Abstract
The effect of grape maturity on proanthocyanidin concentration, composition and transfer into wine was studied. Vitis vinifera L. cv. Pinot noir grapes (Pommard clone) were monitored for three consecutive vintages (2001 to 2003). Proanthocyanidin content was monitored by reversed-phase HPLC after acid-catalyzed cleavage in the presence of excess phloroglucinol (phloroglucinolysis). After three growing seasons, results indicated that an increase in heat summation between fruit set and veraison was associated with an increase in proanthocyanidin content in grapes and wine. Maturity did not have a consistent effect on the total proanthocyanidin content in wine, but the proportion of seed-derived proanthocyanidins extracted consistently increased with maturity.
Astringency is an important aspect of red wine quality, and proanthocyanidins (PAs) are responsible for this attribute (Gawel 1998). Proanthocyanidins are flavonoid compounds consisting of polymeric flavan-3-ol subunits and are extracted from the skins, seeds, and stems during maceration. Proanthocyanidin quantity and composition in wine can vary with respect to production practice and has been associated with differences in astringency quality.
Research on the diffusion of PAs from the grape berry into wine is limited. According to Ribéreau-Gayon (1982), anthocyanin concentration is maximal before the end of fermentation, while PA concentration increases continuously with maceration. Beyond this, however, our understanding of PA extraction is limited.
Environmental factors in the vineyard such as heat, sunlight and vine water status also affect PA quantity and composition. Consequently, vineyard production practices must be considered when optimizing wine astringency quality (De Freitas et al. 2000, Mateus et al. 2001). The specific objectives of this research were to monitor PA development in Pinot noir grapes over three consecutive growing seasons and to investigate their transfer into wine.
Materials and Methods
Chemicals.
All chromatographic solvents were HPLC grade. Acetonitrile, methanol, ethanol, acetic acid, ascorbic acid, potassium metabisulfite, and potassium hydroxide were purchased from J.T. Baker (Phillipsburg, NJ). (+)-Catechin, (−)-epicatechin, and phloroglucinol were purchased from Sigma (St. Louis, MO). Ammonium phosphate monobasic and orthophosphoric acid were purchased from Fisher Scientific (Santa Clara, CA). Hydrochloric acid and anhydrous sodium acetate were purchased from E.M. Science (Gibbstown, NJ) and Mallinckrodt (Phillipsburg, NJ), respectively.
Instrumentation.
A Hewlett-Packard model 1100 HPLC (Palo Alto, CA), consisting of a vacuum degasser, auto-sampler, quaternary pump, diode array detector, and column heater and operated by a computer workstation and Chemstation software, was used for chromatographic analysis.
Weather information.
Cumulative temperature information was taken from the Oregon State University Hyslop monitoring station (lat. 44°63′N; long. 123°20′W; elevation of 68.4 meters above sea level). This data is available online at www.ocs.orst.edu/pub_ftp/weather/hyslop.
Sampling of grapes.
Fruit (Vitis vinifera L. cv. Pinot noir, self-rooted Pommard clone) was collected from vines grown at the Oregon State University experimental vineyard located in Alpine, Oregon. Six rows containing approximately 50 vines each and two rows containing 20 vines were selected for study. Vineyard operations were consistent with commercially accepted practice. Over three consecutive vintages (2001 to 2003), five cluster samples (x five replicates) were randomly collected for analysis each week beginning approximately four weeks before veraison and continuing through commercial harvest. Berries were kept at 4°C until processed (within one week of collection). To process, and for each replicate, berries were removed from the cluster, randomized, and then separated into two groups, the first group for phenolic analysis and the second for pH, titratable acidity, and sugar analysis.
Extraction of phenolics.
Berries reserved for phenolic analysis were counted, weighed, and kept at −20°C until extraction and analysis. For extraction, seeds and skins were separated by hand from frozen berries and rinsed with distilled deionized water. Seeds were counted, weighed, freeze-dried, and then reweighed; skins were also freeze-dried and weighed. Tissues were then extracted separately in 2:1 v/v acetone:water (1 mL/g berry wt) at room temperature for 24 hr on a platform shaker. After extraction, extracts were filtered through Whatman #1 filters and evaporated under reduced pressure at 38°C to remove acetone. The remaining aqueous solution was adjusted to a volume of 100 mL with water. Extracts were kept at −20°C until analyzed.
Wine production.
For wine production and in 2001, grapes were harvested at two maturities: Sept 27 and Oct 4 with corresponding soluble solids of 23.7 and 27.0 Brix, respectively. In 2002, grapes were harvested at three maturities: Sept 16, Sept 26, and October 7 with corresponding soluble solids of 20.1, 23.4, and 25.5 Brix, respectively. In 2003, grapes were also harvested at three maturities: Sept 15, 22, and 29 with corresponding soluble solids of 21.9, 24.5, and 26.4 Brix, respectively. In 2002 and 2003, 115 kg lots were divided into three replicate fermentations. The 2001 wine production was not replicated.
Wines were made using a standard procedure designed to duplicate commercial extraction. For all wines, attention was directed to making wines in a consistent manner so that differences in wine composition could be attributed to grape maturity. Grapes were divided into three equal lots, destemmed and crushed in a crusher-destemmer (IPI, was added. Musts were Coccaglio, Italy), and 50 mg/L SO2 kept at 5°C overnight, allowed to warm to room temperature (8 hr), and then inoculated with 1 g/L Lalvin RC 212 Bourgorouge yeast (Lallemand, Rexdale, ON, Canada). Once fermentation commenced, fermenting wines were punched down twice a day. Fermentations were monitored with a hydrometer and temperatures were maintained near 31°C. Upon dryness (~7 to 8 days), wines were pressed (up to 2.5 bars) with a bladder-type press (Willmes, Lampertheim, Germany). Approximately one month after pressing, wines were racked and inoculated with 6.6 mg/L Oenococcus oeni OSU-1 strain, malolactic bacteria (Lallemand). After completion of malolactic fermentation (~4 months), wines were racked and the free SO2 was adjusted to 25 mg/L. Wines were kept at 5°C until bottled. Before bottling, SO2 was added to wines in order to achieve 30 mg/L free SO2.
Analysis of flavan-3-ol monomers.
Total flavan-3-ol monomer content was measured by reversed-phase HPLC. Separation of phenolics was achieved with a LiChrospher 100 RP-18 (4 x 250 mm, 5 μm particle size) (Merck, Darmstadt, Germany) column, protected by a guard column containing the same material, and solvent gradient conditions and mobile phases were prepared as previously described (Lamuela-Raventós and Waterhouse 1994). Aqueous extracts and wines were filtered (0.45 μm; Acrodisc CR13, Pall Corporation, East Hills, NY) before injection, and elution was monitored at 280 nm, with (+)-catechin used as a standard.
Analysis of proanthocyanidins.
Proanthocyanidin composition was determined by reversed-phase HPLC after acid-catalysis in the presence of excess phloroglucinol (phloroglucinolysis) (Kennedy and Jones 2001), using a modified HPLC method (Kennedy and Taylor 2003). To prepare samples for analysis, 3 mL seed or skin extract was freeze-dried and then dissolved in 5 or 3 mL methanol, respectively. A phloroglucinolysis reagent was prepared (0.2 N HCl in methanol, containing 100 g/L phloroglucinol and 20 g/L ascorbic acid), and equal volumes of seed or skin solution were reacted in this solution at 50°C for 20 min and then combined with 5 vol of 40 mM aqueous sodium acetate to stop the reaction. Solutions were then filtered (0.45 μm; Acrodisc CR13) before analysis.
For wines, PA analysis was conducted after malolactic addition. To prepare fermentation, final racking, and SO2 samples, 10 mL wine was evaporated in a centrivap concentrator (Labconco, Kansas City, MO), redissolved in 6 mL water, and then applied to a C18-SPE column (1 g; Alltech, Deerfield, IL), previously activated with 10 mL methanol followed by 15 mL water. The applied sample was washed with 15 mL water, eluted with 12 mL methanol, evaporated, and then dissolved in 2 mL methanol. Phloroglucinolysis was then carried out as described above.
The results of phloroglucinolysis provided information on PA subunit composition, mean degree of polymerization (mDP), and conversion yield. Briefly, the mDP was determined by summing terminal and extension subunit amounts (in moles) and dividing by the terminal subunit amount (in moles). For grape PAs, terminal subunits did not include flavan-3-ol monomers. However, the mDP for wine included flavan-3-ol monomers. To calculate conversion yield, the mass of all subunits was summed and then divided by the approximate mass of the polymeric material obtained from the flavan-3-ol monomer analysis. The proportion of seed and skin PA extracted into wine was determined as previously described (Peyrot des Gachons and Kennedy 2003).
Wine extract.
Wine extract was considered to be material that would not evaporate under reduced pressure (centrivap concentrator) and 50°C after 24 hr and was determined by evaporating a 10 mL sample and weighing the residue. The nonvolatile residue was weighed, divided by the initial volume, and expressed as g/100 mL (that is, % extract). The residue was likely composed primarily of carbohydrates, fixed acids, glycerol, phenolics, and inorganic salts.
Statistical analysis.
SAS statistical software version 8.0 (SAS Institute, Inc., Cary, NC) was used to perform analysis of variance across vintages (α = 0.05).
Results and Discussion
Seed development.
Overall, seed development followed expected trends (Figure 1⇓). The number of seeds per berry (Figure 1⇓ inset) for 2003 was significantly higher when compared with 2001 (p = 0.0023) and 2002 (p = 0.0082). No significant difference was found between 2001 and 2002 (p = 0.6468).
A reduced number of seeds per berry has been observed in vines exposed to low temperatures just before and during flowering (Ebadi et al. 1995). Similarly, Ewart and Kliewer (1977) found that the number of seeds per berry in vines grown (from one week before fruit set through veraison) at 25/10°C and 25/20°C (day/night temperature) were approximately twice that of vines held at 15/10°C. In the present study, and for 2003, the increase in heat summation over 2001 and 2002 (Tables 1⇓ and 2⇓) could explain the increase in seed number.
Knowledge of seed number per berry has important implications from a winemaking standpoint because of the importance of seed-derived PAs in red wine astringency. Roby et al. (2004) have shown that grapes with a higher number of seeds per berry showed an increase in PA amount per berry. Furthermore, Roby and colleagues found that, with the exception of very low berry weights, seed PA amount per berry weight increased with berry weight. Therefore, it would be expected that a more astringent wine would result from an increase in seed PA assuming similar berry weight and winemaking practice.
Based upon fresh seed weight, seed development occurred earlier in 2003 than in 2001 and 2002. For all vintages, the highest seed weight occurred approximately 1 to 2 weeks before veraison. In 2003, and in contrast to 2001 and 2002, the fresh seed weight did not change during the harvest period. The 2002 and 2003 dry seed weights increased markedly before veraison, were maximal soon after, and remained constant thereafter. A similar pattern was observed in 2001, but the increase before veraison was not as marked. In all years, dry seed weight was similar after veraison and at harvest (p = 0.2050).
On a per seed basis, and for all years, the flavan-3-ol monomers (Figure 2A⇓) reached their maximum amount near veraison, and then declined. For 2001, flavan-3-ol monomer amount was greater than in other years, and 2002 had the lowest amount. At commercial ripeness, no differences were found between 2002 and 2003; however, 2001 had a significantly higher flavan-3-ol monomer concentration (p < 0.01).
The flavan-3-ol monomer proportion changed after veraison (Figure 3A⇓). Five weeks before veraison, the proportion of (+)-catechin (C) was ~50% (mol%), while (−)-epicatechin (EC) and (−)-epicatechin-3-O-gallate (ECG) were 23% and 27%, respectively. After veraison, these proportions changed. For EC and ECG, these proportions around harvest were 40% and 4%, respectively. Overall, similar flavan-3-ol monomer proportions have been reported in Pinot noir (Santos-Buelga et al. 1994) and in other varieties (De Freitas et al. 2000, Kennedy at al. 2000, Mateus et al. 2001).
Like the flavan-3-ol monomers, the PA extension subunits were maximal near veraison. The amount of total extension subunits in seeds (Figure 2B⇑) was similar for 2001 and 2002, and, on a per seed basis, these two years experienced a decrease in extension subunits immediately after veraison. In 2003 PA amount remained nearly constant for 4 weeks after veraison before declining. At harvest, the extension subunit content was significantly higher (p < 0.0151) for 2003 in comparison with 2001 and 2002.
Seed extension subunit composition remained relatively constant in 2002 and 2003 (Table 3⇓). Near veraison, 77% (by mol) of total extension subunits were EC, while C accounted for 13%, and ECG comprised the remaining 10%.
The seed terminal subunit amount on a per seed basis for all three years generally remained constant after veraison (Figure 2B⇑). At harvest, the terminal subunit content in seeds was significantly higher in 2003 than in 2001 (p = 0.0219) or 2002 (p = 0.0032). No significant difference in terminal subunit content was found from 2001 to 2002.
(+)-Catechin made up nearly 60% of total terminal subunit concentration across the season (Figure 3B⇑). Five weeks before veraison, ECG had the second highest proportion, representing 31% of the total terminal subunits. At harvest, the proportion of ECG decreased to 18% while EC increased to 30%.
In 2001 and 2003, the mDP for seed PAs decreased slightly when comparing veraison with harvest, while the mDP for 2002 remained relatively constant (Table 4⇓). At harvest, 2003 had the lowest mDP in seeds. In contrast, 2002 had the highest mDP. The mDP at harvest was significantly different (p = 0.010) between 2002 and 2003. No significant difference was found between 2001 and either 2002 or 2003.
The conversion yield for seed PAs followed similar patterns in 2001 and 2002. For both years, the conversion yield declined from about 90% (preveraison) to 83% near harvest. The conversion yield for 2003 was greater in comparison (~100%).
The overall seed phenolic development observed in this study is similar to other studies on grape seed (Kennedy et al. 2000, Downey et al. 2003). This includes the pattern of accumulation before veraison and the decline after veraison. However, the conversion yields for cv. Pinot noir were much higher at harvest (83% to 100%) than has been reported for other varieties (Kennedy et al. 2000, Downey et al. 2003).
Skin development.
Fresh skin weight development was similar for 2002 and 2003, although for 2003, the observed increase just before veraison was earlier than for 2002 (Figure 4⇓). At harvest, berries from both years had similar fresh skin weights (p = 0.72). Dry skin weight development was similar for both years, and is similar to previous results (Pirie and Mullins 1977).
Consistent with previous work (Souquet et al. 1996), skin extension subunits accounted for much of the skin PA content (Figure 5⇓). Monomers and terminal subunits in skins were responsible for 0.5 and 3.0% (mol %) of the total skin PA amount, respectively, and this proportion remained nearly constant across maturity and vintage.
On a per berry basis, skin PA for 2001 and 2002 was similar in both pattern of accumulation and amount (Figure 5⇑). At harvest, 2003 skin PA amount was higher than 2001 (p = 0.0214) and 2002 (p = 0.0041) (Table 4⇑). In all three years, skin PA amount at harvest declined to 65% of the maximum observed near veraison. Although the nature of this decline remains unclear, previous studies suggest that the formation of stable associations with cellular components occur (for example, polysaccharides or proteins), which could result in PAs becoming less extractable (Escribano-Bailón et al. 1995, Kennedy et al. 2001).
Consistent with other studies (Escribano-Bailón et al. 1995, Kennedy et al. 2001), C was the only flavan-3-ol monomer observed in skin. (−)-Epigallocatechin (EGC), C, EC, and ECG were present as extension subunits. Extension subunit composition was primarily EC and EGC. From the total extension subunit pool, ~62% (by mol) corresponded to EC and 33% to EGC (Table 3⇑). Skin extension subunit composition changed little during the growing season, consistent with previous work (Kennedy et al. 2001).
The mDP for skin PA across the three vintages varied (Table 4⇑). While 2001 and 2002 increased from veraison to harvest (20.1 to 27.0 and 32.1 to 42.0, respectively), 2003 experienced a decrease (34.9 to 28.9). Both trends have been observed in other studies and varieties (Souquet et al. 1996, Kennedy et al. 2001, Downey et al. 2003). At harvest, 2002 had the highest skin mDP, which was higher (p < 0.001) when compared to 2001 and 2003.
Total fruit flavan-3-ols.
The total PA amount by berry weight (Figure 6⇓) developed in parallel for all vintages, with the highest observed in 2003 and the lowest in 2001. Overall, and by berry weight, there was a decline in total PA amount from several weeks before veraison until two weeks postveraison. After this time, the PA concentration remained constant. The overall drop during this period can be explained by a combination of changes in PA extraction and berry weight increase.
With regard to berry development (Table 1⇑), heat summation (Table 2⇑), and the production of phenolics (Tables 4⇑ and 5⇓), some relationships were found. The year 2003 was warmer and produced more PAs. Crippen and Morrison (1986) found that soluble phenols were significantly higher in sun-exposed berries when expressed on a per berry basis, and this increase may have been due to heat differences. Similarly, Swain and Williams (1970) reported that light greatly enhanced the synthesis of phenolic compounds in plant tissues. Overall cluster environment has been shown to influence grape phenolics (Wicks and Kliewer 1983, Mateus et al. 2001). Recent work separating light and heat indicates that temperature plays a critical role in the accumulation of some berry phenolic classes (Spayd et al. 2002). With the ability to monitor PAs directly, additional controlled studies need to be conducted to understand the role that light and heat play in berry PA accumulation.
Wine flavan-3-ols.
Grape soluble solids at different maturities for 2002 and 2003 along with other parameters are shown in Table 5⇑. In contrast to soluble solid differences, the berry weight was similar at all maturities for both years.
Although there were large differences in soluble solids, differences in wine PA concentration with maturity were more subtle. For 2002, PA concentration was significantly lower in the late maturity wine, and there was no significant difference in PA concentration in 2003. Flavan-3-ol monomer concentration increased with maturity in 2003 but varied in 2002. Given that most of the flavan-3-ol monomer is seed-derived, it is interesting that seed PA differences were not consistent with flavan-3-ol monomer differences.
During the 2003 growing season, wines made from increasingly mature grapes (mid- versus late maturity) resulted in an increase in PA concentration. Perez-Magariño and Gonzalez-San José (2004) reported that dimer and trimer flavan-3-ols tended to reach higher amounts in wines made from increasingly ripe grapes. An opposite pattern was observed in wines made during the 2002 growing season. Although grapes experienced a slight increase in terminal and extension subunit amount for the last two pick dates, the PA concentration in wines decreased with grape maturity, particularly with the late-maturity grape-wine comparison. These results suggest that factors other than fruit PA amount influence wine PA concentration (such as physiological integrity of the fruit).
The mDP for all wines was significantly lower than that observed in seed and skin regardless of maturity, varying from 2.3 to 2.9 (Table 4⇑). The highest value corresponded to 2003. The differences in PA mDP observed between grapes (seeds and skin) and wine are likely to be due to differences in extraction. The acetone-water extraction solvent used for grape PA extraction is a much more effective extraction system than that observed in wine production, and therefore, under winemaking conditions, a subfraction of the grape PA is extracted. The subfraction properties (lower amount and mDP) are consistent with a nonequilibrated diffusion-dependent process.
The wine industry tends to associate riper grapes with improved astringency quality, often known as “tannin ripeness.” Much of the impetus for conducting this study was to explore this observation. A common explanation for the observed improvement in astringency quality as a function of berry ripeness is that there is an increase in skin PA proportion as a result of berry maturity. However, our results indicate that wines made with riper grapes have a higher proportion of seed PA (Table 5⇑).
Some authors have suggested that the improvement in astringency perception in fruits could be due to factors other than PAs. Accepting that the quality of red wine astringency improves with fruit maturity, all indications are that astringency quality cannot depend solely upon PA concentration or composition.
The dramatic increase in wine extract with berry ripening suggests that other factors may also be important in determining astringency quality (Table 5⇑). Recent studies suggest that polysaccharides affect PA astringency perception (Vidal et al. 2004). Given that during fruit ripening the cell wall softens and polysaccharides become more soluble (Silacci and Morrison 1990), riper grapes would be expected to contain a higher concentration of soluble polysaccharides, and the differences in wine extract may partially be explained by differences in polysaccharide content.
Conclusion
Heat summation was associated with an increase in seed number per berry and total proanthocyanidin concentration per berry. Assuming that berry volume is approximately 50% of its harvest volume at veraison and that PA differences at veraison existed at harvest, it may be possible to make early wine proanthocyanidin concentration predictions. Clearly, winemaking influences could easily compensate for the differences in proanthocyanidin concentration observed.
If the quality of astringency perception in wine is dependent upon other factors in addition to PA concentration and composition, and many of these factors also change with berry development, then it is important to better understand the large wine solute changes that accompany berry development. The dramatic increase in extract observed in this study indicates that additional knowledge of compounds other than proanthocyanidin concentration and their impact on astringency perception is required.
Footnotes
Acknowledgments: The Oregon Wine Advisory Board provided funding for this project.
The authors thank Jeff Cygan, Scott Robbins, and the Department of Horticulture at OSU for vineyard management operations and Barney Watson for helpful discussions regarding wine production.
- Received November 2004.
- Revision received April 2005.
- Revision received December 2005.
- Copyright © 2006 by the American Society for Enology and Viticulture