sanitary conditions, with amounts reaching up to 30 mg/L. Moreover, turbidity was evaluated, and wines from older vintages generally exhibited turbidity values equal to or less than 2 NTU, with the vintages of 1997 and 1998 showing values under 1 NTU and can therefore be considered clear (Ribéreau-Gayon et al., 2006). An exception is the wine of 2017, with 97.5 NTU. White wines naturally contain grape proteins, which can cause cloudiness and deposits if they flocculate. After alcoholic fermentation, the wine is not clear, showing some cloudiness due to yeast lees, proteins, pectins and other polysaccharides, phenolic compounds and salts (Ribéreau-Gayon et al., 2006; Marchal & Waters, 2010). In a study conducted by Mesquita et al. (2001), involving single-varietal wines from the Moscatel grape cultivar and other Portuguese single-varietal wines, higher protein concentrations and greater susceptibility to the development of protein cloudiness were observed in wines from the Moscatel grape cultivar. Silva (2011) investigated the turbidity that may occur after bottle opening and concluded that it results from tannin polymerisation upon exposure to oxygen. As barrels are small-volume containers, they facilitate the natural sedimentation of unstable colloidal matter in wine, as well as high molecular weight constituents (from condensation and polymerisation phenomena, involving phenolic compounds, proteins and polysaccharides), contributing to the stability and clarity of the wine, resulting in lower turbidity (Ribéreau-Gayon et al., 2006). Thus, the lower turbidity values observed in wines with longer ageing times can possibly be explained by the phenomena that occur in the barrel during wine ageing. Nevertheless, the possibility that accidental stirring during barrel sampling caused sediment resuspension cannot be excluded.

For each vintage year and parameter, the results are expressed as mean ± standard deviation of the corresponding analytical replicates. Means followed by the same letter in a column are not significantly different at the 0.01** or 0.001*** level of significance. For volatile acidity and turbidity parameters, statistical analysis was conducted using the Kruskal–Wallis test.

The chromatic characterisation of the wines was first assessed by measuring absorbance at 420 nm (Figure 2). This wavelength corresponds to the maximum absorbance of yellow and brown pigments, which are widely used as an indicator of browning in wines (Singleton & Kramling, 1976; Chaves et al., 2007). As expected, the fortified wines in this study subjected to longer oxidative barrel ageing exhibited higher absorbance values (ranging from 1.318 to 7.264 a.u.), with the highest absorbance, and consequently the most intense colour, observed in the oldest sample, from the 1985 vintage (Figure 2). According to Comissão Vitivinícola Regional da Península de Setúbal (CVRPS, 2025), aged and superior Moscatel de Setúbal wines, matured for five years or more, display a golden hue that varies from topaz to amber. The emergence of this colour, evidenced by the increase in absorbance at 420 nm, may be primarily attributed to the duration of barrel ageing. As previously mentioned, during this period, a slow and continuous permeation of oxygen through the wood pores and diffusion in the wine promotes the gradual oxidation of phenolic compounds, leading to the progressive development of brownish tones (Chaves et al., 2007; Bravo, 2008; Del Álamo-Sanza & Nevares, 2018; Pfahl et al., 2021).

The progressive increase in absorbance at 420 nm from younger to older vintages, and the corresponding shift towards brownish hues, may also result from the formation of melanoidins (Tredoux & Silva Ferreira, 2012; Oliveira, 2014; Starowicz & Zieliński, 2019). This hypothesis is supported by the research of Milheiro et al. (2023) in white Port wines, where a positive correlation was observed between 5-hydroxymethylfurfural levels and the presence of small polymeric pigments absorbing at 420 nm, which increased with wine age (10, 20, and 40 years). These results suggest that browning in fortified wines may largely arise from sugar-derived reactions such as Maillard and caramelisation processes, rather than phenolic compound oxidation.

A high correlation between absorbance at 420 nm and glucose plus fructose concentrations (y = 22.348 x + 165.31, R² = 0.7783) was observed, with wines from the 1985, 2000, and 2001 vintages showing the highest values. Nevertheless, these observations should be interpreted with caution, as the direct effect of glucose and fructose on colour was not specifically investigated within the scope of the present study. Therefore, further investigation would be valuable to clarify the potential contribution of these sugars to colour development.

Furthermore, the colour intensity may be enhanced during the extended maceration period that precedes barrel ageing, which promotes the diffusion of phenolic compounds and may contribute to the yellowish hues characteristic of these white wines (Feliciano et al., 2009). In the present study, two of the highest absorbance values correspond to the highest total phenols content, above 600 mg of GAE/L. Nevertheless, only a low correlation was observed between colour and total phenolic content (y = 46.125 x + 326.72, R² = 0.4072)

Studies on Sherry wines reported values between 1.27 and 1.57 a.u. for Oloroso wines aged for 7 years (Fabios et al., 2000) and around 2.2 and 4.4 for sweet Pedro Ximénez wines aged for 1.3–11.5 years, and 1.2 a.u. for non-aged wines (Chaves et al., 2007). In this study, Moscatel de Setúbal aged in barrels for 7 years (2015 harvest) presented a higher absorbance value at 420 nm (2.664 a.u., while the 2016 harvest aged for 6 years presented a lower value (1.32 a.u.), similar to the values recorded for Jerez Oloroso wines subjected to a similar ageing period.

Regarding the Portuguese Generosos, in the study by McCallum et al. (2019) on Carcavelos fortified wines, aged in barrels for 8 years, absorbance values of between 0.939 a.u. and 1.211 a.u. were obtained at 420 nm. In contrast, Moscatel de Setúbal fortified wine aged in barrels for 8 years (2014 harvest) exhibited a higher absorbance at 420 nm (1.712 a.u.). In the study by Milheiro et al. (2023), in white Port wines with an age indication of 10 and 20 years, the absorbance at 420 nm was on average 0.361 a.u. and 0.311 a.u., respectively. In comparison, the absorbance values observed in Moscatel de Setúbal fortified wines aged in barrels for 10 years (2012 harvest) and 20 years (2002 harvest) were 1.963 a.u. and 3.116 a.u., respectively (Table 4). The absorbance values measured in Moscatel de Setúbal fortified wines are higher than those observed for Carcavelos and white Port fortified wines, which can be explained by the possible enrichment in colour due to the long maceration period involved in its winemaking technology, or by the formation of pigments through Maillard reactions and the caramelisation of sugars (Oliveira, 2014; Milheiro et al., 2023; IVV, 2026). The findings of the present study are, however, in line with those from a study on Madeira fortified wines, specifically Malvasia (white cultivar) reported by Pereira et al. (2013), who performed chromatic characterisation by measuring absorbance at 420 nm after subjecting the wines to a warming process (0, 1, 2, and 3 months at 45 °C, and 1 month at 70 °C). The heating process was shown to accelerate the kinetics of reactions typically occurring during long-term ageing; consequently, wines subjected to thermal treatment exhibited chromatic characteristics similar to those of wines aged naturally for several years. The absorbance values at 420 nm after heating at 45 ºC for 2 months are comparable to those obtained for Moscatel de Setúbal fortified wines aged for 2 years in barrels (2020 harvest), suggesting that thermal treatment contributes to the development of similar chromatic features as thermal amplitude suffered by Moscatel de Setúbal subject to ageing with high thermal amplitudes. Moreover, the absorbance recorded for Madeira wine after 1 month at 70 °C was substantially higher than in the other treatments and closely resembled that of Moscatel de Setúbal fortified wines aged for 10 years in barrels (2012 harvest), reinforcing the role of heating/high thermal amplitudes in accelerating browning reactions.

The CIElab method is among the most widely applied approaches for assessing the colour characteristics of wines and monitoring their evolution (Pérez-Magariño & González-Sanjosé, 2003), being an official method (Resolution Oeno 1/2006) of OIV (OIV, 2025). This approach has previously been used to assess colour in Madeira wines (Carvalho et al., 2015), Porto wines (Milheiro et al., 2023) or Sherry wines (Chaves et al., 2007), but to the best of our knowledge, there are no scientific studies regarding Moscatel de Setúbal. The results obtained from the CIELab system for the samples of Moscatel de Setúbal fortified wines are shown in Figure 3. Analysing parameter by parameter, we can observe a decrease in brightness (L*) with the years of ageing. The minimum value (87.7) was observed in the oldest wine (1985), and the maximum value (98.5) was observed in the wine from the 2013 harvest, with a noticeable trend towards darkening as the harvest year becomes older (Figure 3). As for the red/green component (a*), with increased barrel ageing, the red component decreases (in most cases) and gives way to the green component. In fact, most wines from more recent harvests have reddish tones, while wines from older harvests mostly have greenish tonality (Figure 3). The general trend towards a decrease in a* values during wine ageing may be related to the oxidation reactions of tannins, with the formation of new polymeric complexes, which are also present in white wines, but generally in smaller quantities than in red wines (Pereira et al., 2013; Teissedre & Jourdes, 2013; Pereira et al., 2022) or in white wines subject to maceration. This trend has been observed in white fortified Madeira wines (Pereira et al., 2013; Pereira et al., 2022).

In the yellow/blue component (b*), all values obtained are higher than zero, meaning that all wines have only a yellow component (Figure 3). The yellow component intensifies with older vintages having a yellow component 30 units higher, as in the case of the 1985 vintage (40.16), when compared to the yellow component of wines from more recent vintages, such as the 2017 vintage (9.44).

This trend was also observed in white fortified Madeira wines in the studies by Pereira et al. (2013) and Pereira et al. (2022) and has been associated with the oxidation of phenolic compounds and the formation of new polymeric complexes. Pereira et al. (2013) found that in fortified Madeira wine, yellow pigments were always predominant, as expected for a white grape cultivar.

The total colorimetric difference (ΔE*) was also calculated, based on the average of L*, a* and b* parameters as follows: ΔE* = [(ΔL*)² + (Δa*)² + (Δb*)²]^1/2. Generally, the human eye can discriminate between two colours when ΔE ≥ 1. However, when wine is observed through a glass, colour discrimination is reduced, and the visual threshold may increase to differences of up to approximately ΔE* = 5 (Pérez-Magariño & González-SanJosé, 2003). Consistently, in a study with Madeira wine, Pereira et al. (2013) considered colour differences to be higher than 3 units perceptible by humans.

The results (Table 3) revealed that the 1985, 2000 and 2001 vintages exhibited the most distinct colour profiles, showing higher values of b*. It is also possible to observe a separation between older and younger wines, denoted by darker colour cells indicating greater differences. Thus, according to the results obtained in this study, and bearing in mind the colour difference of 3 units, it was easy to discriminate colour with the human eye, for example between 1985, 2000 and 2001 and all the other years, but it was not possible to discriminate between 2000 and 2001 with each other or between 2018, 2019 and 2020 vintages. Thus, the results indicate that the chromatic variations observed among the samples are sensorially relevant. ΔE values exceeding the threshold of 3 units reflect significant colour evolution. These results suggest that both vintage and ageing conditions influence the chromatic characteristics of the wines, leading to perceptible differences in colour that may affect visual quality assessment.

For each vintage year and parameter, the results are expressed as mean ± standard deviation of the corresponding analytical replicates. Means followed by the same letter in a column are not significantly different at the 0.01** or 0.001*** level of significance. n.s. – non-significant differences. Columns without letters indicate that the statistical analysis was conducted using the Kruskal–Wallis test.

The total phenol contents found in the different samples of Moscatel de Setúbal ranged from 257.84 mg GAE/L (2016) to 717.84 mg GAE/L (2001) (Table 4). Figure 4 shows that older Moscatel de Setúbal wines exhibit slightly higher values than more recent vintages.

As previously mentioned, during the ageing stage of Moscatel de Setúbal fortified wine, it is placed in used oak barrels, which facilitate oxygen permeation and diffusion (Pasteur, 1873; Cutzach et al., 1999; Cutzach et al., 2000). Although successive use of the barrel reduces its capacity to enrich the wine with volatile compounds and hydrolysable tannins, the contribution of small amounts of hydrolysable tannins and other phenolic compounds, both volatile and non-volatile, cannot be completely excluded, even if their contribution is expected to be negligible (Cutzach et al., 2000; McCallum et al., 2019; Zamora, 2019). In addition, the potential contribution of volatile phenols remaining from beverages previously aged in the barrel should also be considered (Coelho et al., 2019). Also, the phenolic composition of grapes varies between vintages, since grape characteristics are influenced by the climatic conditions of each respective year (De Oliveira et al., 2019).

In addition to sweet fortified wines traditionally being aged in used barrels, these are often partially empty due to the evaporation of water and ethanol through the pores of the wood, and in cellars with variable temperatures (Pasteur, 1873; Cutzach et al., 1999; Cutzach et al., 2000; Gómez-Plaza & Bautista-Ortín, 2019; Pfahl et al., 2021). The evaporation of water and ethanol has the effect of concentrating the constituents present in the wine, and this phenomenon is possibly responsible for the trend observed in the total phenols content of Moscatel de Setúbal fortified wines. Indeed, it is known that there are no two equal barrels and the evaporation is, among other factors, dependent on it (construction, vedation) (Del Álamo-Sanza & Nevares, 2018; Pfahl et al., 2021). In addition, the differences in total phenol values may also be due to the harvest year (state of grape ripeness at harvest) and the variable maceration time, since these are located mostly in the solid parts of the bunch, and their transfer to the must and wine is mainly determined by maceration during the winemaking process (Sun et al., 1999; Cheynier, 2005; Feliciano et al., 2009). Franco and Singleton (1984) determined the total phenols content of 24 fortified Moscatel de Setúbal wines aged between 0 and 77 years in used oak barrels. The values ranged from 649 mg expressed as gallic acid/L in wines aged for 8 years to 718 mg expressed as gallic acid/L in wines aged for 41 years. When compared with those presented in Table 4, these values reveal the same trend as well as a similar order of magnitude, with the wine aged for 8 years (2017 harvest) presenting 276.56 mg expressed in gallic acid/ L and wine aged for 40 years (1985 harvest) 604.60 mg/L expressed in gallic acid. These values are slightly lower than those reported by Franco and Singleton (1984). Bravo et al. (2006), analysing wines from different producers, found values between 400 mg/L expressed in GAE/L (7 months of maceration) and 1700 mg/L expressed in gallic acid (8 months of maceration). Higher values were obtained in the wines from grapes harvested in earlier times and without complete destemming. Feliciano et al. (2009), in a study involving different producers and different vintages, found values between 796 and 1863 mg/L expressed in gallic acid. The authors suggest that this difference can be attributed to different winemaking techniques, but it must also be seen that the RSD associated with the mean values is, in the first case, 25.8 % and, in the second, 37.3 %. Mean values result from the analysis of different vintages from the same producer, so it can possibly be said that vintage year is also a factor of variability. However, a direct comparison between the results of this study and those reported in the literature is not possible, since, despite some methodological variations, most other authors used the Folin–Ciocalteu method.

According to Feliciano et al. (2009), the Folin–Ciocalteu method is widely used for the evaluation of total phenolic content in wines. However, it exhibits low selectivity owing to interference from non-phenolic compounds that can react with the Folin reagent, as well as sugar interference (Ricardo-da-Silva et al., 1987). In the study by Milheiro et al. (2023), the phenolic compound content of white Port wines aged 10 and 20 years was approximately 160 mg gallic acid/L and 136 mg gallic acid/L, respectively. The results for total phenols observed in this study (Table 4) for Moscatel de Setúbal fortified wines aged 10 years (2012 harvest) and 20 years (2002 harvest) were 397.36 mg of gallic acid/L and 632.24 mg of gallic acid/L, respectively. These values are higher and show a contrary trend to those reported in white Port wines. However, factors such as the state of ripeness of the grapes at harvest, the grape varieties used and possibly some more interactions with proteins, polysaccharides and phenols in specific years in the case of white Port wine, the little or no maceration time (Moreira & Guedes de Pinho, 2011), as well as the time spent in barrels (phenomena of oxidation, condensation, polymerisation, and concentration that occur with ageing in barrels) may be possible explanations for the opposite trend found in this study of white Port wine. Thus, using the Prussian Blue Method, by spectrophotometric measurement at 720 nm, it is not possible to compare the results obtained for total phenols with those of other white fortified wines due to the scarcity of information in the literature.

To estimate the potential astringency of Moscatel de Setúbal fortified wines, their tanning capacity was determined. The tanning capacity values observed in this study are shown in Table 4, ranging from 3.86 NTU (2013) to 25.78 NTU (2001). As shown in Figure 5, older vintages exhibit higher and more variable tanning capacity values, whereas more recent vintages display lower values that are relatively consistent. The trend observed for this parameter may be related to the grape characteristics across vintages, inherent differences among oak barrels, and fluctuations in temperature across the years, in addition to ageing time. Changes in the polymerisation profile of condensed tannins during ageing may also contribute to the observed trend, as the tannins that remain soluble in the wine may exhibit different reactivity because of their structural evolution. Another possible explanation is the prolonged extraction of hydrolysable tannins from the oak wood, which may remain relevant even when previously used barrels are employed.

It is not possible to compare these results with those of other Moscatel de Setúbal fortified wines due to a lack of information. However, comparisons can be made with studies of other white fortified wines that have used the same method to determine the parameter. McCallum et al. (2019) studied the effect of botanical species and barrel toasting conditions on Carcavelos fortified wines aged in barrels for 8 years, observing tanning capacity values between 3.58 NTU and 0.49 NTU (converting from NTU/mL, as indicated in the original reference to Freitas and Mateus (2001), to NTU, the form of expression used in this study). These values are lower than the tanning capacity observed in Moscatel de Setúbal wine aged for 8 years (2014 harvest), equal to 4.20 NTU. In contrast to Carcavelos fortified wines, Moscatel de Setúbal has a long maceration time, during which the solid parts of the grape are extracted, leading to the diffusion of phenolic compounds, contributors to the sensation of astringency (Feliciano et al., 2009; IVV, 2026).

Une image contenant texte, capture d’écran, Police, nombre Le contenu généré par l’IA peut être incorrect.

For a better characterisation of the phenolic composition of Moscatel de Setúbal fortified wines, monomeric flavanols and condensed proanthocyanidins/tannins were determined according to their degree of polymerisation, namely: the monomeric fraction (F1), the oligomeric fraction (F2) and the polymeric fraction (F3). The monomeric flavanol contents found in the fortified wines analysed are shown in Table 4, ranging from 16 mg/L (2006) to 445 mg/L (2000), with significant differences among vintages. The monomeric fraction of the older vintages showed higher values, although they vary among themselves, unlike the monomeric fraction of the more recent vintages, which have lower values and are generally close to each other. The 1985 wine has a concentration of monomeric flavanols close to the values obtained for the most recent harvests (Figure 6).

The levels of condensed oligomeric tannins present in Moscatel de Setúbal wines are shown in Table 3. Statistically, the samples are not significantly different from each other (p > 0.05). The levels of condensed oligomeric tannins found in Moscatel de Setúbal vary between 8 mg/L (1985) and 400 mg/L (1997), with the latter value standing out from the rest as it is much higher than the average of the results obtained. In general, the older vintages show higher values when compared to the more recent ones, except for 1985, which has the lowest value of condensed oligomeric tannins (Figure 6). The levels of condensed polymeric tannins present in Moscatel de Setúbal are shown in Table 4. Statistically, the samples are not significantly different from each other (p < 0.05). The condensed polymeric tannins obtained for Moscatel de Setúbal fortified wines vary between 20 mg/L (1985) and 396 mg/L (1997). As can be seen in Figure 6, the concentration of condensed polymerised tannins tends to be lower in wines with longer ageing periods. According to Somers and Pocock (1991), white wines produced without maceration contain very low amounts of proanthocyanidins. However, the production of wine from white grapes with maceration promotes their extraction from the solid parts of the grape, resulting in wines with higher levels of these compounds, leading to their precipitation.

The dominance of the monomeric and oligomeric fraction over the polymeric fraction in older wines may be related to the occurrence of depolymerisation phenomena during barrel ageing. In fact, the depolymerisation of tannins in an acidic environment (pH 3.40), resulting in the release of monomeric units, has been reported (Haslam, 1980). Over time, tannins polymerise until they reach such a high molecular weight that they precipitate, with older wines having a lower importance of the polymeric fraction, with only tannins with low molecular weight, namely oligomeric tannins and monomeric flavanols, being soluble (Haslam, 1980; Dallas et al., 2003; Sun et al., 2013). In the case of the 1985 wine, the long ageing period of 37 years may explain the overall decrease in all flavanol fractions. Thus, there appear to be two distinct profiles among the 27 vintages of Moscatel de Setúbal fortified wine (Table 4). A first profile, covering the 1985 and 1995 to 2001 vintages, except 1997, where the monomeric fraction (F1) dominates, and a second profile, covering the 1997 and 2003 to 2020 vintages, except 2002, where the polymeric fraction (F3) dominates. While the second profile is the one most commonly observed and described in the literature, the first profile seems to reveal the consequences of prolonged ageing in used oak barrels on the degree of polymerisation of tannins, although bearing in mind also the possible influence of grape characteristics from each vintage, inherent differences among oak barrels, and fluctuations in temperature across the years in addition to ageing time.

The Principal Components Analysis provided further perspective into the variables contributing to the differentiation of the wines (Figure 7). The first principal component (PC1) explained 65.37 % of the total variance, while the second principal component (PC2) accounted for a further 12.09 %, together explaining 77.46 % of the total variance.

Une image contenant ligne, diagramme, Tracé, texte Le contenu généré par l’IA peut être incorrect.

PC1 was mainly associated with the physicochemical and chromatic characteristics of the wines. Most physicochemical parameters, with the exception of ABV and pH, together with absorbance at 420 nm, the CIELab parameters, tanning power, flavanols and total phenols, influence the special distribution of the samples along this component. The parameters exerting the greatest influence on this distribution, in descending order of importance, were dry extract, density, colour at 420 nm, glucose plus fructose, chroma (C*), b*, total acidity, fixed acidity, volatile acidity, L*, total sulphur, free sulphur, a*, monomeric flavanols (F1) and H*. The variables associated with the negative side of PC1 included total acidity, volatile acidity, sugar-free extract and colour-related parameters, particularly absorbance at 420 nm, b* and chroma (C*), as well as tanning capacity (TC), flavanols (F1) and total phenols (TPh). In contrast, L* was strongly associated with the right side of PC1. This pattern allows the differentiation between the older vintages (1985 and 1995–2001), which were generally characterised by higher colour intensity, a more pronounced yellow hue, higher chroma, higher tanning capacity and a higher proportion of monomeric flavanols, and the more recent vintages (2002–2020), which were mainly associated with higher lightness (L*).

PC2 explained 12.09 % of the total variance and was mainly associated with ABV and pH. The wines from 1997–1999 were mainly located in the upper-left region of the score plot, whereas the wines from 1985 and 1995 were positioned towards the lower-left region. Among the more recent vintages, 2006, 2010, 2012 and 2013 were mainly located in the upper-right region, while 2005, 2009 and 2015 were positioned in the lower-right region. The most recent vintages, particularly 2018–2020, were also located in the lower-right region.

Overall, the PCA suggests that the differentiation of the wines was primarily associated with their physicochemical, chromatic and phenolic characteristics, with PC1 largely reflecting the distinction between older and more recent vintages. Nevertheless, the dispersion observed within each vintage group, particularly along PC2, indicates that wine age alone does not fully explain the variability. This pattern may also reflect vintage-related differences in grape characteristics and composition that reflect environmental and climatic conditions, as well as slight variations in the winemaking conditions between vintages.