Sulfur Dioxide Management during Aging Is an Important Factor for the Development of Rosé Wine Color

REPORT Sulfur Dioxide Management during Aging Is an Important Factor for the Development of Rosé Wine Color Caroline P. Merrell 1 and James F. Harbertson 1,2 * Cite this article: Merrell CP and Harbertson JF. 2017. Sulfur dioxide management during aging is an important factor for the development of rosé wine color. Catalyst 1:80-87. 1 Washington State University, 2710 University Drive, Richland, WA 99354-7224; and 2 Associate Professor of Enology, Washington State University, Viticulture and Enology Program, School of Food Science, 2710 Crimson Way, Richland, WA 99354. *Corresponding author (jfharbertson@wsu.edu) Acknowledgments: Ste. Michelle Wine Estates are thanked for fruit donation. Richard Larsen and Colin Hickey are thanked for winemaking assistance, and Maria Mireles is thanked for harvest laboratory analysis. The Wine Research Advisory Committee, the Washington Wine Commission, the Washington Grape and Wine Research Program, and the WSU Agricultural Research Center are thanked for their generous funding of this research. Manuscript submitted Feb 2017, revised Aug 2017, accepted Aug 2017 Copyright 2017 by the American Society for Enology and Viticulture. All rights reserved. doi: 10.5344/catalyst.2017.17003 Summary Goals: The goal of this project was to determine the impact of sulfur dioxide (SO 2 ) and berry maturity on the color of Syrah rosé wines through fermentation and accelerated aging. This project also set out to determine the best method for measuring potential color of rosé wines by measuring anthocyanin bound to SO 2. Key Findings: Acetaldehyde was found to be the most suitable reagent for reacting with SO 2 bound to anthocyanin and releasing potential color of rosé wines. Color in rosé wines was affected by fruit maturity and decreased during fermentation. SO 2 treatment had no impact on rosé color by the end of fermentation. During aging, absorbance at 520 nm increased over time and was affected by fruit maturity and SO 2 treatment. Final absorbance at 520 nm after eight weeks of accelerated aging was accurately predicted by a reaction with acetaldehyde prior to aging. When analyzed with tristimulus color measurements, fruit maturity and SO 2 content affected visual intensity of red and yellow hue over time. Impact and Significance: This study provides winemakers with tools to measure apparent and potential color in rosé wines. In both 22 and 24 Brix fruit, wines retained only 38 and 27%, respectively, of the initial color by the end of fermentation. The level of SO 2 added to wine had no impact on postfermentation color, but levels of SO 2 added prior to aging did impact long-term color. The potential for a rosé wine to darken over time was influenced by how much color the wine initially had, and SO 2 concentrations had a larger impact on color development in wines with more color initially. Key words: color, color stability, polymeric pigment, rosé wine, sulfur dioxide, wine aging Overview The color of rosé wines is a major quality indicator and largely affects consumer preference. Anthocyanins, which are extracted from grape skins during maceration, are responsible for color in winegrapes and, therefore, rosé wines. 1 Once anthocyanins are extracted into solution, they are relatively unstable and react with sulfite and water to form colorless compounds. 2 Over time, anthocyanins react to form more stable color compounds, called polymeric pigments, which are resistant to bleaching by sulfite. In red wine, polymeric pigments are primarily the result of reactions between anthocyanins and tannins. 3,4,5 However, in rosé wine, there is limited extraction of tannins due to limited skin and seed contact. Therefore, polymeric pigments in rosé wine are primarily low molecular weight compounds, which are formed by reactions between anthocyanins and yeast metabolites. 6,7,8 While these compounds are not all polymers, this nomenclature is commonly used to describe this diverse group of pigments. page 80

Sulfur Dioxide and Rosé Wine Color 81 Additionally, co-pigmentation is a phenomenon that occurs in red wines when anthocyanins interact with other molecules, called co-factors, and that results in an overall color that is greater than anthocyanin concentration alone accounts for. 9 However, the concentration of anthocyanins must be greater than 18.5 mg/l, which is not normally reached in rosé wines. 9 There are multiple ways to produce a rosé wine, from simply using a shorter maceration time, to saignée of red juice, to blending red and white wines together. Regardless of the way the rosé wine is produced, anthocyanins are extracted from the grape skins. With anthocyanins present in the wine, polymeric pigments can form over time. 4 The formation of polymeric pigments over time, as well as the loss of SO 2 and the release of bleached monomeric anthocyanins, contributes to the darkening color of rosé wine over time. Studies have examined the effects of various winemaking treatments on rosé wine color, such as maceration time, temperature, and the addition of enzymes. 10,11,12 However, there is limited work on the effect of SO 2 on color during fermentation and aging. Additionally, many of these previous studies have not tracked color during both fermentation and wine aging. Since color is a major quality indicator, this study set out to determine the extent of color change with different levels of initial color (resulting from different fruit maturities) and SO 2 levels. Additionally, research on rosé wine color measurements has focused on developing CIELab or tristimulus values, 13 or high-performance liquid chromatography (HPLC) techniques. 11 However, due to cost, many wineries do not have access to these techniques. This work also sought to determine a simple absorbance method for measuring rosé wine color, which could be easily implemented in wineries. Major Observations and Interpretations Color measurement nomenclature. Throughout this study, we measured color via absorbance at 520 nm because of the simplicity of this measurement and the ability to implement these readings into a practical winery setting. Other methodology to measure anthocyanin content also relies on absorbance at 520 nm, since the pigments absorb at this wavelength. 14 Polymeric pigments were measured by reading absorbance at 520 nm after bleaching monomeric anthocyanin with SO 2. Potential color was measured by reacting acetaldehyde with wine to release any anthocyanin that was bound to and temporarily bleached by SO 2. Finally, CIELab data was collected in order to gather more descriptive information on color changes. Fermentation results. Syrah fruit was harvested at ~22 and 24 Brix (Table 1). Fruit did not have significantly different levels of anthocyanin, but postpress juice did differ significantly (Figure 1). For all the wines, color dropped dramatically throughout fermentation. In both 22 and 24 Brix fruit, low- and high-so 2 treatments had the same color by the end of fermentation. In the wine made from the 22 Brix fruit, the high-so 2 treatment had significantly less color throughout fermentation. However, by day 8 of fermentation, the color was the same as in the low-so 2 treatment. For the wine made from the 24 Brix fruit, the low- and high-so 2 treatments had the same color by day 4 of fermentation. Polymeric pigment (pigment that is resistant to bleaching by SO 2 ) and potential color measurements also dropped throughout fermentation (data not shown). The amount of color retained from press to end of fermentation was only ~38% for 22 Brix fruit and ~27% for the 24 Brix fruit. Development of potential color measurements. After fermentation, the color of rosé wines cannot always be measured directly by absorbance at 520 nm, due to some anthocyanins being bound and bleached by SO 2. Therefore, total potential color can be obtained by adding a strong electrophile such as hydrogen peroxide (H 2 O 2 ) or acetaldehyde to free the bound SO 2. Because of the cost and availability of H 2 O 2 compared with acetaldehyde, the former is a more attractive option. However, H 2 O 2 released bound anthocyanin only briefly, before continuing to oxidize and degrade the anthocyanin. This was evident as a decline in absorbance Table 1 Fruit chemical analysis from each harvest date indicates an increase in fruit sugar over the two pick dates, but no other significant changes. Harvest a Brix b ph TA Avg berry wt (g) Anthocyanin (mg/g fresh wt) DOY 242 22.1 ± 0.2 b 3.68 ± 0.04 6.1 ± 0.3 1.03 ± 0.02 1.09 ± 0.18 DOY 265 24.6 ± 0.4 a 3.72 ± 0.06 6.1 ± 0.3 1.12 ± 0.10 1.14 ± 0.14 a DOY, day of year. b Values in a column not sharing a letter are significantly different at p < 0.05 (Fisher s least significant difference).

82 Merrell and Harbertson after a peak was reached (Figure 2). Moreover, when higher H 2 O 2 quantities were added to wine, this degradation occurred even faster. In contrast, the reaction with acetaldehyde reached a maximum after ~45 min, and absorbance did not decline. The absorbance was also measured 24 hrs after addition and only decreased by ~5%, suggesting that the reaction could be left overnight. Finally, the max absorbance after H 2 O 2 addition was less than what was reached with acetaldehyde. Therefore, H 2 O 2 was not a suitable reagent for determining potential (bound) color, but acetaldehyde could be used. Aging results. In both wines, absorbance at 520 nm (Figure 3) and polymeric pigment content (Figure 4) increased over time. For the wine made from 22 Brix fruit, absorbance at 520 nm increased slightly but then leveled off over time. At week 8 of incubation, all treatments (Brix and SO 2 level) significantly differed in absorbance at 520 nm and in polymeric pigment content. Wine made from 24 Brix fruit had a higher absorbance than wine made from 22 Brix fruit, and the low-so 2 treatments were always higher in absorbance than the high-so 2 treatments. For both 22 and 24 Brix wines, the potential color (A 520nm ) measured prior to incubator aging (week 0) was approximately equal to the final color after eight weeks in the incubator (Figure 5). While there were statistical differences between some of the potential and final color readings, there was not a large practical or visual difference. The statistical significance was due to the small standard error (max of 1.5%). However, the largest variation between predicted and final color was only 9%. Figure 2 Reaction kinetics for releasing bound pigment with acetaldehyde and hydrogen peroxide (H 2 O 2 ) at different concentrations show that H 2 O 2 degrades the anthocyanin pigment. Means are presented, with error bars indicating standard error (n = 3). Figure 1 Fermentation graphs of 22 Brix fruit (A) and 24 Brix fruit (B) indicate that color drops dramatically during fermentation. Means are presented, with error bars indicating standard error (n = 3). Figure 3 Absorbance values over incubator aging indicate a general increase in color for wine made from 24 Brix fruit, but a leveling off for wine made from 22 Brix fruit. Means are presented, with error bars indicating standard error (n = 3). At week 8, data points sharing an asterisk are not significantly different at p < 0.05 (Fisher s least significant difference).

Sulfur Dioxide and Rosé Wine Color 83 For wines made from 24 Brix fruit, time and SO 2 treatment were statistically significant for red and yellow hue of wines. Red hue decreased slightly over time for wine made from 22 Brix fruit, and differences among the SO 2 treatments were negligible (Figure 6). For wine made from the 24 Brix fruit, redness increased over time and reached a steady value. After eight weeks aging, the low-so 2 treatment had a significantly higher intensity of red hue. Yellow hue also increased for both maturities as wine aged, and similar to red hue, the effects were much greater for wines made from 24 Brix fruit (Figure 7). At week 8, the wines made from 22 Brix fruit had similar levels of yellow hue, while in the wine made from 24 Brix fruit, the low-so 2 treatment was more yellow than the high-so 2 treatment. Wine made from 22 Brix fruit was significantly lower in red and yellow hue intensities than wine made from 24 Brix fruit. As discussed above, polymeric pigment formation, which most likely corresponded to this yellowing, also increased over time, and treatment trends were similar Figure 4 Formation of polymeric pigments (pigments resistant to bleaching by SO 2 ) increased over time for all treatments. Means are presented, with error bars indicating standard error (n = 3). At week 8, data points sharing an asterisk are not significantly different at p < 0.05 (Fisher s least significant difference). Figure 6 Tristimulus color measurements for red hue show an increase in red hue over time for wine made from 24 Brix fruit, but not from 22 Brix fruit. Means are presented, with error bars indicating standard error (n = 3). At week 8, data points sharing an asterisk are not significantly different at p < 0.05 (Fisher s least significant difference). Figure 5 For each maturity and SO 2 treatment, potential color (measured immediately before aging) corresponds to final color after eight weeks of incubator aging. Means are presented, with error bars indicating standard error (n = 3). Bars sharing an asterisk are not significantly different at p < 0.05 (Fisher s least significant difference). Figure 7 Tristimulus color measurements for yellow hue show an increase in yellow hue over time. Means are presented, with error bars indicating standard error (n = 3). At week 8, data points sharing an asterisk are not significantly different at p < 0.05 (Fisher s least significant difference).

84 Merrell and Harbertson for polymeric pigment formation (Figure 4) and yellow hue increase (Figure 7). The lightness attribute slightly decreased over time, but no major changes were observed (data not shown). This indicated that the changes in red and yellow hues corresponded to the perceived darkening over time rather than to an actual decrease in lightness. Figure 8 illustrates the color changes over time from CIELab data. Color changes in the wine made from 22 Brix fruit were minimal; however, the wine made from the 24 Brix fruit became more intensely colored over time. Broader Impact Pinking is a phenomenon characterized by the development of a pink color in wines made from white grape cultivars. This phenomenon occurs in wine made from white grapes, such as Pinot gris or Gewürztraminer, that have a small amount of anthocyanin pigment in their skins. In white wine, it has been observed that pinking occurs over time when SO 2 levels decrease, which releases bound anthocyanin that was present in the wine. 15 Over time, this free anthocyanin reacts to form polymeric pigments that are then resistant to bleaching by SO 2. This evolution of white wine pinking is extremely similar to the chemical evolution of rosé wine color, which leads to the darkening of color over time. To measure potential pinking in white wines, the use of H 2 O 2 has been suggested. 15,16 However, H 2 O 2 cannot adequately predict the total color of wines due to the oxidative degradation of anthocyanins. For both potential pinking due to the presence of anthocyanins, as well as potential color of rosé wines, acetaldehyde is potentially a better option because of the persistence of the color. Since the majority of color (including bound anthocyanin) was lost during fermentation, measuring potential color during active fermentation was not beneficial. A decline in anthocyanin concentration in the absence of grape skins has been observed previously and can be Figure 8 CIELab color representations of rosé wines after incubator aging.

Sulfur Dioxide and Rosé Wine Color 85 partially attributed to interactions with yeast. 17 In addition, fermentation in the absence of skins and seeds leads to wines without significant extraction of tannins. Reactions between anthocyanin and tannin have been observed to increase stability of the pigments. 18 Therefore, this loss of anthocyanin during fermentation is due to both adsorption on lees and the instability of the pigment. Although not beneficial during fermentation, potential color measured immediately prior to aging (after SO 2 addition) was a good predictor of final wine color (Figure 5). While tristimulus color measurements continue to be the most descriptive tool for measuring rosé wine color and change, absorbance at 520 nm can also be used to track and estimate final color. Absorbance values are also more realistic for wineries to measure than CIELab or HPLC, making this methodology practical for predicting and measuring color changes. The use of SO 2 is a valuable tool to control color in rosé wines. Levels of SO 2 prior to fermentation did not affect postfermentation color. However, the level of SO 2 added before aging had a significant impact on the absorbance and hue of the wine. Moreover, polymeric pigment content depended on both fruit maturity and SO 2 treatment. Wines higher in SO 2 had a lower formation of polymeric pigment. With increasing SO 2 levels, there is an increased number of anthocyanins weakly bound to SO 2. Our results suggest that anthocyanins that are bound to SO 2 do not appear to react with other compounds to form polymeric pigments. Additionally, wines that began with more color postfermentation had more polymeric pigment formation due to an increased anthocyanin amount. In wines made from 24 Brix fruit, the increase in red hue for the low-so 2 treatment corresponded to more monomeric (and not sulfite-bleached) anthocyanin, and the increase in yellow hue corresponded to higher polymeric pigment content. Due to this impact on polymeric pigment formation, hue, and absorbance at 520 nm, SO 2 management before bottling is the most critical time point for controlling long-term rosé wine color. The wine made from 24 Brix fruit had slightly different aging patterns than the wine made from 22 Brix fruit. This indicated that when wines have more color initially, they have more potential to change color in the bottle. Wines made from 24 Brix fruit had significantly higher absorbance after fermentation and a higher potential color at week 0 of incubator aging, indicating that these wines had more color than wines made from 22 Brix fruit. The wine made from the 22 Brix fruit did not drastically increase in color in the bottle, and SO 2 levels had less of an impact on color changes. However, the wine made from the 24 Brix fruit had large color changes in the bottle, and the SO 2 treatment had a major impact on color. Special attention should be paid to SO 2 decisions to limit color darkening in the bottle when fruit is riper or has more initial color. Incubator aging has been utilized in previous studies to accelerate the effects of aging. 3,19 In red wine, it has been observed that approximately one month in the incubator corresponds to one year in the cellar. 19 Although red and rosé wines age slightly differently because of differences in phenolic content, the eight-week time point roughly correlated to two years in the cellar. In addition, plastic tubes used for aging in this study allow slightly more oxygen into the samples than glass wine bottles. Tubes were flushed with argon prior to aging to minimize these effects. However, this increase in oxygen content may oxidize anthocyanins, lowering absorbance at 520 nm in incubated samples, compared with cellar-aged samples. From a practical winemaking point of view, winemakers can set final color absorbance goals and adjust winemaking decisions accordingly. In this study, picking decisions were varied to manipulate the starting wine color. However, the pressure reached during the press cycle or the time on skins can also change extraction levels. Since wines are expected to retain only between 27 and 38% of postpress color, final absorbance values can be estimated from initial, postpress color. Riper fruit may be desirable for rosé wine production because of the increase in ripe aromas and flavors. However, wine made from less ripe fruit, which finished fermentation at an absorbance value below 0.1, did not drastically change color during aging, whereas wine with more initial color, which finished fermentation at an absorbance of 0.2, did darken substantially in the bottle. If riper fruit is desired, the potential color reading could be used before bottling to predict final color on the basis of anthocyanin release from SO 2 over time, and the SO 2 could be adjusted accordingly. While lowering initial wine color through fruit maturity was the most effective way to limit darkening over time, the use of SO 2 can also help a winemaker s control over color increase during aging. Experimental Design Winemaking. The experiment was conducted during the 2016 growing season in the Columbia Valley AVA of Washington State. Syrah was manually harvested at 22 and 24 Brix (day of year 242 and 265, respectively) in order to obtain fruits with different levels of initial color. The fruit went directly to press (bladder press;

86 Merrell and Harbertson Zambelli Enotech) with a final pressure of 2 bar. Winemaking was conducted in triplicate on a 20 L scale at the Washington State University Wine Science Center. Potassium metabisulfite was added directly after juice was pressed to reach a final SO 2 concentration of 25 and 50 mg/l (low and high treatments, respectively). KS enzyme was added prior to inoculation at 0.026 ml/l juice (Scott Laboratories). The juice was inoculated with Lalvin EC 1118 at 10 6 cells/ml (Lallemand). Fermaid K (0.25 g/l) and GoFerm (0.3 g/l in yeast rehydration) were added to each carboy immediately after inoculation (Scott Laboratories). Fermentations were sampled every other day and were monitored with a portable density meter (Anton Paar). When the fermentation had finished, wines were cold-stabilized at 0 C. After cold stabilization, samples were collected and potassium metabisulfite was added to reach 0.4 and 0.8 mg/l free molecular SO 2 (low-and high- SO 2 treatments, respectively). SO 2 levels were measured enzymatically (Admeo), and further additions were made until the free SO 2 concentration was stable. Four 50 ml samples from each fermentation replicate were then placed in centrifuge tubes, topped with argon gas, and placed in a 30 C incubator (VWR) in order to simulate aging. One set of sample tubes was analyzed every two weeks for eight weeks, giving triplicate aging samples for each maturity and SO 2 treatment. Color measurements. Acetaldehyde has long been used in excess to react with SO 2 and free and bound anthocyanin pigments. 20,21 H 2 O 2 has chemical properties similar to acetaldehyde 15,16 and is an inexpensive option. To test the suitability of both acetaldehyde and H 2 O 2 for this reaction, different concentrations were added to wine samples, and absorbance readings were taken every five minutes. During fermentation and aging, absorbance was measured at 520 nm (Agilent). Potassium metabisulfite was added to wine samples (80 μl in 1 ml wine, 10 min incubation) to measure total polymeric pigment. 22 Acetaldehyde was added to samples (10 μl of 10% acetaldehyde solution in 1 ml wine, 45 min incubation) to measure total potential color. 20,21 During incubator aging, samples were also measured for tristimulus color values using a CR-400 Chroma Meter (Konica Minolta). Fermentation samples were not measured with tristimulus color due to limited sample volume. Statistical analysis. Statistical analysis was performed with one- and two-way analysis of variance using Minitab 17. Separation of the means was accomplished using Fisher s least significant difference with a significance value established as p < 0.05. References and Footnotes 1. Mazza G and Francis FJ. 1995. Anthocyanins in grapes and grape products. Crit Rev Food Sci Nutr 35:341-371. 2. Fulcrand H, Dueñas M, Salas E and Cheynier V. 2006. Phenolic reactions during winemaking and aging. Am J Enol Vitic 57:289-297. 3. Salas E, Fulcrand H, Meudec E and Cheynier V. 2003. Reactions of anthocyanins and tannins in model solutions. J Agric Food Chem 51:7951-7961. 4. Somers TC. 1971. The polymeric nature of wine pigments. Phytochemistry 10:2175-2186. 5. Remy S, Fulcrand H, Labarbe B, Cheynier V and Moutounet M. 2000. First confirmation in red wine of products resulting from direct anthocyanin-tannin reactions. J Sci Food Agric 80:745-751. 6. Fulcrand H, Cameira dos Santos P, Sarni-Manchado P, Chenyier V and Favre-Bonvin J. 1996. Structure of new anthocyanin-derived wine pigments. J Chem Soc, Perkin Trans 1 1:735-739. 7. Fulcrand H, Benabdeljalil C, Rigaud J, Cheynier V and Moutounet M. 1998. A new class of wine pigments generated by reaction between pyruvic acid and grape anthocyanins. Phytochemistry 47:1401-1407. 8. Bakker J and Timberlake CF. 1997. Isolation, identification and characterization of new color-stable anthocyanins occurring in some red wines. J Agric Food Chem 45:35-43. 9. Boulton R. 2001. The copigmentation of anthocyanins and its role in the color of red wine: A critical review. Am J Enol Vitic 52:67-87. 10. Puértolas E, Saldaña G, A lvarez I and Raso J. 2011. Experimental design approach for the evaluation of anthocyanin content of rosé wines obtained by pulsed electric fields. Influence of temperature and time of maceration. Food Chem 126:1482-1487. 11. Kelebek H, Canbas A and Selli S. 2007. HPLC-DAD-MS Analysis of anthocyanins in rosé wine made from cv. Öküzgözü grapes and effect of maceration time on anthocyanin content. Chromatographia 66:207-212. 12. Salinas MR, Garijo J, Pardo F, Zalacain A and Alonso GL. 2003. Color, polyphenol and aroma compounds in rosé wines after prefermentative maceration and enzymatic treatments. Am J Enol Vitic 54:195-202. 13. Ayala F, Echávarri JF and Negueruela AI. 1997. A new simplified method for measuring the color of wines. I. Red and rosé wines. Am J Enol Vitic 48:357-363. 14. Harbertson JF and Spayd S. 2006. Measuring phenolics in the winery. Am J Enol Vitic 57:280-288. 15. Andrea-Silva J, Cosme F, Ribeiro LF, Moreira ASP, Malheiro AC, Coimbra MA, Domingues MRM and Nunes FM. 2014. Origin of the pinking phenomenon of white wines. J Agric Food Chem 62:5651-5659. 16. Simpson RF. 1977. Oxidative pinking in white wines. Vitis 16:286-294.

Sulfur Dioxide and Rosé Wine Color 87 17. Medina K, Boido E, Dellacassa E and Carrau F. 2005. Yeast interactions with anthocyanins during red wine fermentation. Am J Enol Vitic 56:104-109. 18. Singleton VL and Trousdale EK. 1992. Anthocyanin-tannin interactions explaining differences in polymeric phenols between white and red wines. Am J Enol Vitic 43:63-70. 19. Merrell CP. 2017. Phenolic evolution in wine: Determination of new methodology to measure phenolic content and the impact of maturity, alcohol and sulfur dioxide on wine color during aging. Thesis, Washington State University, Richland, WA. 20. Somers TC and Evans ME. 1977. Spectral evaluation of young red wines: Anthocyanin equilibria, total phenolics, free and molecular SO 2, chemical age. J Sci Food Agric 28:279-287. 21. Levengood J and Boulton R. 2004. The variation in the color due to copigmentation in young Cabernet Sauvignon wines. In Red Wine Color. Waterhouse AL and Kennedy JA (eds.), pp. 35-52. American Chemical Society, Washington, DC. 22. Harbertson JF, Picciotto EA and Adams DO. 2003. Measurement of polymeric pigments in grape berry extracts and wine using a protein precipitation assay combined with bisulfite bleaching. Am J Enol Vitic 54:301-306.