Production of blueberry wine and volatile characterization of young and bottle‐aging beverages

Abstract The aim of this study was the production of blueberry wine and the characterization of the volatile compounds of fermented and aging in bottle products. Multivariate data analysis indicated similarity of volatile compounds released when fermentations were conducted at laboratory‐scale and midscale, with the exception of one replicate creating a distinctive group characterized by low concentrations of acetaldehyde, methanol, 1‐hexanol, and ethyl hexanoate, and the production of polyalcohols such as 2,3‐butanediols. This experiment was the only one where no adjustments of YAN were performed. Some of the major volatile compounds (acetaldehyde, ethyl acetate, 2‐methyl‐1‐butanol, 3‐methyl‐1‐butanol, and 2‐phenylethanol) were found above their perception thresholds. Esters and terpenic compounds were the groups of volatiles expressed the most in blueberry wines, followed by volatile fatty acids, alcohols, and norisoprenoids (3‐hydroxy‐7,8‐dihydro‐β‐ionone, 3‐oxo‐α‐ionol, and 3‐hydroxy‐7,8‐dihydro‐β‐ionol). The wines that experienced bottle‐aging are characterized by high concentrations of ethyl esters, diethyl succinate, ethyl lactate, and diethyl malonate. The results contribute for deeper knowledge of the technological procedure, analytical characteristics, and volatile compounds of blueberry wines, reinforcing the interest in this beverage and opening perspectives for further studies on the production of new blueberry‐based products with differential characteristics that value its nutraceutical and functional properties.

Blueberries are mainly commercialized as fresh fruits, but with the expansion of this highly perishable fruit, farmers have tried to find other strategies to preserve it. In addition to the freezing of blueberries, the transformation of blueberries into juices, jellies, and powders is still an option that guarantees the availability of this fruit in the market, keeping the essential bioactive compounds so desired by consumers (review in Michalska & Łysiak, 2015). In fact, comparative studies demonstrate that the antioxidant activity of anthocyanin extracts from blueberries is not significantly altered in fresh, dried, or frozen blueberries (Lohachoompol, Srzednicki, & Craske, 2004).
The high antioxidant activity of these fruits has been correlated with health-promoting benefits, conferring the title of functional food to blueberries (Huang, Zhang, Liu, & Li, 2012). These properties mainly result from the high content of phenolic compounds, including anthocyanins, chlorogenic acids, flavonols, and procyanidins (Kim, Kim, Kim, & Park, 2013). The major flavonoids in berries and red grapes are anthocyanins and flavonols, mainly present in glycosylated forms (Cho, Howard, Prior, & Clark, 2004). The ranges of total anthocyanin and total flavonol contents (expressed in mg/kg) varied in blueberries from 1435.2 to 8227.3 and from 172.5 to 327.5, respectively, and in red grapes from 380.9 to 7904.7 and from 21.0 to 322.2 (Cho et al., 2004). Similarly, in red grapes, anthocyanins are mainly present in the skins (Riihinen, Jaakola, Ka¨renlampi, & Hohtola, 2008), while tannins and phenolic acids predominate in seeds (Santos et al., 2016). In addition to phenolic compounds, ripe blueberries also contain sugars, but their average content is half of that found in grapes, that is, usually less than 100 g/kg of fresh fruit (Kalt & McDonald, 1996), with glucose and fructose being the greatest percentage of total soluble carbohydrates (Skrovankova, Sumczynski, Mlcek, Jurikova, & Sochor, 2015) without sucrose (Kalt & McDonald, 1996). In a completely different location, sugar content (°Brix) ranged from 8.3 to 14.30 in six different cultivars (Kim et al., 2013), indicating that the amount of sugars is dependent on the cultivar, the degree of ripeness, and location. Thus, total sugar content can be a major criterion for ripeness and consequently for selecting the optimal harvesting point. The organic acids reported in blueberries are citric, malic, succinic, and quinic acids, and the two studied species are distinguishable based on their acid profiles (Ehlenfeldt, Meredith, & Ballington, 1994). Several amino acids other than proline have also been found (Zhang et al., 2014). At normal maturity, an average blueberry fruit consists of approximately 84% water, 0.4% to 0.5% fat, 1.5% to 3.5% dietary fiber, 9.7% carbohydrate, 3.5% cellulose (Keservani, Sharma, & Kesharwani, 2016), 0.7% soluble pectin, several minerals (Zhang et al., 2014), and 0.6% to 0.7% of protein (Michalska & Łysiak, 2015), as mass per mass of fresh fruit. The vitamins in blueberry are vitamin A, niacin, vitamin C (Michalska & Łysiak, 2015), and folic acid (Skrovankova et al., 2015); ascorbic acid shows significant variability between cultivars and species, ranging from 13 to 164 g/kg, with an average of 125 g/ kg (Prior et al., 1998), which corresponds to 1/3 of its daily recommended intake (Keservani et al., 2016). In fact, blueberry is considered one of the greatest sources of antioxidants among all fruits and vegetables (Prior et al., 1998). Continuous expansion of blueberry production incentivizes farmers to search for new technological options for added-value products. Thus, the main goal of this study was to produce an alcoholic beverage from blueberry fruit following the traditional technology used to produce wine and to characterize its volatile composition using gas chromatography-mass spectrometry (GC-MS) for the determination of minor compounds and gas chromatography-flame ionization detection (GC-FID) for major compounds. Few studies have investigated the production of blueberry wines and the volatile composition of this fruit wine remains poorly understood, although aroma is one of the most important qualities for wine products. We anticipate that this study of blueberry wine processing and the evaluation of the aroma compounds of this beverage can contribute to the additional use of this fruit for the production of "fruit wines" or other derivatives.

| Strain and inoculum preparation
Saccharomyces cerevisiae Lalvin QA23 (Lallemand, Montreal, Canada) was used in this study as an active wine dry yeast. Starter cultures were prepared by rehydration of the active dry yeasts according to the manufacturer's instructions to obtain cell populations (as colonyforming units-CFU) of 10 8 ml −1 . This preculture was used to inoculate blueberry juice with an initial cellular concentration of 10 6 ml −1 .

| Preparation of blueberry juice for fermentation
Blueberry wine was produced from Vaccinium corymbosum "Duke," acquired from Cuvelo, Vila Pouca de Aguiar, in the northeastern region of Portugal, from a local farmer who produces approximately 7 t of fruit and exports 60 % of it as fresh berries. In this farm, blueberry fruits are grown in accordance with the principles of organic farming. After harvest, the fruits were kept frozen until the beginning of the experiments. Thus, immediately before the trials, the fruit was thawed, selected, and manually pressed for must preparation. Sulfur dioxide, in a 6% solution, was added up to a concentration of 50 mg/L to prevent unwanted microbial growth. Then, the juice was analyzed to determine the following parameters: ºBrix, pH, titratable acidity (TA), and yeast assimilable nitrogen (YAN). Based on the results, adjustments of sugar, TA, and nitrogen content were performed to optimize yeast performance during alcoholic fermentation. TA, as tartaric acid, was adjusted to 3.5 g/L with L-(+)-tartaric acid (Panreac, Spain). The nitrogen content, as YAN, was adjusted to 120 to 140 mg/L with diammonium hydrogen phosphate (DAP-Merck, Germany). To obtain an alcoholic beverage with an alcoholic strength (by volume) of approximately 11 %, blueberry-juice ºBrix was corrected to 20. The parameters ºBrix, pH, TA, and YAN concentration were determined after the adjustments.
Insoluble solids maintained contact with the fermented juice until 20 to 30 g (1.020 of specific gravity) of sugar remained. At the end of alcoholic fermentation, unclarified juice was centrifuged at

| Laboratory-scale fermentations (LabSFs)
The mixture was subdivided into 1-L flasks filled to 2/3 of their volume and inoculated with the selected wine yeast strain. Two fermentations were conducted at approximately 22 to 25°C without shaking in the presence of 2/3 of the skins (LabSF 1 and LabSF 2).
Fermentations were monitored daily by weight loss as an estimate of CO 2 production.

| Midscale fermentations (MSFs)
Another two fermentations were performed later using the same procedure in 50-L fermenters of stainless steel (MSF 1 and MSF 2).
In sum, fruit berries were pressed in a manual vertical press, and after the addition of SO 2 (50 mg/L), juice composition was adjusted based on the results previously obtained under laboratory conditions. Alcoholic fermentation was conducted in contact with all the skins. Fermentation was monitored daily by determination of temperature and specific gravity; when the specific gravity reached 1.020, the fermented juice was removed from the skins, and the fermented product was kept separately in full 20-L glass bottles until the end of total sugar breakdown.
At the end of alcoholic fermentation, samples were taken from all fermented media to determine the production of minor volatile compounds by GC-MS and major compounds by GC-FID.

| Post-fermentation wine handling, bottling, and aging
At the end of alcoholic fermentation, the temperature of the wines MSF 1 and MSF 2 was decreased to 4°C, and the yeast was allowed to settle for 10-15 days. Sulfur dioxide was added to

| Analytical determinations
Total and free SO 2 , pH, titratable acidity (TA), volatile acidity (VA), and alcoholic strength by volume (AS v ) were determined according to validated standard methods (OIV, 2015). YAN was determined by the formaldehyde method as previously described (Aerny, 1996). All physicochemical analyses were performed prior to and at the end of alcoholic fermentations.

| Organic acids and glycerol
Organic acids and glycerol were quantified by high-performance liquid chromatography (HPLC Flexar, PerkinElmer, Shelton, Connecticut, USA) using a system equipped with an ion exchange column (Aminex HPX-87H, Bio-Rad Laboratories, Hercules, CA, USA) and refractive index and UV/Vis detectors. The eluent was sulfuric acid (5 mmol/L) at a flow rate of 0.6 ml/min. The column was maintained at 60°C, and a sample volume of 6 μl was injected.
The components were identified, using the PerkinElmer Chromera Software, through their retention times relative to respective standards. Citric acid was spectrophotometrically quantified using an enzymatic assay procedure (Y15, © Biosystems S.A., Barcelona Spain).

| Major compounds
Major volatiles were analyzed using a CP-9000 chromatograph (Chrompack) equipped with a Meta-WAX capillary column (30 m × 0.25 mm; 0.20 μm film thickness) and a flame ionization F I G U R E 1 Flowchart showing the experimental design (a) and the major steps of blueberry wine production (b) detector (FID). Helium was used as the carrier gas with an initial flow rate of 1 ml/min. Samples were mixed with 4-nonanol as an internal standard at a final concentration of 60 mg/L, and 1 μl of mixture was injected. The temperatures of the injector and detector were both maintained at 250°C. The column was initially at 50°C, then heated to 177.5°C at a rate of 5°C/min and finally heated to 230°C at 10°C/ min, which was held for 15 min. Quantification was performed using Star Chromatography Workstation version 6.41 (Varian) software with response factors and retention times determined with pure standards. for 2 min, then raised from 60°C to 234°C at a rate of 3°C/min and from 234°C to 260°C at 5°C/min and finally maintained at 260°C for 10 min. The injector temperature was 250°C with 30 ml/min split flow. Compounds were identified using MS Workstation version 6.9 (Varian) software by comparing mass spectra and retention indices with those of pure standards. Minor volatile compounds were quantified as 4-nonanol equivalents.

| Statistical analysis
The data are presented as the mean values with their standard deviation. One-way analysis of variance (ANOVA) and comparison of treatment means (LSD, p < 0.05) of each volatile compound were performed using Statistica 6.1 software (Statsoft, Tulsa, OK, USA).
The Tukey HSD (honest significant difference) test was used for paired comparisons. Differences between the volatile compound profiles of wines were explored using principal component analysis

| General physicochemical characterization of blueberry juice and wine
Freshly squeezed blueberry juice of Vaccinium corymbosum "Duke" had a relatively high pH and consisted of sugars, which were characterized by a °Brix of approximately 13, and a limited concentration of assimilable nitrogen, as displayed in Table 1. The levels of sugars (°Brix) are in accordance with the higher values reported for other cultivars in Korea, which ranged from 8.3 to 14.30 (Kim et al., 2013), but are relatively higher than those reported in California, where the mean values ranged from 77.2 to 103.5 g/kg in ripe or overripe blueberries (Kalt & McDonald, 1996), indicating the dependence of this parameter on the cultivar, degree of ripeness, and location. TA values obtained herein (1.5-4.0 g/L) are in agreement with results previously reported for lowbush blueberry cultivars (Kalt & McDonald, 1996). In this study, YAN ranged from 52.5 to 126 mg/L (Table 1), concentrations similar to those reported in grape berries (from 50 to 450 mg/L). A mean value of 140 mg/L is the value given as sufficient TA B L E 1 Physicochemical characteristics of blueberry juices and respective wines after alcoholic fermentation by the yeast strain Saccharomyces cerevisiae QA23 in laboratory-scale fermentations (LabSFs) and midscale fermentations (MSFs) or after 16 months in bottle

Before corrections
After corrections for complete fermentation of reasonably ripened grapes (Agenbach, 1977). One strategy used by technologists for nitrogen-limited juice is the addition of nitrogen supplements such as inorganic forms like diammonium phosphate prior to grape-juice fermentation (Mendes-  Table 1. Accordingly, the alcoholic strength by volume ranged from 10.2 % to 11.9 %, slightly different than expected (11.7 %), suggesting that substances other than sugars interfered with the measurement of °Brix by refractometry in this type of matrix. Only traces of assimilable nitrogen were found in the final wines, indicating that all the YAN was necessary for the strain S. cerevisiae Lalvin QA23 to complete sugar breakdown. The pH values of the final fruit wines varied from 3.00 to 3.26, with values very similar to those of the corrected juice.
In contrast, TA varied between 5.1 and 6.9 g/L, approximately the double of the total acidity of the corrected juices (3.0 to 4.0 g/L) prior to alcoholic fermentation. This increase in acidity can be attributed to the production of succinic acid by yeasts during alcoholic fermentation (  (Boulton et al., 1996). The concentration of glycerol produced by yeast was independent of the fermentation conditions.
Acetaldehyde constitutes approximately 90 % of the total aldehyde content in wine; it can be formed by yeasts and acetic acid bacteria or as a result of the auto-oxidation of ethanol and oxidation of phenolic compounds . This carbonyl compound also contributes to the persistence of the color in red wine  and, when present in excess, imparts an undesirable green, grassy, apple-like aroma (Osborne, Mira de Orduña, Pilone, & Liu, 2000;Swiegers et al., 2005). Acetaldehyde was the major volatile, and its concentration decreased the most during aging in bottles.
In wine, the decrease in acetaldehyde concentration after alcoholic fermentation can be associated with the occurrence of malolactic fermentation (Osborne et al., 2000), condensation with wine phenolic compounds and reduction to form acetoin and 2,3-butanediol TA B L E 2 Concentration of organic acids (C) in blueberry wines after alcoholic fermentation by the yeast strain Saccharomyces cerevisiae QA23 in laboratory-scale fermentations (LabSFs) and midscale fermentations (MSFs) or after 16 months in bottle Treatment Citric acid C/ (g/L)

Acetic acid C/(g/L) Glycerol C/(g/L)
LabSF Methanol was also found in blueberry wine at concentrations higher than those typically found in wine (approximately

| Minor volatiles
Minor volatile compounds were quantified in blueberry wines by GC-MS, and the results are presented in Table 4. Esters and terpenic compounds were the groups of volatiles expressed the most in blueberry wines, followed by volatile fatty acids, alcohols, and norisoprenoids.
Esters found in fruit wines can derive from either the fruit or sugar fermentation. For instance, ethyl butyrate and ethyl hexanoate have been reported in blueberry fruits (Farneti et al., 2017) but can also be derived via fermentation by yeast, which is responsible for producing several other esters such as isoamyl acetate and ethyl octanoate.
Isoamyl acetate, with a banana-like aroma, ranged from 42.28 to 116.77 μg/L, while ethyl hexanoate and ethyl octanoate, which impart fruity/floral characteristics (Lilly, Lambrechts, & Pretorius, 2000), decreased after aging in bottle, although that decrease was only significant for ethyl octanoate (Table 4). Esters can also derive from chemical esterification of acids when combined with alcohols, which can justify the appearance of ethyl lactate, diethyl malate, and diethyl and monoethyl succinate in the blueberry wines (Bartowsky & Pretorius, 2009). Ethyl esters of succinate, diethyl, and monoethyl, whose descriptors are fruity-apple or stewed fruit aromas, were the major esters in blueberry wines, and their concentrations were also increased after aging. This result is particularly relevant, given that the formation of ethyl esters of lactic and succinic acid has been considered one of the principal chemical changes during aging, with their contents regarded as indices of the length of aging (Shinohara, Shimizu, & Shimazu, 1979) or as enhancers of fruity characteristics in the alcoholic beverages (Swiegers et al., 2005).
Terpenic compounds found in blueberry wines were also diverse. Several terpenic compounds had been previously reported in blueberry fruits, namely linalool, linalool oxide, and α-terpineol, were identified (Beaulieu, Stein-Chisholm, & Boykin, 2014; Du & TA B L E 3 Concentration of major volatile compounds (C) detected and quantified by GC-FID in blueberry wines after alcoholic fermentation by the yeast strain Saccharomyces cerevisiae QA23 in laboratory-scale fermentations (LabSFs) and midscale fermentations (MSFs) or after 16 months in bottle Rouseff, 2014; Farneti et al., 2017). The evolution of terpenes during aging is of particular importance for understanding their content in blueberry wines. The linalool concentration decreased significantly during aging, probably due to chemical phenomena, especially oxidation reactions. During fermentation, certain enzymes of S. cerevisiae have been described to cyclize linalool to α-terpineol (Takoi et al., 2010), which is evident due to the increase in its concentration. As a result of the high linalool content, several other related compounds were found in blueberry wine, which can have varietal origins; these compounds include 8-hydroxylinalool, whose concentration did not TA B L E 4 Concentration of minor volatile compounds (C) detected and quantified by GC-MS in blueberry wines after alcoholic fermentation by the yeast strain Saccharomyces cerevisiae QA23 in laboratory-scale fermentations (LabSFs) and midscale fermentations (MSFs) or after 16 months in bottle for example, the concentration of linalool hydrate increased significantly during storage. Additional terpenic compounds such as transcarveol, β-citronellol, and p-1-menthen-9-ol were found in blueberry wines, but these compounds can also originate from the fruit due to its terpenoid composition. Among these compounds, β-citronellol concentration decreased significantly during aging and is believed to be correlated with the degradation of this compound because it is prone to oxidation by light or hydrogen peroxide (Komaszylo née Siedlecka et al., 2016).
Medium-chain fatty acids (MCFAs) are produced through lipid metabolism by yeasts and are usually associated with unpleasant aromas such as fatty, sweaty, rancid, or cheese aromas (Ferreira, López, & Cacho, 2000). In this study, octanoic acid was the only MCFA present in all wines, but it was present in significantly different concentrations, ranging from 0.58 mg/L (in wines before aging in bottle) to 1.76 mg/L (in wines after aging in bottle).
In grape berries, 4-vinylguaiacol is present as glycosylated precursors; thus, it can be speculated that something similar occurs in blueberry. This compound can be produced from ferulic acid, a natural constituent of berries, by the activity of hydrocinnamate decarboxylase, which converts it to 4-vinylguaiacol (Boulton et al., 1996). In F I G U R E 2 Two-dimensional plot from principal component analysis based on the concentrations of volatile aroma compounds identified and quantified by GC-MS (for minor compounds) and GC-FID (for major compounds) in blueberry wines after alcoholic fermentation by the yeast strain Saccharomyces cerevisiae QA23 in laboratory-scale fermentations (LabSFs) and midscale fermentations (MSFs) or after 16 months in bottle. • LabSF; ■ MSF; X Aged in bottle blueberry wines, 4-vinylguaiacol was detected in concentrations ranging from 24.28 to 80.02 μg/L, and these levels are below the threshold value of 426 μg/L (Chatonnet, Dubourdieu, Boidron, & Pons, 1992); exceeding this level leads to unpleasant mousy, animal, horsy, barnyard, smoky, spicy, or medicinal phenolic character (Heresztyn, 1986;Chatonnet et al., 1992).
To obtain a more abridged view of the relationship among blueberry wines and their aroma composition, PCA was applied to all data obtained (Table 3 and Table 4). This approach allowed the identification of the compounds that better discriminated the wines. The first two principal components accounted for 61.5 % of the total variance. PC1 accounted for 34.5 %, and PC2 accounted for an additional 27 % of the total variability ( Figure 2 that the global aroma of the wines could separate them into two well-defined quadrants (aged and nonaged wines) and a third group separated by differences in the initial nitrogen source.

| CON CLUS IONS
This work reveals that the fruit wines obtained from blueberry present a set of compounds that usually arise from unmodified fruit.
Indeed, several terpenic compounds have been identified, including linalool, cis-furan linalool oxide, α-terpineol, trans-carveol, β-citronellol, and p-1-menthen-9-ol. The presence of nor isoprenoids, namely 3-hydroxy-7,8-dihydroβ-ionone, 3-oxoα-ionol, and 3-hydroxy-7,8-dihydroβ-ionol, appears to be a varietal characteristic of this fruit. A set of other compounds came directly from the metabolic activity of yeast; this set can be called the "fermentation bouquet" as it is in wine, and it is assumed to account for the aroma and mouthfeel perception. This fermentation bouquet consists of alcohols, esters, acids, and carbonyl compounds, which are typical of an alcoholic beverage and mostly contribute positively to wine quality. Interestingly, the levels of some of the alcohols in this fruit wine were above the described perception threshold (2-methyl-1-butanol, 3-methyl-1-butanol, and 2-phenylethanol), and even acetaldehyde surpassed the concentrations usually found in wines. Esters of succinate, generally associated with aging, increased in blueberry wines after 16 months of aging in bottles. Nevertheless, further study is still needed to confirm the sensory contributions of the panoply of the volatile compounds identified in blueberry wines. The results obtained anticipate the potential of this fermented beverage in a market where innovation and new products are increasingly sought by consumers.

FEDER-000004) funded by the European Regional Development
Fund under the scope of Norte2020-Programa Operacional Regional do Norte.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

E TH I C A L S TATEM ENT
This article does not involve any studies with human participants or animal testing performed by any of the authors.