Introduction

Aging in oak barrels is a conventional technique used to improve the quality of red wines. Wine aging in oak barrels involves some complex and slow changes that have a positive impact on its flavor, aromatic complexity, and colour stability. During the aging process, the oak barrel allows a slow and continuous entry of oxygen and contributes to the gradual release of polyphenolic and volatile compounds into the wine. The content of volatile compounds, ellagitannins, hydroxybenzoic and hydroxycinnamic acids, and aldehydes of the oak heartwood varies significantly depending on the oak species and the geographical origin of the wood [1]. Wine aging in oak barrels includes many chemical reactions and changes in wine phenolic composition that may stabilize the red wine colour and decrease its astringency [2]. These reactions are due to the moderate wine oxidation in the oak barrel [3], to reactions of wine compounds with compounds extracted from the wood [4], and to adsorption of wine phenolic compounds on the wood surface [5].

Various studies on the chemical composition of oak woods have pointed out the great variability of phenolic extractable compounds as a function of their forest origin [6,7,8]. In fact, it has been observed that the content of ellagitannins allows the distinction of French and East European oak woods (Quercus robur and Quercus petraea) [6], as well as Portuguese and Spanish oak woods (Quercus pyrenaica) [9]. In addition, Feuillat and co-workers [10] also observed significant differences in the ellagitannin content from oak wood (Quercus robur and Quercus petraea) of different French forests. Geographical origin also affected the content of phenolic compounds of woods from Portuguese cooperage (Quercus pyrenaica) [7, 8].

Nevertheless, there are a limited number of studies on the influence of oak forest origin on the wine phenolic composition, being French oak woods the better studied in this regard. Therefore, the effect of oak forests from different French locations [11, 12] showed that wines aged in Limousin barrels had the highest concentration of ellagitannins and total polyphenols. Moreover, it has also been reported that wines aged in Spanish oak barrels showed higher levels of monomeric anthocyanins compared to those aged in French and Central European oak barrels, which may be attributed to the relatively weaker polymerization reactions in wines aged in Spanish oak, which are likely linked to less oxidative conditions [13]. Significant differences were also observed in the content of monomeric phenolics in wines aged in Spanish and French Quercus robur and Quercus petraea barrels [14].

One of the main oak species used for wine ageing is Quercus alba, which grows in various areas across the United States [15], encompasses the whole eastern part of the United States, stretching north into southern Ontario and Quebec, Canada, from the Gulf Coast to the Great Lakes and from the Atlantic Seaboard in the east to the Missouri River in the west [16].

Although Quercus alba is one of the most common and used oak in the eastern United States, there is little information about its region of origin. According to our knowledge, no studies have been published evaluating the influence of Quercus alba barrels from different geographical origins in the United States on the phenolic composition, colour characteristics and sensory analysis of wines. A recent study of our research group has shown that Quercus alba oak barrels from different geographical origins in the USA produced wines with different aromatic characteristics, being the wines aged in barrels from Missouri and Kentucky forests were the ones with the highest scores on global aromatic perception [17].

Therefore, this study aims to evaluate the influence of the USA geographical origin of Quercus alba oak barrels on the phenolic composition, colour properties, and visual and gustatory sensory properties of Tempranillo wines after twelve months of barrel aging. For this, Quercus alba oak barrels from Missouri, Ohio, Kentucky, and Pennsylvania forest were employed. Moreover, wines were grouped based on their initial polyphenolic content, total polyphenol index and colour intensity to assess whether the barrel origin had a different effect depending on the initial polyphenolic composition of the wines.

Materials and methods

Barrels

225-L oak barrels were manufactured in Murua Cooperage (Logroño-La Rioja, Spain) in 2018 as previously described [17]. Quercus alba staves from four different forests of the USA: Missouri (M), Kentucky (K), Ohio (O) and Pennsylvania (P) were used. Raw staves had the following measurements: 950 mm high × 27 mm thick × 50–100 mm wide with natural seasoning at the cooperage for 24 months. Barrels were introduced in water for 40 min to a maximum of 3 h at high temperatures to decrease bitter tannins and favors the toasting process. Finally, the barrels were toasted at medium intensity on oak firewood.

Vinifications and wine samples

Vinifications were carried out in 12 wine cellars from Spanish Qualified Designation of Origin (D.O.Ca) Rioja and Designation of Origin (D.O.) Ribera del Duero using Vitis vinifera cv Tempranillo by traditional red winemaking process [17]. Once malolactic fermentation was completed, wines were aged in triplicate in four different geographical origins of Q. alba oak barrels for a period of 12 months. The storage conditions were maintained at a temperature of 14–16 oC and a relative humidity of 70–75%. As 12 wine cellars were part of the investigation, 144 Q. alba oak barrels were employed. Samples were taken in the initial wines and after 12 months of aging. Wines after 12 months of aging were named M, K, O and P wines.

Determination of oenological parameters and colorimetric characteristics of wines

The classical oenological parameters, total polyphenol index (TPI) and colour intensity (CI) were analysed according to methodologies exposed by the OIV (2003) [18]. Malic acid and the sum of glucose and fructose were analyzed by the autoanalyzer BioSystems Y15 (Biosystem, Barcelona, Spain). The method described by Ayala and co-workers [19] was used to assess the CIELAB parameters. The colour differences between wines were determined as: ΔE* = (ΔL*2 + Δa*2 + Δb*2)0.5 [20]. All analyses were made in triplicate.

Analysis of wine monomeric phenolic compounds

Monomeric phenolic compounds were analyzed by high-performance liquid chromatography (HPLC) (Agilent Technologies, Waldbronn, Germany) with photodiode array detection (DAD) according to the methodology described in Gómez-Alonso and co-workers [21].

A modular 1100 Agilent liquid chromatograph (Agilent Technologies, Waldbronn, Germany) equipped with one G1311A quaternary pump, an on-line G1379A degasser, a G1316A column oven, a G1313A automatic injector, and a G1315B photodiode-array detector (DAD) controlled by the Chemstation Agilent software was used. Separation was done in an ACE HPLC column [5 C18-HL, particle size 5 μm; 250 × 4.6 mm (Teknokroma, Barcelona, Spain)]. The solvents used were: (A) 50 mmol/L (NH4H2PO4 at pH 2.6), (B) acetonitrile/solvent A (80:20% v/v) and (C) 200 mmol/L (o-phosphoric acid at pH 1.5), establishing the following gradient: isocratic 0% B and 0% C during 5 min, from 0 to 8% B in 12 min, from 8 to 14% B and 0% to 86% C in 5 min, from 14 to 18% B and 86% to 82% C in 7 min, from 18 to 21% B and 82% to 79% C in 26 min, from 21 to 33% B and 79% to 67% C in 15 min, from 33 to 50% B and 67% to 50% C in 8 min, from 50 to 80% B and 50% to 0% C in 8 min, at a flow of 1 mL/min. Phenolic compounds were identified according to the retention times of pure compounds and the UV–visible characteristics obtained from chemical standards. Quantification was made using DAD chromatograms recorded at 280 nm (hydroxybenzoic acids and flavan-3-ol), 320 (hydroxycinnamic acids and stilbenes), 360 nm (flavonols), and 520 (anthocyanins). The sum of total hydroxybenzoic acids, hydroxycinnamic acids, flavonols, flavan-3-ol, stilbenes, ellagitannins, and anthocyanins was used to determine the total phenolics. All analyses were done in triplicate.

Analysis of wine ellagitannins

The ellagitannin content, expressed as castalagin, was evaluated by HPLC–DAD after acid hydrolysis according to the method described by Peng and co-workers [22] with little modifications. Briefly, 10 mL of wine were concentrated using a rotavapor until dryness. Next, it was redissolved in methanol, and a fraction was analyzed using HPLC–DAD for the determination of free ellagic acid. Another fraction was hydrolyzed in an acidic HCl/MeOH medium and subsequently analyzed using HPLC–DAD for the determination of total ellagic acid. Separation was done in an Agilent XDB-C18 column (150 × 4.6 cm × 5 µm), with a wavelength of 370 nm.

Wine sensory analysis for the visual and gustatory phase

Sensory evaluation was performed in wines after 12 months of aging in a test room designed according to the ISO 8589:2010 Standard. K, M, O and P wines were tasted separately for each wine cellar. The tasting panel was formed by a group of 12 oenologists from different wineries in the D.O.Ca Rioja (7 males and 5 females, 25–40 years old). A standardized tasting sheet of descriptive analysis method ISO 11035 was used. Before the sensory analysis, the panelist established a common set of attributes for the visual and the gustatory phases by determining both qualitative and quantitative criteria.

For both the visual and gustatory phases, a structured numerical scale with 6 points was utilized. A score of 0 corresponded to the absence of intensity while a score of 5 indicated the highest intensity. The Geometric Mean (GM %) was utilized to classify the visual and gustatory descriptors, in accordance with ISO 11035 from the International Organization for Standardization.

Statistical procedures

Statistical analyses were performed using SPSS Statics 23 (IBM Corp., Armonk, NY, USA). Multivariate analysis of variance (MANOVA) was used to determine differences among barrel origins considering wine mean values as it considers both the factor of wine and the barrel origin (Table 3, 5 and 7). Analysis of variance (ANOVA) was used to determine differences among barrel origins considering individual data for each wine cellar (Table 4 and 6). Post hoc Duncan (p < 0.05) was used to determine the significant differences between samples.

Moreover, wines were grouped based on their initial polyphenolic content, total polyphenol index and colour intensity to assess whether the barrel origin had a different effect depending on the initial polyphenolic composition of the wines. This clustering was carried out using the k-means method for three groups. These results were computed using R version 4.1.0 (R Core Team, Vienna, Austria). Heatmaps were also performed with the R version.

Results and discussion

Oenological parameters and colorimetric characteristics of the initial wines

General oenological parameters and colorimetric characteristics of the wines before aging are shown in Table 1.

Table 1 General oenological parameters and colorimetric characteristics of Tempranillo wines before wine aging
Full size table

The results for volatile acidity after malolactic fermentation showed no microbial alterations. The values of ethanol content, pH, titratable acidity and CIELAB parameters were like those obtained in other Tempranillo wines before barrel aging [23, 24]. The total polyphenol index (TPI) and colour intensity (CI) values of the initial wines were both high, indicating favorable aging potential.

Phenolic composition of the initial wines

Table 2 presents the content of the different families of monomeric phenolic compounds in the wines before aging.

Table 2 Phenolic composition (mg/L) of the wines before aging
Full size table

The main hydroxybenzoic acid detected in wines was gallic acid (Table 2), and it was in the range reported [23, 24]. Ellagic acid was not detected in the wines before aging since this compound originates from hydrolysable tannins, predominantly derived from oak tannins [25]. Caftaric and coutaric acid were the primary hydroxycinnamic acids detected in the wines before aging. Concentration of tartaric esters and free acids were in the range published for Tempranillo wines after malolactic fermentation [26]. Wines showed values in flavonol content in the range obtained by other authors [24, 27]. Quercetin and myricetin-type flavonols were the main flavanols detected, representing together 87–95% of the total flavonol content in the wines. The content of kaempferol, isorhamnetin and their derivatives accounted for 5–13% of total flavonols in wines. Similar percentages were observed by Gómez-Alonso et al. [21] in Tempranillo young wines. The presence of flavonols in grape skins is exclusively in the form of glycosides [28]; and therefore, the occurrence of free flavonols in wines may result from the hydrolysis of their glycosides during the winemaking process. In fact, the average percentage of free flavonols found in wines was 30% to 57% of the total flavonols, which is in the range reported for red wines [29]. Catechin was the only flavan-3-ol detected in wines. Its concentration was slightly higher than that described in Tempranillo wines after malolactic fermentation [26]. The content of stilbenes agreed with the values found described in the bibliography [27]. Trans-piceid (resveratrol glucoside or polydatin) was the main stilbene in the wine samples, representing around 50–86% of the total content of stilbenes. Ellagitannins were not found in the initial wines because they are detected in wines as a result of oak barrel aging or the use of oak chips in stainless steel tanks [30], representing up to 10% of the dry weight of the wood core [31]. The total concentration of anthocyanins in the wines was within the range of values reported for young Tempranillo wines [26]. Therefore, malvidin 3-O-glucoside and its derivatives were identified as the most prevalent anthocyanins in the wines, being around 53–62% of the total anthocyanin content. Non-acylated anthocyanins were 81–87% of the total anthocyanin content. Within the acylated anthocyanins, p-coumaroyl derivatives were predominant, as typically observed in this grape variety [21].

Colour parameters and total polyphenol index of the wines after 12 months of aging

Table 3 shows the mean values of colour parameters and total polyphenol index of the 12 wines aged in K, M, O and P barrels for 12 months. Table 4 shows the individual values of all the wines from the 12 wineries.

Table 3 Mean colour parameters and total polyphenol index of the wines after 12 months of aging
Full size table
Table 4 Colour parameters and total polyphenol index of all the wines after 12 months of aging
Full size table

All the wines showed similar CIELAB and CI parameters after 12 months of aging than those shown in the bibliography for Tempranillo wines aged in oak barrels for 12 months [20, 32].

Barrel origin significantly affected the CIELAB parameters of the wines. K, M, O and P wines showed differences in their CIELAB parameters, both when considering the mean values (Table 3) and the individual data of the wines (Table 4). Therefore, L* values showed significant differences among origins in the 12 wines studied, while b* showed differences in 11, h* in 10, and a* and C* in 8 of the wines.

Wines aged in K and O barrels showed significantly higher mean chroma (C*) values than wines aged in M and P barrels (Table 3), which means higher colour vividness. Therefore, wines aged in K and O barrels showed the highest C* values in most of the wineries which showed significant differences in C* (Table 4). In contrast, wines aged on P barrels had the lowest mean chroma (C*) and hue angle (h*), but the highest lightness (L*) values (Table 3). These findings showed that P wines were more luminous (higher L* values) but less colorfulness (lower C*) wines, and were in agreement with the data observed for the wines in each individual winery. Hence, wines aged in P barrels showed the highest L* value in 11 of the 12 wines studied (Table 4).

The a* CIELAB parameter showed significant differences in the mean values among the wines (Table 3). Hence, K and O wines showed the highest mean contribution of redness (a*) component, but this effect was not observed when observing each individual wine (Table 4). The lowest mean a* value was obtained in P wines (Table 3), which also showed the lowest values in half of the wines with differences in the a* value (Table 4). Higher a* values could be associated with a higher formation of anthocyanin-derived pigments, which could contribute to the stabilization of the flavylium red-colored form in wines aged in K and O barrels [33]. The lowest mean a* values in wines aged in P barrels could be related to a higher degradation of anthocyanins and related pigments, as well as the precipitation of insoluble polymeric anthocyanin-derived pigments in these wines [34]. With regards to the b* component, wines aged in M barrels showed the highest b* values in 7 of the 11 wineries with differences (Table 4), which could be attributed to a greater loss of copigmentation effects which may be accompanied by the formation of red-orangish pigments derived from anthocyanin as pyranoanthocyanins [35], as well as higher oxidation phenomena of red wine pigments [34]. The lowest mean b* value in wines aged in P barrels (Table 3) could be due to the fact that these wines maintained more of the anthocyanin-alkyl-catechin pigments that contribute to the purple colour [36].

Barrel origin also significantly affected the colour intensity (CI) of the wines but not the tonality (T). Hence, CI showed significant differences among origins in 9 of the wines studied, while T values showed differences in only 2 wines (Table 4). K, O and M wines showed the highest mean CI values (Table 3), in good agreement with the data obtained for each individual wine. Therefore, wines aged in M and O barrels showed the highest CI in the majority of wineries studied, followed by K wines (Table 4).

The colour difference values (ΔE*) were calculated to determinate the general colorimetric differences among the K, M, O and P wines. These values were in the 0.22–2.13 CIELAB unit range (data not shown). In general, the human eye can differentiate between two colours when the difference in ΔE* is equal to or greater than 1.0 CIELAB unit [37]. Therefore, the colour differences between K/M, M/O and O/K pairs were not detectable by the human eye (0.34, 0.55, and 0.22, respectively), and only the K/P, M/P, and O/P pairs of barrels could be clearly detected (2.09, 2.02, and 2.13, respectively).

Variations in the chromatic properties of wines observed at the end of the aging process have been associated with variations in the structural and chemical properties of the used wood [14, 32].

Our results indicated that P wines showed lower a* CIELAB values, and colour intensity, but the highest luminosity, which could indicate that these barrels have larger pore size and/or lower content of ellagitannins, and both factors impact the oxidation processes, as well as the polymerization and condensation reactions of the anthocyanins [32]. In previous studies, we have already observed that P barrels had a finer grain than the rest of barrels used [17]. In fact, the OTR (oxygen transfer rate) is affected by grain size as smaller grains allow for greater oxygen influx. Moreover, ellagitannins derived from oak wood have a significant role in wine colour properties. They function as antioxidants, due to their capability to consume large amounts of oxygen, thus regulating the oxidation reactions [32].

Although no significant differences were observed in mean TPI values (Table 3), K, M, O and P wines showed differences in TPI in 10 of the wines analyzed (Table 4), indicating that barrel origin significantly affected the TPI of the wines. Wines aged in K barrels showed the highest TPI in 6 of the wines studied (Table 4), while P wines showed the lowest value in the majority of the wines.

Phenolic composition of the wines after 12 months of aging

Table 5 shows the mean content of phenolic compounds of the 12 wines aged in K, M, O and P barrels for 12 months. Table 6 shows the individual values of all the wines from the 12 wine cellars. Data of individual phenolic compounds are shown in Annex Table 8.

Table 5 Mean phenolic composition (mg/L) of the wines after 12 months of aging
Full size table
Table 6 Composition of phenolic families (mg/L) of all the wines after 12 months of aging
Full size table

Results showed that the influence of the geographic origin of Q. alba on the monomeric phenolic compounds was mainly significant for anthocyanins, stilbenes and ellagitannins. Barrel origin significantly affected the content of anthocyanins in 10 of the 12 wines studied, and ellagitannin and stilbene content was affected in 11 of the wines (Table 6). However, barrel origin only influenced the content of hydroxybenzoic acids and hydroxycinnamic acids in 5 and 3 of the wines studied, respectively, and flavonol and flavan-3-ols content showed differences in only 4 of the 12 wines (Table 6).

In general, P wines showed the lowest concentration of total hydroxycinnamic, hydroxybenzoic acids, flavonols, stilbenes, ellagitannins, and anthocyanins, both in mean and individual data (Tables 5 and 6), and K and O wines showed in total the highest concentration of ellagitannins in 8 of the wines (Table 6). This may be explained by the different types of compounds that are released from wood originating from different geographic regions [7, 12]. Moreover, the content of low molecular weight compounds and their transfer to the wine have been associated with the amount of oxygen that the wines are exposed to during the process of wood aging [38], suggesting our results a different oxygen permeability among barrels.

Regarding the concentrations of anthocyanins, a decreased was observed in most of the wines after 12 months of aging (Tables 2 and 6). The mean anthocyanin content decreased 17 to 19% in K, M and O wines, and 24% in P wines (Tables 2 and 5). The reduction in the concentration of anthocyanins in red wines aged in oak wood barrels has also been reported in the bibliography [39]. Our results indicated that there were more aging reactions leading to a decrease in anthocyanins in P wines than in K, M and O wines. Therefore, P wines showed a quicker evolution than wines from the other barrels because there was a higher loss of anthocyanins. Therefore, P wines showed significantly lower values of total anthocyanins in 9 of the wineries studied (Table 6), which was related to their lowest colour intensity, C* and a* values (see Sect. "Colour parameters and total polyphenol index of the wines after 12 months of aging").

Regarding stilbenes, the mean concentration of trans-resveratrol and its glucoside decreased during oak aging (Tables 2 and 5). The decrease in the content of stilbenes, attributed in the bibliography to their enzymatic conversion to their isomers cis-resveratrol and cis-resveratrol glucoside [40] and to adsorption processes on the oak surface [5], was more accused in P wines (23 to 26% for K, M and O wines vs 32% for P wines) (Tables 2 and 5). After 12 months of aging, wines aged in P barrels had the lowest contents of stilbenes in the majority of the wines, which could indicate a higher adsorption capacity of P barrels.

Wines from K were the richest in ellagitannins in 6 of the 11 wines which showed significant differences (Table 6), probably due to the different content in the heart wood, which undergo alterations based on the different geographic origin [7, 11]. Once ellagitannins are extracted into wine, they undergo a series of reactions, e.g. reacting with anthocyanins or flavanols to form anthocyanin-ellagitannin or flavano-ellagitannins; consuming oxygen, regulating oxidation reactions due to their strong antioxidant properties; and they can also be hydrolyzed [4, 41]. Our results seemed to indicate that the release of ellagitannins from the barrels was higher that its consumption by participating in the reactions resulting in an increase in their concentration. The lowest ellagitannin content in P wines (Tables 5 and 6) may be due to a higher oxygen permeability of the barrel or to the fact that the barrel releases less ellagitannins. The higher oxygen supply in P wines could result in more intense oxidation reactions. Thus, ellagitannins could be more easily oxidized, and as result, the decrease in their concentration would be more pronounced than in others barrels. Our previous results indicated that P barrels probably had a finer grain [17], and thus, they would be more permeable to oxygen.

As observed in bibliography [2, 38], the most abundant benzoic acid was gallic acid and its content was higher in aged wines than in non-aged wines (Tables 2 and 5). Gallic acid may be released by hydrolysis of gallate esters from hydrolysable tannins present in the wood and exhibits strong antioxidant properties even at very low concentrations [42]. Generally, a higher content of gallic acid was found in K, M and O wines (Annex Table 8). Recent studies have shown that the wines aged in low OTR barrels extracted more gallic acid from the wood, and at a faster rate when compared to wines aged in high OTR barrels. This may be attributed to the fact that wood classification by OTR involves a distinction of compositional and anatomical and traits [38]. Ellagic acid increased in the wines during oak aging because of the contact with wood (Tables 2 and 5). The most pronounced increase was observed in O and K wines. The oak wood has a high content of water-extractable ellagitannins, which can be extracted into wine during the oak aging process. However, the mean content of syringic acid decreased with respect to the initial wines (Tables 2 and 5), being the wines aged in O barrels the ones with the lowest concentrations in 6 of the 12 wines (Annex Table 8). A decrease in the concentrations of syringic acid during oak aging has been previously reported [43]. The decrease in the concentrations of syringic acid may be attributed to the equilibration of concentrations between the oak wood and the wine. Additionally, adsorption on the barrel surface, as well as a relatively slow penetration of phenolic compounds into the oak barrels could occur during the wine aging process [43]. The reduction of syringic acid during oak aging could also be attributed to enzymatic or chemical degradation [43].

Hydroxycinnamic acids evolved differently during oak aging. Fertaric acid content increased in all the wines (Tables 2 and Annex Table 8), probably due to aging in oak barrels favors the esterification processes [44]. Caftaric and coutaric acids diminished or maintained their concentration in most of the wines (Tables 2 and Annex Table 8), as they are described to be highly reactive compounds that contribute to oxidation processes [42]. An increase in the mean value of caffeic and p-coumaric acids was also observed in the wines during aging (Tables 2 and 5). The increase in the content of caffeic and p-coumaric acids could be attributed to various factors, including the hydrolysis of hydroxycinnamoyl-tartaric acid grape precursors in wine to produce caffeic and p-coumaric acids. Additionally, the release of these acids from oak wood and their consumption during the formation of pyranoanthocyanins, which result from the reaction between free hydroxycinnamic acids and anthocyanins, may also contribute to this phenomenon [45].

Regarding barrel origin, P wines exhibited the lowest mean concentrations of hydroxycinnamic acids (Table 5), but only 3 wines showed significant differences when considering the individual wines (Table 6), indicating that barrel origin had very little effect on these compounds.

With regards to the flavonol content, quercetin, and myricetin-type flavonols were the most abundant in all the wines. A decrease in the mean values of both flavonol glycosides and aglycones content was observed after 12 months of aging (Tables 2 and 5). As previously reported [1], these results suggest that flavonol glycosides underwent significant acid hydrolysis during the oak aging process. However, an increase in the aglycone content was not observed, probably due to the involvement of flavonols in oxidation and condensation reactions, as well as their insolubility [46]. The mean content of catechin also decreased in the wines during aging (Tables 2 and 5). Flavan-3-ol content in the wines after 12 months of aging was in the range described by other authors [14, 42]. The loss of catechin during wine barrel aging may be attributed to its participation in polymerisation and condensation reactions with anthocyanins, which are favored by the diffusion of oxygen from the barrels and its adsorption on the wood surface [38]. Regarding barrel origin, the higher mean content of flavanols and flavan-3-ols were obtained in K, M and O wines (Table 5). However, only 4 wineries showed significant differences in these compounds when considering the individual wines (Table 6), indicating that barrel origin practically did not affect the content of these compounds.

The results of this research revealed that the origin of barrels made of Quercus alba had a great effect on the content of anthocyanins, stilbenes and ellagitannins but it practically did not affect the rest of the phenolic compounds. The lower content of TPI and phenolic compounds in P wines suggested a slower release of compounds during aging, and/or a higher adsorption capacity and/or oxidation level than barrels from K, M and O origins. By contrast, the higher ellagitannin concentration of K and O wines would provide greater colour stability during wine maturation and aging, protecting the wine against oxidation.

Wine sensory analysis after 12 months of aging

K, M, O and P wines were tasted separately for each winery. ANOVA was applied to the mean data of the 12 wines to analyze differences among wines according to barrel origin. All the sensory attributes evaluated showed GM > 40% (data not shown), so all of them were included in the statistical analysis. The ANOVA results are shown in Fig. 1. Figure 1 shows the mean values of the visual and gustatory phases of sensory analysis of Tempranillo wines aged for 12 months in K, M, O, and P oak barrels. Figure 2 shows the heatmap plot of all the individual wines identifying in red the highest values for each wood origin.

Fig. 1
figure 1

Visual (A) and gustatory (B) phases of sensory analysis of the Tempranillo wines aged for 12 months in oak barrels from Kentucky, Missouri, Ohio and Pennsylvania forests. Mean values of the 12 wines are shown. For each parameter, values with different letters are significantly different between the samples (p ≤ 0.05). Blue line: wines aged in Kentucky barrels; red line: wines aged in Missouri barrels; Green line: wines aged in Ohio barrels; Violet line: wines aged in P barrels

Full size image
Fig. 2
figure 2

Heatmap plots identifying for each barrel origin the highest values of the visual and gustatory attributes of the wines after 12 months of aging. The rows in the heatmap represent sensory attributes and the columns indicate wines. Red colour of the heatmap cells indicates the highest value of parameter across different samples. K: wines aged in Kentucky barrels; M: wines aged in Missouri barrels; O: wines aged in Ohio barrels; P: wines aged in P barrels

Full size image

The Tempranillo wines used in this study were characterized by high colour intensity, limpidity, and brightness (Fig. 1A). Regarding the judge's comments on colour, it is worth noting that the wines were described as exhibiting ruby and cherry red tonalities.

Despite the wines aged in the different barrels showed significant differences in the colorimetric parameters (see Sect. "Colour parameters and total polyphenol index of the wines after 12 months of aging"), no significant differences were observed in any descriptor of visual phase when analyzing the mean values (Fig. 1A). Expect for brightness, the origin of the barrel also showed no clear effect on the visual parameters of colour when considering the individual results of the 12 wines (Fig. 2). This fact may be attributed to the diminished ability of tasters to discern colour variations of up to five units in a to ΔE* when observing the wine through a glass [47]. M wines showed the highest brightness scores in 7 of the 12 wines studied.

Regarding the gustatory phase (Fig. 1B), wines showed high values of smoothness, tannin level, body, length, and balance. Acidity, tannin level, astringent tannin, and balance descriptors showed significant differences among wines aged in barrels from different origins. P wines achieved the mean highest scores in astringent tannins. The higher astringent tannin perceived in P wines increased the perception of acidity with respect to other wines. These results agreed with the sensory data of the individual wines. Therefore, P wines achieved the highest punctuations in astringent tannins and acidity in 6 and 7 of the 12 wines, respectively (Fig. 2).

K and O wines showed the mean highest punctuations of tannin level, and the highest values for balance descriptor (Fig. 1B), in agreement with the individual data where K and O wines showed in total the highest values of tannin level in 8 of the wines, and ripe tannin and balance in 10 and 11 of the wines, respectively (Fig. 2). Wines from P barrels showed the lowest mean scores for the balance descriptor. The higher levels of tannins and ripe tannins in O and K wines made wines more appreciated by judges for their higher balance. The lower content of hydroxybenzoic acids, hydroxycinnamic acids, flavonols, flavan-3-ol, stilbenes, ellagitannins, and anthocyanins in P wines, and their higher values in astringent tannins and acidity, could have led to a lower score for the balance descriptor.

Finally, the tasting panel assessed the overall perception for both the visual and gustatory phases. All wines obtained high punctuations for global perception in the visual phase, 3.64, 3.63, 3.70 and 3.68 for K, M, O, and P, respectively, and no significant differences were observed (data not shown). However, in the global perception of the gustatory phase, the wines differed significantly as the panelist rating for overall acceptance ranged from 3.64 to 2.85. K and M wines had the highest scores (3.64 and 3.26, respectively), followed by the O wines (3.09), and finally, the P wines (2.85) (data not shown). Figure 2 shows that wines aged in K barrels were the bests valued un the gustatory phase in 9 of the wines tasted while M wines were the best valued in 3 of the 12 wines. It is important to highlight that a previous work of our research group showed that the K wine, together with the M wine, were the most highly valued regarding global olfactory perception [17], indicating that these forests would be the most suitable for making barrels providing wines with higher sensory qualities.

Classification of the wines according to their initial phenolic composition and effect of barrel origin

To evaluate the influence of the barrel forest origin according to the initial phenol composition of the wines, the wines were grouped by their initial polyphenol content (measured by HPLC–DAD), TPI and CI parameters. The clustering analysis was performed by the so-called k-means method, for k = 3 groups. The number of groups was previously set by the authors but also confirmed as appropriate by the “elbow” method.

Group 1 was composed of wines with the highest content of polyphenols (870.35 mg/L) as well as the highest value of TPI (90.94) and CI (18.85). Group 2 of wines was characterized by a low content of polyphenols (545.70 mg/L) and the lowest TPI (74.54) and CI (13.11). Group 3 of wines was characterized by low content of polyphenols (538.24 mg/L) and high TPI (82.85) and CI values (18.41).

Table 7 shows the concentration of phenolic compounds in the different groups of wines. MANOVA analysis confirmed the impact of the barrel´s origin, wine, and their interaction. The effect of the origin of the barrel (percentage of variance calculated by MANOVA analysis) increased when the wines were classified according to their initial phenolic (Table 7). As previously stated, the content of ellagitannins was the most affected by the barrel origin. Generally, the barrel origin had a greater influence on the wines with the highest content of phenolic compounds, and CI and TPI values. When considering all the wines, P barrels produced the wines with the lowest content of phenolic compounds (hydroxycinnamic acids, hydroxybenzoic acids, flavanols, flavan-3-ol, stilbenes, ellagitannins and anthocyanins) and CI values (Tables 3 and 5). K and O wines showed the highest content of ellagitannins (Table 5), which would enhance the colour stability of the wines during maturation and aging and protecting the wine against oxidation. This same effect was observed in the wines from group 2 and 3, but not in group 1. Wines forming group 2 were characterized by the lowest initial TPI and CI values. In this group, the effect of the barrel origin on the phenolic composition was the same as it was observed when considering all the wines. Wines from group 3 showed intermediate values of TPI and CI. In this group, P wines showed low contents of hydroxycinnamic acids, hydroxybenzoic acids, and stilbenes, and K and O wines had again the highest content of ellagitannins. In group 1 of wines, characterized by the highest initial polyphenol content, TPI and CI values, no significant differences were observed in the total content of hydroxybenzoic acids, hydroxycinnamic acids, flavanols and anthocyanins among the wines aged in the different barrels. However, wines aged in M barrels had the highest content of ellagitannins and stilbenes, indicating that M barrels could be the most appropriate for aging wines with a higher content of total polyphenols. Ellagitannins possess natural antioxidant properties and can protect other wine compounds, as stilbenes, from oxidation by reacting effectively with oxygen. Regardless of the origin of the barrel, the content of ellagitannins was the highest in wines of group 1. The content of ellagitannins would depend on an equilibrium between their release form the wood and its consumption, which is liked to its oxidation. Nikolantonaki and co-workers [48] observed that the time required for reaching the highest concentration of ellagitannin in the wine is related to its matrix.

Table 7 Composition of phenolic families (mg/L) in the 3 groups of wines after 12 months of aging. Multivariate analysis of variance and percentage of variance attributable (%) to barrel origin, wine and the interaction of both (barrel origin × wine)
Full size table

Conclusions

This work studied the phenolic composition, colour parameters, and visual and gustatory attributes of Tempranillo red wines after 12 months of aging in Quercus alba oak barrels from Missouri (M), Ohio (O), Kentucky (K) and Pennsylvania (P) forests. The results showed that the forest origin of Quercus alba affected the colour parameters of the wines, except for tonality, the total polyphenol index, and the phenolic composition of the aged red wines. Hence, barrel origin significantly affected the content of anthocyanins, stilbenes and ellagitannins but had little effect on the rest of phenolic compounds. Wines aged in P barrels showed the lowest colour intensity and content of most phenolic families and ellagitannins, but the highest luminosity. Wines aged in K and O barrels showed the highest content of ellagitannins. The forest origin of wood did not affect the visual attributes of the wines but produced important effects on their gustatory profile. Wines aged in K and O barrels showed the highest content of tannin level and ripe tannins, and were the most balanced. K wines were the best valued in the overall perception of the gustatory sensory analysis, followed by M wines. P wines were characterized by acidity and astringent tannins and obtained the lowest score in the overall perception of the gustatory phase. Finally, the influence of the barrel origin on the phenolic composition of the wines was evaluated according to their initial phenolic composition. Wines with the highest initial CI and phenol content had the highest content of ellagitannins after barrel aging. The results of the present paper indicated that K barrels were the most appropriate for the aging of Tempranillo wines, as was previously observed by our work group in terms of aromatic quality.