Abstract
In this study, grape cultivar, fruit maturity, and alcohol were varied to determine their impacts on polymeric pigment formation in artificially aged wines. Polymeric pigments are formed primarily, but not exclusively, through the reaction of anthocyanins and tannins during wine aging. Two cultivars (Syrah and Cabernet Sauvignon) that differ in their native anthocyanin and tannin content were harvested at three maturities (20, 24, and 28 Brix) in order to vary the initial anthocyanin and tannin content and their ratio (A:T). At each harvest, juice sugar content was manipulated in the winery to simulate the two other respective maturities. The range of initial wine A:T varied by cultivar (Cabernet Sauvignon, 0.36 to 0.93; and Syrah, 1.3 to 2.1). Results of two-way ANOVAs indicated that fruit maturity and wine age significantly affect anthocyanin content and decline, as well as small polymeric pigment formation. The anthocyanin decline could be modeled by an exponential decay regression, with R2 values ranging from 0.996 to 0.999, depending on maturity and cultivar. Winery alcohol treatment and time significantly impacted the formation of large polymeric pigments. The initial wine anthocyanin content had the strongest correlation to total polymeric pigment content in wines (R2 of 0.735 and 0.670 for Syrah and Cabernet Sauvignon, respectively), while the relationship between A:T and total polymeric pigments was relatively poor (R2 of 0.042 and 0.405 for Syrah and Cabernet Sauvignon, respectively). Measuring wine phenolic hydrophobicity is also proposed to evaluate overall phenolic characteristics, and it was significantly impacted by cultivar, fruit maturity, and aging time.
Red wine quality heavily depends on color and mouthfeel characteristics such as astringency. In Vitis vinifera grapes, color stems from the presence of monomeric anthocyanins (Mazza and Francis 1995), whereas bitterness and astringency stem from monomeric flavan-3-ols and tannin polymers, respectively (Arnold et al. 1980, Gawel 1998). Once extracted into wine, anthocyanins react to form polymeric pigments, which are responsible for stable, long-term color (Somers 1971, Cheynier et al. 2006). While anthocyanins can react with yeast metabolites to form vitisins, the majority of polymeric pigments are due to the reactions of anthocyanins and tannins (Somers 1971, Timberlake and Bridle 1976, Bakker and Timberlake 1997, Salas et al. 2004). In addition to providing stable color, the formation of polymeric pigments is also thought to lead to a decrease in astringency, as isolated anthocyanins and polymeric pigments do not elicit any astringency or bitterness when tasted in model wine solutions (Vidal et al. 2004).
Due to the formation of polymeric pigments, it has been suggested that increasing the anthocyanin concentration of wines increases the solubility of tannin (Singleton and Trousdale 1992). Singleton and Trousdale (1992) also observed that an increase in tannin solubility with the addition of anthocyanin was successful only up to a certain concentration, leading to the hypothesis that an ideal anthocyanin to tannin ratio (A:T) exists. The importance of A:T for polymeric pigment formation is also proposed by multiple authors (Cheynier et al. 1998, Brossaud et al. 1999, Fulcrand et al. 2004, Bindon et al. 2014); however, few studies have examined this effect because of the difficulty of manipulating this ratio while keeping other winemaking and grapegrowing factors constant.
One method of manipulating anthocyanin and tannin levels, and therefore the ratio between the two, involves varying grape maturity and wine alcohol. Anthocyanin synthesis and accumulation in grapes begin at veraison and continue throughout ripening (Boss et al. 1996). Therefore, riper fruit contains more anthocyanins, leading to wines with more anthocyanins (Pérez-Magariño and González-San José 2004, Casassa et al. 2013a).
Although ripening has a more straightforward impact on anthocyanin development, the same is not necessarily true for tannin development. On a per berry basis, skin tannin remains relatively stable throughout ripening, while seed tannin declines slightly (Harbertson et al. 2002). Unfortunately, no relationship exists between the tannin per berry and the amount of tannin in the resulting wines (Adams and Scholz 2008). While tannin extraction cannot be easily predicted from berry ripeness, winemaking techniques can manipulate the concentration of phenolics extracted into the wine (Sacchi et al. 2005). Tannin extraction is complex but generally increases with increased skin and seed contact, such as extended maceration, and can also be increased by increasing fermentation temperature and alcohol (Singleton and Draper 1964, Gawel et al. 2001, Harbertson et al. 2002, Casassa et al. 2013b). Therefore, in this study, alcohol content was selected as a variable in order to manipulate tannin concentration in wines.
Multiple authors have previously examined the effects of alcohol or maturity on the extraction of phenolics (Pérez-Magariño and González-San José 2004, Canals et al. 2005, Casassa et al. 2013a). Generally, more ripe fruit leads to more anthocyanin pigment, and higher alcohol leads to higher tannin concentrations. However, these studies examined these effects over a relatively short term and did not evaluate the impact of initial phenolic extraction on the formation of polymeric pigments over time. Additionally, studies of phenolic compounds during aging and polymeric pigment formation tend to examine isolated systems with just a flavan-3-ol monomer and anthocyanin, or tannin polymer in the absence of anthocyanins (Timberlake and Bridle 1976, Dallas et al. 1996, Francia-Aricha et al. 1997, Fulcrand et al. 1998, Vidal et al. 2002, Salas et al. 2004).
The goal of this project was to determine the most important factors for the formation and stability of polymeric pigments and their phenolic precursors in a complete model wine environment over time. This study also utilized cultivars that inherently yield wines with different tannin concentrations. In this case, we selected Cabernet Sauvignon (CS) and Syrah (SY) on the basis of a combination of availability, economic importance, and known tannin concentration differences (Harbertson et al. 2008). Each cultivar was picked at different maturities to vary anthocyanin concentration and alcohol in order to vary tannin concentration within the same cultivar and vineyard. Additionally, we manipulated the initial soluble solids concentration by chaptalization or removing juice (saignée) prior to fermentation and adding water to have each ethanol concentration represented at each grape maturity. By varying these factors (cultivar, maturity, and ethanol concentration), we also obtained a range of A:T ratios. Manipulating these factors within one cultivar and vineyard allowed for the manipulation of anthocyanin, tannin, and the A:T ratio while keeping all growing conditions constant. Although previous studies have examined maturity and alcohol effects on resulting wines and studied the A:T ratio within isolated systems, this work combines both areas. This study also set out to explore the hydrophobicity of wine samples in order to understand solubility differences of anthocyanins, tannins, and polymeric pigments during accelerated aging.
Materials and Methods
Materials
Amberlite XAD-7, formic acid, hydrochloric acid, potassium bitartrate, octanol, and all HPLC-grade solvents were purchased from VWR. Reagents for the determination of protein precipitable tannin and total phenolics are reported elsewhere (Harbertson et al. 2003, 2015).
Site location and winemaking
The experiment was conducted during the 2015 growing season in the Columbia Valley AVA of Washington State. SY and CS were each manually harvested from the Columbia Crest (45°57′N; 119°36′W) and Cold Creek vineyards (46°57′N; 119°89′W) of Ste. Michelle Wine Estates. For each cultivar, 1.5 tons of fruit was harvested at ~20, 24, and 28 Brix (SY day of year (DOY) 231, 252, and 286; CS DOY 233, 260, and 289). At each harvest date, the fruit was divided into three separate lots. Sugar adjustments were performed on two of the lots in order to replicate the sugar concentration of the other harvest dates (Figure 1). On each harvest date, tanks were adjusted either with saignée (bleed off), followed by chaptalization, or with water-back, in order to achieve 20, 24, or 28 Brix. At each pick date, one lot was kept as a control. Water for the water-back treatments contained 5 g/L tartaric acid, and chaptalization was done with an 80 Brix sugar solution. Saignée prior to water-back or chaptalization maintained the original juice-to-skins ratio and total tank volume.
Winemaking was conducted in triplicate in 200 L stainless steel tanks (58 cm diam with jacketed cooling; Spokane Industries) at the Washington State University Wine Science Center. Fruit was processed the same day it was harvested with Armbruster equipment (vibrating hopper, vibrating sorting table, elevating conveyer, and destemmer/crusher; Scott Laboratories). To each tank, 120 L of must was added, and must, water, and sugar adjustments were performed within 4 hrs of crushing. Sulfur dioxide was added immediately after adjustment at a rate of 50 mg/L. Following these adjustments, all tanks were inoculated with Lalvin EC 1118 (Lallemand) at 106 cells/mL. Fermaid K (0.25 g/L), diammonium phosphate (200 mg/L), and GoFerm (0.3 g/L in yeast rehydration) were added to each tank immediately after inoculation (Scott Laboratories). Wines were coinoculated with malolactic bacteria (Oenococcus oeni) strain Lalvin VP41 (Lallemand) ~48 hrs post-primary inoculation at a rate of 10 mg/L. Tanks were fermented on skins for 10 days and were pumped over six times per day for 5 min each (Cypress Semiconductor). Soluble solids, cap, and must temperatures were monitored multiple times per day, using Cypress fermentation computer systems (Cypress Semiconductor). Consistent with previous experiments, sugar adjustments did not drastically affect the rate of fermentation (Casassa et al. 2013a). Samples (500 mL) were collected at day 10 of fermentation for additional analysis and the accelerated-aging experiment.
The remaining wine was bottled ~6 mos postfermentation and stored at 18°C. Cellar samples were collected 6 and 12 mos postfermentation and were analyzed to determine the validity of the incubator-aging experiment.
Sample preparation and aging
Remaining sugars and organic acids were removed from samples collected after fermentation to eliminate variation between the maturity and sugar treatments, and to ensure that there were no solubility issues during storage. These components were removed with low-pressure chromatography on an Amberlite XAD-7 resin (70 mm × 460 mm). The column was washed with 0.1% (v/v) formic acid for ~15 min. Wine samples were eluted with 80% (v/v) ethanol containing 0.1% formic acid. Wine samples were collected, and the elution solvent was removed by rotary evaporation. The remaining products from the wine samples after evaporation were then dissolved in 500 mL of model wine solution (14% [v/v] ethanol, 5 g/L potassium bitartrate, adjusted to pH 3.5 with HCl) in order to maintain the original wine concentration. Samples were held at −80°C until aging began. Aging was performed similarly to that described by Salas et al. (2003). Samples were divided into 50 mL aliquots for each point of aging (0, 1, 2, 3, and 4 mos), and each tube was sparged with argon and incubated at 30°C (VWR). Individual sample tubes were removed once every 4 wks for 4 mos and centrifuged (5000g for 5 min) prior to analysis to remove any sediment that had formed during incubation. Every wine produced was incubator aged, resulting in 270 samples (including both cultivars and winemaking replicates).
Fruit analysis
Four replicates that each contained eight fruit clusters were collected prior to crushing. For each replicate, 30 berries were randomly selected and blended using an IKA A11 analytical mill (Fisher Scientific). The resulting juice was transferred to a 50 mL centrifuge tube and centrifuged (5000g for 5 min). The supernatant was analyzed for pH, titratable acidity (TA; Mettler Toledo), and Brix (Atago pocket refractometer; Cole Parmer). Berry anthocyanin concentration was analyzed according to Harbertson et al. (2009).
Wine analysis
Ethanol concentration of wines was measured with a near-infrared spectrophotometer (Anton Paar). Analysis of wine polymeric pigment, anthocyanin, tannin, and total phenolic content was performed by protein precipitation, bisulfite bleaching, and ferric chloride reactions as described previously (Harbertson et al. 2003, 2015). Spectrophotometric measurements were performed on an Agilent 8453 UV-vis spectrophotometer (Agilent Technologies).
Octanol-water partitioning
The hydrophobicity of the wine samples was determined by octanol-water partitioning as modified from Mueller-Harvey et al. (2007) and Hagerman et al. (1998). To a 15 mL centrifuge tube, 1 mL wine, 2 mL water, and 3 mL octanol were added. The samples were placed in a 35°C water bath in an N-EVAP nitrogen evaporator (Organomation), and the samples were mixed with a constant stream of nitrogen for 40 min. The tubes were centrifuged (5000g) to separate the layers, and 0.5 mL of each layer was analyzed via high-performance liquid chromatrography (HPLC) on an Agilent 1200 HPLC system. A Zorbax SB-C18 column (3.5μm, 4.6 mm × 150 mm) was used at 30°C for the entire run. Detection was carried out with a diode array detector at 280 nm. The injection volume was 10 μL, and separation utilized a linear gradient with 0.1% (v/v) aqueous trifluoroacetic acid (TFA) (A) and 0.1% TFA (v/v) in acetonitrile (B). Gradient conditions were 0 min, 25% B; 7 min, 75% B; and 8 min, 25% B, followed by a 2 min postrun wash with 25% solvent B. Total peak area was determined by integration with the flat baseline technique. Partition coefficients (Kow) and recoveries were determined with the following equations:
Statistical analysis
A one-way analysis of variance (ANOVA) was used to analyze fruit composition as a result of pick date (df = 2). A two-way ANOVA with interaction and a 5% level for rejection of the null hypothesis were used to analyze the effects of maturity treatment, alcohol level, and their interaction (df = 8). Aged samples were analyzed with a three-way ANOVA for maturity, alcohol level, and time, as well as with a two-way ANOVA at each sample time point. Separation of the means was accomplished with Fisher’s least significant difference (LSD) with a significance value established as p < 0.05. Linear and nonlinear regressions were used to determine correlations between chemical data and polymeric pigment formation. Data analysis was performed using Minitab 17. The dataset was also evaluated with principal component analysis (PCA), using the FactoMineR package in RStudio.
Results
Basic fruit chemistry
In the 2015 season, grape-ripening points were compressed due to unusually warm growing season temperatures; however, ripeness pick dates were separated by ~3 wks. The fruit composition is outlined in Table 1. SY berries dehydrated as the fruit became overripe, but the CS fruit did not. In both SY and CS, anthocyanin (mg/g fresh weight [FW]) increased from unripe to ripe fruit, but decreased slightly in the overripe fruit.
Initial wine chemistry
For chemical analysis, wine-making treatments were grouped according to fruit maturity (unripe, ripe, or overripe) or by sugar adjustment (20, 24, or 28 Brix) (Tables 1, 2, and 3). The sugar adjustment treatment will be referred to as alcohol treatment (low, medium, and high) because of the significant impact of initial sugar concentration on final wine alcohol (Table 2). Higher alcohol concentrations significantly increased wine pH, whereas TA was unaffected.
For wine phenolics, there was no initial two-way interaction between alcohol and harvest date; alcohol concentration consistently affected phenolic extraction for all berry maturities (Table 3). For both cultivars, anthocyanin content depended on fruit maturity and was independent of wine alcohol. Wines made from riper fruit had a higher anthocyanin content. Wine alcohol treatment did not impact anthocyanin extraction, but higher alcohol concentrations did increase tannin extraction. For CS, tannin concentration increased only when alcohol concentration was above 17%, whereas in SY, tannin concentration increased when alcohol concentration was at or above 15.3%. Tannin concentration was largely independent of pick date. However, unripe CS had significantly more tannin than ripe or overripe fruit.
Within a varietal, anthocyanin content varied up to 2.1-fold, and tannin concentration varied up to 1.5-fold due to ripeness or alcohol, respectively. Because of these concentration differences, there was a 2- to 3-fold difference in the A:T ratio within the same cultivar. These wines provided a wide range of anthocyanin, tannin, and A:T for the study of polymeric pigment formation over time.
Accelerated aging
CS and SY followed similar trends relating to the losses of anthocyanin and tannin and the formation of polymeric pigments over time in the wine-aging studies. Over time, anthocyanin content declined in all wines. For both cultivars, the loss of anthocyanin could be modeled by exponential decay (R2 values ranged from 0.996 to 0.999, depending on cultivar and maturity; Figure 2). The decay occurred regardless of tannin concentration or cultivar. This logarithmic decay of anthocyanin was consistent with previous studies (Bakker et al. 1986). Tannin concentration decreased slightly over time in both cultivars (Supplemental Figure 1).
Over time and in both cultivars, anthocyanin and small polymeric pigment (SPP; pigments that cannot precipitate protein and are resistant to SO2 bleaching) content depended only on fruit maturity (Figure 3) and was not affected by wine alcohol (data not shown). Large polymeric pigment content (LPP; pigments that can precipitate protein and are resistant to SO2 bleaching) in CS depended only on wine alcohol content and was independent of fruit maturity (Supplemental Table 1). In SY, alcohol treatment was still significant, but maturity also had an impact on LPP formation (overripe fruit had significantly more LPP than unripe or ripe fruit; Supplemental Table 1). When SPP and LPP were added together to obtain total polymeric pigment (TPP) concentration, both berry maturity and alcohol treatment were significant (Figure 4). In CS, wines made from unripe fruit had significantly less polymeric pigment than ripe or overripe fruit treatments. In SY, all three maturities were significantly different, and polymeric pigment content increased with increasing ripening.
Maximum polymeric pigment formation for CS occurred after 1 mo in the incubator and after 2 mos for SY (Figure 3). When a max concentration was reached, concentrations then leveled off or decreased very slightly. Anthocyanin conversion to polymeric pigment was greater in unripe fruit, even though the resulting polymeric pigment was still less than that in ripe or overripe fruit (Table 3). Additionally, increasing alcohol also increased the anthocyanin retention.
In both SY and CS, initial wine anthocyanin concentration was the strongest predictor of max TPP in the wine (R2 of 0.735 and 0.670, respectively; Figure 5). Conversely, when initial tannin concentration was used as the predictor, there was a slight correlation with TPP in SY, but no relationship in CS (R2 of 0.392 and 0.020, respectively). This is most likely attributed to CS containing relatively high concentrations of tannins. The A:T ratio was an extremely poor predictor of TPP formation (R2 of 0.042 for SY and 0.405 for CS).
PCA identified the overarching relationships among anthocyanin, tannin, and polymeric pigment (Figure 6; Supplemental Figure 2). Treatment confidence ellipses could be established for both maturity and alcohol treatments, with the first component primarily being influenced by maturity, and the second by wine alcohol (Figure 7 and Supplemental Figure 3). The first two components represented 90.3% and 89.3% of the variation for SY and CS, respectively.
Hydrophobicity
The partition coefficient Kow, representing hydrophobicity, significantly differed in the CS and SY samples by cultivar berry maturity and time, but was not affected by wine alcohol (Figure 8). Due to the significance on berry maturity, wines were grouped by this treatment. In CS, partition coefficients for each maturity treatment were significant, whereas in SY, unripe and ripe partition coefficients did not significantly differ during the majority of the aging. In both cultivars, wines from ripe fruit were the most hydrophobic, and hydrophobicity typically increased as wine aged across all ripeness treatments. Average hydrophobicity recovery was 103% ± 5%, and recoveries were slightly over 100% due to increased solubility once phenolic compounds were fractioned into two phases.
Comparison to cellar-aged samples
Cellar-aged wine samples were analyzed 6 and 12 mos after winemaking to determine the validity of the incubator aging. Tannin content after 1 mo of incubator aging was approximately equivalent to 12 mos of cellar aging, but anthocyanin content was slightly higher in the cellar-aged samples (Table 4). Polymeric pigment content was also slightly higher in the incubated samples, indicating that the loss in anthocyanin content in incubated samples may be a result of the accelerated formation of polymeric pigments due to temperature. Additionally, incubation increased the rate of sedimentation of phenolics, which also contributed to the aging difference.
Discussion
The main goal of this study was to determine which factors most influenced the formation of polymeric pigments in wines. To determine the influence of anthocyanin on polymeric pigment formation, it was necessary to represent concentration of both on the same scale. However, polymeric pigments are often represented in terms of absorbance units rather than concentrations due to the wide range of compounds grouped by this classification and the lack of a known molar extinction coefficient. For the sake of simplicity in the present study, absorbance units were converted to malvidin-3-glucoside concentration equivalents, since the malvidin-3-glucoside portion of the polymeric pigment contributes the color. Malvidin-3-glucoside is typically the most abundant anthocyanin in V. vinifera wines (Wulf and Nagel 1978), and most assays for color present data as malvidin-3-glucoside equivalents (Harbertson et al. 2009, Somers and Evans 1977). This conversion allows easier tracking of anthocyanin concentrations over time. Although characterizing polymeric pigments, which are a heterogeneous group of compounds with various polymer lengths and subunit compositions, by one colored compound is an oversimplification, the conversion systemically adjusted the data so that overall trends and statistical differences were unchanged.
Manipulating fruit maturity and wine alcohol levels yielded wines with a wide range of anthocyanin and tannin concentrations and ratios. It has been suggested that the A:T ratio plays an important role in the development of polymeric pigments, since both compounds are required for their formation (Fulcrand et al. 2004, Singleton and Trousdale 1992). However, in the present study, A:T was a very poor predictor of polymeric pigment content. Instead, initial wine anthocyanin content was the best single predictor of polymeric pigment formation over time. This result contradicted the importance placed on A:T in previous literature. However, Brossaud et al. (1999) made claims about A:T after harvesting fruit from various vineyards, therefore varying more factors than just A:T. Additionally, other previous claims about A:T were largely theoretical and were not accompanied by wine-making experiments (Fulcrand et al. 2004).
The TPP can be predicted by initial anthocyanin content because of the strong relationship between SPP and initial anthocyanin (R2 = 0.854 for CS and 0.862 for SY). The correlations between SPP and other factors, such as tannin concentration, were not significant. However, LPP depended only on wine tannin as a result of alcohol extraction, and wine tannin depended on both grape maturity and wine alcohol. Therefore, while maturity was statistically insignificant for LPP formation, the impact of maturity on tannin led to large standard deviations when wines were grouped by alcohol treatment. The correlation (R2) between wine tannin and LPP content was 0.364 and 0.435 for CS and SY, respectively. Finally, wine tannin was correlated with LPP, and wine anthocyanin was correlated with SPP. However, due to the stronger correlation between wine anthocyanin and SPP, TPP was more strongly correlated with initial anthocyanin content.
While initial wine anthocyanin content was a strong predictor of polymeric pigment formation, fruit anthocyanin content did not directly translate to wine anthocyanin content. In both SY and CS, ripe fruit had the most anthocyanin. However, wine made from overripe fruit had equal or greater anthocyanin content than wine made from ripe fruit. Therefore, only wine anthocyanin content, and not fruit anthocyanin, can be used as an indicator for wine polymeric pigment formation.
For both SY and CS, anthocyanin retention improved with increasing alcohol content due to the increased tannin concentration (Table 3). This observation indicated that within a cultivar, increasing initial concentrations of both anthocyanin and tannin will increase polymeric pigment formation (Figure 4), and that the initial concentration of these factors is more important than the ratio between the two. Additionally, when both initial anthocyanin and tannin concentrations were factors in a regression model for max polymeric pigment, the resulting R2 values were higher than when either factor was analyzed individually (0.859 and 0.767 for SY and CS, respectively). This result was consistent with, and expands on, previous literature. Bindon et al. (2014) observed an increase in SO2-resistant pigment over time in wines with higher anthocyanin and tannin concentrations. However, these authors studied only two wines, one high and one low in anthocyanin and tannin. The present study could vary anthocyanin and tannin concentrations over a larger sample size containing more controlled variables.
However, initial anthocyanin and tannin contents were not the only factors that determined the polymeric pigment content. SY and CS had similar levels of initial anthocyanin; however, SY had a higher formation of polymeric pigments than CS. This was counterintuitive, since LPP formation depended on tannins, and CS had higher tannin concentrations than SY. Other factors that affect the formation of polymeric pigments will need further investigation.
Measuring phenolic hydrophobicity provides new insight into phenolic solubility and polymer length. The traditional method to determine the size of tannin polymers by the mean degree of polymerization (mDP) is performed in the presence of acid and a strong nucleophile such as phloroglucinol or toluene-α-thiol (Souquet et al. 1996, Guyot et al. 2001, Kennedy and Jones 2001). However, pigmented tannin is resistant to thiolysis, and yields drop considerably after incubation aging (Salas et al. 2003). Therefore, thiolysis and phloroglucinolysis are not acceptable methodologies for evaluating aged wine samples. Hydrophobicity is a relatively underused method to determine tannin size and sensory characteristics. As tannin polymers increase in polymer length, they become more efficient at precipitating protein (Harbertson et al. 2014) and become more astringent (Arnold et al. 1980, Peleg et al. 1999). Additionally, the larger the tannin polymer, the more hydrophilic it becomes (Hagerman et al. 1998). Finally, individual flavan-3-ol subunits have different hydrophobicity characteristics (Hagerman et al. 1998), and the hydrophobicity theoretically changes with the addition of an anthocyanin to form polymeric pigments. Therefore, the measure of hydrophobicity of a wine provides insight into both polymer length and sensory/astringency characteristics of the wine. In this study, hydrophobicity depended only on fruit maturity and increased with wine age.
As samples aged, concentration of tannin decreased, but polymeric pigment remained stable. Also, as mentioned previously, anthocyanin retention increased with increased tannin concentration. Therefore, the addition of the anthocyanin with a sugar group to the tannin molecule resulted in increased solubility. This finding was consistent with those of Singleton and Trousdale (1992). However, the hydrophobicity of the samples increased over time (Figure 8). The hydrophobicity measurement of the wine sample examines the phenolic system as a whole. Smaller tannin polymers are more hydrophobic than large ones (Hagerman et al. 1998). Therefore, the increase in hydrophobicity over time could potentially be explained by a decrease in tannin polymer size. This would be consistent with the findings of Vidal et al. (2002), who observed a decrease in mDP over time in an isolated system. The authors speculated that the decrease in polymer length was responsible for the decrease in astringency as wines age. However, this system was developed in the absence of anthocyanin.
Over time, the loss of monomeric anthocyanin also affects the hydrophobicity of the system. Because the retention of the anthocyanin is only between 40 and 60%, the loss of hydrophilic monomeric anthocyanins could affect the hydrophobicity more than the change in tannin polymer length. The increasing hydrophobicity over time could also be due to sedimentation and therefore loss of larger tannin polymers, consistent with what has been reported by Zimman and Waterhouse (2004). Over time, the peak area of the octanol phase remained relatively unchanged, while the area of the water phase decreased. This resulted in the shift in increased hydrophobicity. Since the area of the octanol phase did not increase, it is likely that the changing hydrophobicity of the wines was due to the loss of hydrophilic tannin polymers and anthocyanins, rather than to a shortening of tannin polymer length.
Hydrophobicity was slightly different in incubator samples than in cellar-aged samples (Table 4). The higher incubator temperatures increased the rate of sedimentation, and therefore compounds that would fraction into the water phase were lost more rapidly. Since the partition coefficient is a ratio, a slight loss of area in the water phase considerably altered the final value toward greater hydrophobicity. It is also possible that the sample preparation (XAD) removed other phenolic compounds that would contribute to hydrophobicity.
Conclusion
The present study examined the effects of fruit maturity, wine alcohol, and age on the concentration of anthocyanins and tannins, and on the formation of polymeric pigments in wines. In CS and SY, initial wine anthocyanin content was the strongest positive predictor of polymeric pigment content in wines. Although no sensory analysis was performed in this study, it is anticipated that wines with a higher concentration of polymeric pigment, given a constant tannin concentration, will be less astringent than those with lower polymeric pigment content. Therefore, the results of this study suggest that, assuming similar tannin concentrations, wines made from riper fruit would produce less astringent wines than wines made from less ripe fruit.
Phenolic hydrophobicity was also explored in this study and provided a new technique to evaluate the polymer size and chemical characteristics of wine phenolic systems as a whole. In this study, we found that hydrophobicity depended on fruit maturity but not on wine alcohol. Therefore, even though wines resulted in greater tannin content as a result of higher alcohol (and no difference in anthocyanins), the type (subunit composition and polymer length) of tannin extracted depended only on fruit maturity. For all wines made in this study, phenolic hydrophobicity increased over time as hydrophilic tannin and anthocyanin decreased in concentration. Both the loss of tannin and the formation of polymeric pigment would be expected to decrease the perceived astringency over time; however, more research is necessary to confirm this expectation.
Acknowledgments
The Washington State Wine Commission funded this project. The authors thank Ste. Michelle Wine Estates for the donation of the fruit. Thank you also to Christopher Beaver for assistance with statistical analysis, Maria Mireles for laboratory analysis, and Colin Hickey for wine-making assistance.
Footnotes
Supplemental data is freely available with the online version of this article at www.ajevonline.org.
- Received April 2017.
- Revision received August 2017.
- Revision received October 2017.
- Accepted October 2017.
- Published online December 2017
- ©2018 by the American Society for Enology and Viticulture