EFFECTS OF ETHANOL, TANNIN AND FRUCTOSE ON THE SENSORY AND CHEMICAL PROPERTIES OF WASHINGTON STATE MERLOT ANNE CAROLYN SECOR

Transcription

1 EFFECTS OF ETHANOL, TANNIN AND FRUCTOSE ON THE SENSORY AND CHEMICAL PROPERTIES OF WASHINGTON STATE MERLOT By ANNE CAROLYN SECOR A thesis submitted in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE IN FOOD SCIENCE WASHINGTON STATE UNIVERSITY School of Food Science MAY 2012

2 To the Faculty of Washington State University: The members of the Committee appointed to examine the thesis of ANNE CAROLYN SECOR find it satisfactory and recommend that it be accepted. Carolyn F. Ross, Ph.D., Chair Charles G. Edwards, Ph.D. Jeffri C. Bohlscheid, Ph.D. ii

3 ACKNOWLEDGMENTS I would like to thank Dr. Carolyn Ross for her support throughout this project. Her advice and encouragement have been invaluable throughout this process. The other members of my graduate committee, Dr. Charles Edwards and Dr. Jeff Bohlscheid, have also been immensely helpful. Thank you to Snoqualmie Winery, for their collaboration in dealcoholizing the wine for this project. Thank you to Karen Weller, Scott Mattinson, and Jodi Anderson, for keeping me on track in all aspects of graduate life. To Medy Villamor, I am grateful for the support, the collaboration, and the endless positivity. Also, thank you to the rest of the Ross lab, for your help with sensory panels and flexibility in the lab, and to the Edwards lab, for the honorary work-space and support. Finally, this project would not have been possible without the support of my parents, William and Tammi Secor; my fiancé, Brandon Zwink; and my dearest friends. Thank you. iii

4 EFFECTS OF ETHANOL, TANNIN AND FRUCTOSE ON THE SENSORY AND CHEMICAL PROPERTIES OF WASHINGTON STATE MERLOT ABSTRACT By Anne Carolyn Secor, M.S. Washington State University May 2012 Chair: Carolyn F. Ross The relationship between matrix components and sensory properties of red wine was examined. A Washington State Merlot was dealcoholized to 3.2% and alcohol was added back to four ethanol levels: 3.2%, 8%, 12% and 16% ethanol (v/v). Within each treatment, wines were maintained at the original tannin (211 mg/l CE tannin) and fructose (120 mg/l fructose), or brought to 1500 mg/l CE tannin and/or 2000 mg/l fructose (n=16 solutions). The wines were spiked with the same concentrations of three aroma compounds: 3-methyl-1-butanol, 2- phenylethanol, and eugenol. These wines were then evaluated by a trained panel (n=10) for the intensity of aromas and flavors ( caramel, rose and clove ), tastes ( bitterness and sourness ), and mouthfeel ( astringency and heat ). Gas chromatography/mass spectrometry was used to quantify aroma compounds. PCA was used for correlation between sensory and analytical results. All data were analyzed using analysis of variance (p<0.05) and Fisher s Least Significant Difference. Analytical results showed that ethanol significantly reduced the relative headspace recovery of all three compounds. The interaction effects between ethanol, tannin and fructose varied based upon the aroma compound and the ethanol content. In standard red wine ethanol concentrations (12 to 16%), volatile recovery was not influenced by tannin or fructose. iv

5 However, in low ethanol wines, high tannin concentration negatively impacted the relative recovery of 3-methyl-1-butanol, 2-phenylethanol, and eugenol. An increase in fructose concentration when ethanol and tannin concentrations were low reduced the recovery of 3- methyl-1-butanol, but increased the recovery of 2-phenylethanol. The trained panel sensory evaluation results showed that increasing ethanol concentrations increased clove flavor, and heat, and decreased sourness intensity. High fructose concentration increased rose aroma and flavor scores, and decreased clove aroma scores. Tannin concentration positively affected clove flavor while perceived drying and bitterness were impacted by ethanol*tannin. PCA separated treatments based on ethanol, tannin, and fructose concentrations, and chemical analyses of aroma compounds were not correlated with perceived aromas or flavors. This study demonstrated the complexity of relationships within the wine matrix, indicating chemical and sensory effects that winemaking techniques such as saigneé, the addition of water, and dealcoholization may have on wine quality. v

6 TABLE OF CONTENTS Page ACKNOWLEDGMENTS... iii ABSTRACT... iv LIST OF TABLES... viii LIST OF FIGURES... x CHAPTER I: INTRODUCTION... 1 CHAPTER II: LITERATURE REVIEW... 4 Importance of Wine to Washington State... 4 Current Trends in Increasing Alcohol Content in Wines... 4 Methods of Alcohol Reduction... 6 Saigneé and water addition... 6 Dealcoholization by reverse osmosis... 7 Wine Sensory Attributes... 8 Alcohol burn... 8 Astringency... 9 Sourness Bitterness Aromas Flavors Physiological factors Wine Matrix: Volatile and Non-Volatile Components Tannin Fructose vi

7 Ethanol Aromatic volatile compounds Interactions between pairs of components Interactions among three or more components CHAPTER III: MATERIALS AND METHODS Materials Base Wine Volatile Compound Profiling Calibration Curves Wine Treatments Chemical and Volatile Analysis Sensory Analysis Data Analysis CHAPTER IV: RESULTS AND DISCUSSION Chemical Analysis Volatile Compound Analysis Sensory Evaluation Principal Component Analysis and Pearson Correlation CHAPTER V: CONCLUSIONS AND SUGGESTIONS FOR FUTURE RESEARCH LITERATURE CITED vii

8 LIST OF TABLES Table 1. Treatment number and associated ethanol, tannin, and fructose concentration. A total of 16 treatments were evaluated by both GC/MS and sensory methods. Volatile compound concentrations remained constant for each treatment: 93.8 mg/l 3-methyl-1-butanol, 78.4 mg/l 2-phenylethanol, and 0.5 mg/l eugenol Table 2. Taste and aroma standards used in training session. Base wine was Livingston Red Rosé (Modesto, CA) Table 3. Analytical results of Merlot wine, after dealcoholization and prior to treatment modifications, including ph, titratable acidity (g/100ml), ethanol (%), tannin (mg/l CE), residual sugar (%), fructose (mg/l), free SO 2 (mg/l), and total SO 2 (mg/l). Results presented are the mean of triplicate measurements, followed by the standard deviation Table 4. Analytical results of Merlot wines used for sensory evaluation, including ethanol (%), tannin (mg/l CE), fructose (mg/l), ph, and titratable acidity (g/l). Treatment numbers refer to treatments described in Table 1. Values represent a mean of triplicate measurement, followed by the associated standard deviation. Means with different letters within columns differ at p < 0.05 using Tukey s HSD Table 5. Standard curves created for quantification of 3-methyl-1-butanol, 2-phenylethanol, and eugenol in 3.2% ethanol. Measurements were taken as a mean of three measurements, with six points per standard curve for 3-methyl-1-butanol and 2-phenylethanol, and five points in the eugenol standard curve Table 6. Calculated F-values and significant interactions of gas-chromatography/massspectrometry volatile recovery in Merlot wines varying in concentration of ethanol (3.2%, 8%, 12%, and 16%), tannin (211 and 1500 mg/l CE) and fructose (120 and 2000 mg/l). Significance is denoted as * (p<0.1), ** (p<0.05), *** (p<0.01) Table 7. Mean concentrations (mg/l) of volatile compounds in Merlot treatments as analyzed by GC-MS. Each treatment refers to treatments listed in Table 1. Means with different letters within columns differ using Fisher s LSD (p<0.05) Table 8. Comparison of absolute recovery of 3-methyl-1-butanol, 2-phenylethanol, and eugenol based on peak area from GC-MS HS-SPME in 16 treated wines. All wines were compared to initial, untreated, spiked wine (Treatment 1), which was established as Treatment numbers refer to treatments as described in Table Page viii

9 Table 9. Calculated F-values and significant interactions of the trained panel for Merlot wines. Rep: Replicate; Pan: Panelist; EtOH: Ethanol; Tan: Tannin; Fruc: Fructose. Significance is denoted as * (p<0.1), ** (p<0.05), *** (p<0.01) Table 10. Mean intensity ratings for Merlot treatments as determined by a trained panel (n=9) using a 15 cm anchored line scale. Replicate evaluations were made over 7 days. Means with different letters within columns are significantly different (p<0.05) using Fisher s LSD. Treatment numbers refer to treatments described in Table Table 11. Pearson Correlation: correlations between chemical components and sensory attributes of aromas and flavors. Bold text indicates significance (p<0.05). 3-M-1-B: 3-methyl-1-butanol; 2-PE: 2-phenylethanol; EuOL: eugenol ix

10 LIST OF FIGURES Figure 1. Chemical structure of a) 3-methyl-1-butanol, b) 2-phenylethanol, and c) eugenol, adapted from 23 Figure 2. Interaction of ethanol and tannin on headspace concentrations of 3-methyl-1-butanol in (a) 211 mg/l CE tannin; (b) 1500 mg/l CE tannin. Letters that are different in both (a) and (b) indicate significantly different means (p<0.05) Figure 3. Interaction of ethanol and tannin on headspace concentrations of 2-phenylethanol in (a) 120 mg/l fructose; (b) 2000 mg/l fructose. Letters that are different in both (a) and (b) indicate significantly different means (p<0.05) Figure 4. Interaction of (a) ethanol and tannin and (b) ethanol and fructose on headspace concentrations of eugenol in Merlot wine. Different letters within each figure signify significantly different means (p<0.05) Figure 5. Principal Component Analysis of sensory and chemical attributes in Merlot. Blue points indicate treatment and its placement. Red points indicate sensory attributes (UPPERCASE) and chemical attributes (lowercase) Page x

11 CHAPTER I INTRODUCTION The wine industry in Washington State generates over $3 billion of revenue and brings in over 2 million visitors annually ( Recently, global warming, among other causes, has led to an increase in ethanol content of many wines in the state (Jones 2007). This is undesirable for many winemakers as an extra tax is imposed on wines containing greater than 14% ethanol. Ethanol can be decreased during winemaking using various techniques, including water addition prior to fermentation or dealcoholization of the wine after fermentation. Both techniques can drastically impact the macro-component concentrations of the wine, including ethanol, polyphenols, proteins, and polysaccharides. However, concentrations of macro-components also significantly influence the sensory profile of wines. The sensory profile of a wine is critical for consumer acceptance. The sensory profile is impacted by many attributes, including aroma, flavor, taste, and mouthfeel. All of these attributes are affected not only by the concentration of each volatile and non-volatile compound in the wine, but also by the chemical interactions among these compounds. While lower-level interactions have been studied between pairs of components (Conner et al. 1994, Dufour and Bayonove 1999a, Dufour and Bayonove 1999b, Fischer and Noble 1994, Gawel et al. 2007, Martin and Pangborn 1970, Nahon et al. 1998, Scinska et al. 2000, Singleton et al. 1975), more complex interactions have not received much focus. For example, it is known that ethanol reduces the volatility of aroma compounds because it increases their solubility in the liquid portion of the matrix, which reduces their concentration in the headspace (Hartman et al. 2002). It is also known that tannins can bind aromatic volatile compounds, reducing 1

12 their concentration in the headspace (Pozo-Bayon and Reineccius 2009), and increases in monosaccharide concentrations can reduce the solubility of some aromatic volatiles, which can lead to higher concentrations of the volatile compound in the headspace (Godshall 1997). However, it is not known how the interactions among different concentrations of ethanol, monosaccharides, and tannin affect aroma compound volatility. While some previous studies have examined higher-level interactions between wine matrix components (Jones et al. 2008, Robinson et al. 2009, Villamor 2012), they used model wine systems. Although model systems indicate potential interactions between specific components, a real wine has many other complexities not included in a model system that may enhance or interfere with the interaction effects of these specific components. Thus, the present study evaluated these interactions using a dealcoholized wine matrix, selected so as to better reflect the true nature of the wine. The dealcoholized wine served as a base or control wine, to which ethanol, tannin, and fructose concentrations were varied. Three distinct aroma compounds commonly found in Merlot were kept constant in concentration. Therefore, this study defined the interactions among ethanol, tannin, and fructose concentration on sensory and chemical attributes, including aromas, flavors, tastes, and mouthfeel of Washington State Merlot in a dealcoholized wine system. The main subobjectives were as follows: (1) To investigate the influence of ethanol concentration on perceived astringency, sourness, bitterness, and intensity of specific aroma compounds. It was hypothesized that an increase in ethanol would result in an increase in perceived bitterness, and a decrease in sourness and aroma intensities. 2

13 (2) To investigate the interaction among ethanol, tannin, and fructose on perceived astringency, sourness, bitterness, alcohol burn, and intensity of specific aroma compounds. It was hypothesized that perceived intensity of the aromas under study would decrease with increasing ethanol, but the extent of the decrease at each ethanol concentration would be dependent upon fructose and tannin concentrations. Although main effects of ethanol, tannin, and fructose would affect sourness, bitterness, alcohol burn, and astringency, it was expected that both physiological and cognitive interactions may affect each taste and mouthfeel. (3) To investigate the interaction among ethanol, tannin, and fructose on headspace concentrations of three volatile compounds in wine. It was hypothesized that headspace recoveries would decrease with increasing ethanol concentration, but the extent of the decrease would be affected by different concentrations of fructose and tannin. Specifically, tannin would further reduce recovery, and fructose would increase the recovery, although neither would dominate over the impact of ethanol concentration. 3

14 CHAPTER II LITERATURE REVIEW Importance of Wine to Washington State The wine industry in Washington began with the first wine grapes planted in The state saw significant growth in the industry until Prohibition in After the act was repealed, wineries again became a growing industry, as 42 wineries were in business in the state by It wasn t until the 1960s that commercial-scale production began, with the advent of predecessors to wineries of today such as Columbia Winery and Chateau Ste. Michelle. In the 1970 s, the industry was again rapidly expanding, as it still is today. Currently, a new winery opens in Washington State every 15 days ( The state has at least 740 wineries and sells wines of more than 30 varietals. In 2010, 160,000 tons of grapes were harvested, and 12 million cases of wine were produced ( Washington has twelve American Viticultural Areas as defined by the Alcohol and Tobacco Tax and Trade Bureau, and the number is expected to grow in the near future. With over 40,000 acres of wine grapes, Washington is the USA s second largest wine producer. The most notable varieties include Riesling, Chardonnay, Cabernet Sauvignon, Merlot, and Syrah. The industry has created $3 billion of revenue, and employs more than 14,000 people in the state. Wine has become one of the highest tax generators, and tourism has led to an influx of 2 million visitors annually ( Current Trends in Increasing Alcohol Content in Wines The earth has recently seen climate changes impacting growth of grapevines and their fruit. For instance, France has experienced a temperature increase ranging from 0.7 to 1.8 C 4

15 between 1950 and The greatest warming trend (greater than 2.5 C from 1950 to1999), however, was seen in the Iberian Peninsula, Southern France, and sections of Washington and California (Jones et al. 2005). Washington s Columbia Valley observed an increase in 149 growing degree-days between 1948 and 2002, which is similar to the average increase of 171 growing degree-days across the western growing areas of California, Oregon, and Washington (Jones and Goodrich 2008). The largest impact of this climatic change related to wine quality is observed as more rapid plant growth and unbalanced ripening (Jones 2007). Both of these phenomena result in higher concentrations of sugars in the ripe grapes available to be converted to alcohol by yeast. The problem is not solvable by simply harvesting the grapes earlier, as the flavor compounds inherently found in grape varieties are still largely undeveloped until finished ripening (Jones 2007). Thus, viticulturists and winemakers harvest grapes high in sugar and low in acid in order to harvest flavorful grapes. The rise in sugar in the grapes increases the final alcohol content in wine, and higher alcohol content wines have been trending in recent years. For example, Riesling in Alsace has increased in convertible sugars enough to produce a potential alcohol increase of 2.5% v/v in the past 30 years (Duchene and Schneider 2005). Godden and Gishen (2005) found an increase from 12.3% to 13.9% v/v alcohol in Australian red wines and from 12.2% to 13.2% v/v alcohol in white wines between 1984 and The increase in alcohol in red wines from Napa was from 12.5% to 14.8% between 1978 and 2001 (Vierra 2004). While this increase in alcohol is attributed largely to climatic warming, others have speculated other causes. For instance, Vierra (2004) cited stylistic changes resulting from higher consumer demand for bigger, bolder wines. Others cite viticultural practices and decisions. These include the increased planting of grape varieties that produce more sugar 5

16 (Gambuti et al. 2011), and harvest time (Jones 2007). Some viticulturists intentionally leave grapes on the vines for an extended hang-time to produce higher sugar content and more intense flavor development. Others do not have this intention, but see the same result, as grapes harvested earlier in the season, and therefore in the warmer parts of summer, may experience water loss due to the high temperatures. This desiccation results in higher concentrations of sugar (Jones 2007). These studies all indicate the challenges associated with increased sugar content per volume of must, and, therefore, a finished wine higher in alcohol. Although some consumers enjoy wines with higher ethanol content, the increased percentage is less cost-effective for the winemaker. The Alcohol and Tobacco Tax and Trade Bureau (TTB) charges a tax of $1.07 per gallon of wine with less than 14% alcohol. However, the tax increases to $1.57 per gallon for wines over 14% alcohol. Therefore, winemakers have an incentive to sell wines with less than 14% alcohol. Multiple winery processing methods are available to the winemaker to reduce the alcohol content in a final wine. These include a combination of saigneé and water addition, and dealcoholization by reverse osmosis, vacuum membrane distillation, pervaporation, and spinning cone column distillation, among other methods. All result in alterations of macrocomponents in the wine matrix. Methods of Alcohol Reduction Saigneé and water addition Saigneé, which means to bleed in French, is a term used for the removal of a portion of unfermented juice out of the must (Wagner 1976). This increases the ratio of grape skins to juice in the must. Theoretically, the resulting juice may have a higher solids content, which, in turn, increases tannin and fermentable sugars, among other constituents. 6

17 Saigneé may be accompanied by a second step: the addition of water back into the must. The volume of juice removed from the must may be replaced with water, thus decreasing the solids content and the ratio of grape skins to juice. This step reduces ethanol in the final wine. The combination of saigneé and water addition has been shown to reduce ethanol while not significantly changing aromas or flavors. Baiano et al. (2009) found that saigneé significantly increased total phenolic concentration in musts and fermented wine compared to traditional, delestage, delayed punching-down, heating of must, cryo-maceration, and prolonged maceration techniques. Harbertson et al. (2009) observed a decrease in ethanol and an increase in tannin and perceived sourness as the saigneé run-off and water addition volume increased, although no other differences in sensory perception were observed. Other studies involving the effects of saigneé before fermentation have not been found. Dealcoholization by reverse osmosis Dealcoholization by reverse osmosis is a method of removing alcohol in a finished wine. Under pressure, the wine is subjected tangentially via a flow pipe to a semipermeable filter. The filter removes water, alcohol, and other small molecules in the wine, forming two partitions: the permeate and the retentate. The permeate (containing water and alcohol) is distilled to remove alcohol, and the resultant permeate, less the alcohol, is added back to the retentate. Demineralized water can also be used to replace the permeate. It is generally assumed that dealcoholization by reverse osmosis does not remove volatile compounds (Catarino et al. 2006). However, Kavanagh et al. (1991) observed a high loss of volatile compounds in beers dealcoholized by reverse osmosis from 4.80% v/v alcohol to 2.04% and 0.96% v/v alcohol. The volatile compounds studied in that particular experiment 7

18 included esters, alcohols, and organic acids. In contrast to compounds contributing to the aroma, other beverage constituents are much less affected by reverse osmosis. For instance, Gambuti et al. (2011) observed no differences in total phenolic content of four wines (including Merlot) subjected to a decrease in 5% ethanol. Meillon et al. (2010) found a decrease in liking of a wine dealcoholized to 7.9% v/v alcohol versus a 13.4% v/v alcohol control, and attributed this result to decreases in complexity and the aromatic profile. In addition to a decrease in heat, the dealcoholized wine was perceived as more astringent than the control. When 8.44 g/l sugars from concentrated grape juice were added to the dealcoholized wine, the berry attribute increased significantly. Both methods of reducing ethanol in wine, saigneé/water addition and dealcoholization, influence the concentrations of other macro-components in wine, including polyphenols and simple sugars. This also influences sensory attributes of the wine, including alcohol burn, astringency, sourness, bitterness, aroma, and flavor. Wine Sensory Attributes Alcohol burn Alcohol burn may also be described as heat or hotness in wine. In wine, this is generally caused by ethyl alcohol. This burning sensation occurs when a chemical compound stimulates the nerve endings of the trigeminal nerve. The trigeminal nerve is composed of many fibers that surround fungiform papillae and are distributed randomly throughout the oral cavity (Whitehead et al. 1985). This is believed to mediate oral burn. The intensity of the burning sensation is dependent upon the chemical concentration of the irritant. The threshold concentration for irritation by ethanol was found to be 9% by 8

19 Mitchell and Gregson (1968), but 14% by Diamant et al. (1963). Gawel et al. (2007) found that the intensity of the burning sensation in wine increased as ethyl alcohol increased from 11.6% to 12.6% to 13.6% v/v. Another study (Yu and Pickering 2008) showed that the difference threshold in Zinfandel was between 1.08 and 1.14% v/v for orthonasal perception, or 1.31 and 1.32% v/v for retronasal perception. Differences in perception of alcohol burn may be explained by taster status. Bartoshuk et al. (1993) found that super-tasters have more fungiform papillae, and therefore have more trigeminal fibers, leading to more intense oral irritation by ethanol. Duffy et al. (2004) also found a correlation between 6-propyl-2-thiouracil (PROP) tasting status and trigeminal irritation intensity: non-tasters experienced less burn than tasters. Prescott and Swain- Campbell (2000) also observed a significant difference in intensities of ethanol between PROP non-tasters and tasters. They also observed desensitization to the irritation as intensity of alcohol burn decreased for repeated tastings over a ten-minute period. In addition to taster status, other parameters that affect ethanol threshold include sensory panel experience, ethnicity, and wine consumption level (Yu and Pickering 2008). Astringency Astringency, like alcohol burn, is a sensation caused by stimulation of the trigeminal nerve ends. Astringency is perceived when a series of complexing reactions between a compound and the proteins of the mouth and saliva causes the proteins to precipitate out of solution (Noble 1994, Noble 1998). In wine, this is usually a combination of polyphenols (e.g. tannins) and proline, an amino acid found in saliva proteins (Haslam and Lilley 1988). The result is a drying, roughening, or even puckering mouthfeel that can increase in intensity with higher concentrations of the astringent compound (Kallithraka et al. 1997a). 9

20 Perception of astringency can also be influenced by other macro-components. Scinska et al. (2000) found that astringency intensity can be masked and decreased by ethanol. The authors attributed this to the cognitive interactions between the perception of bitter and sweet tastes, whereby certain compounds, such as ethanol, may be perceived as either bitter or sweet. Other authors have also found that perceived astringency decreases as ethanol concentration increases (Fontoin et al. 2008, Gawel 1998). Serafini et al. (1997) claimed the decreased perception of astringency was due to ethanol s interference between the binding reaction between salivary proteins and tannins. Vidal et al. (2004a), who also observed a decrease in perceived astringency with an increase in ethanol, attributed to the result to the disaggregation of tannin complexes by ethanol, resulting in smaller, less astringent molecules. Conversely, Noble (1998) found no effect on perceived astringency by ethanol, and Meillon et al. (2009) found that astringency decreased in wines that had been dealcoholized by reverse osmosis. Other macro-components have the ability to affect astringency, as well. For instance, Vidal et al. (2004a, 2004b) found polysaccharides to interfere with procyanidin aggregations, potentially decreasing astringency. Sourness Sour taste, also termed acidity, is caused by hydrogen ions in foods. When a food is ingested, the acid dissociates into a hydrogen ion and an anion. The hydrogen ion binds to the receptor membrane via ion channels and as concentration increases, intensity increases. When ph is exclusively considered, sourness increases with a decrease in ph (Fischer and Noble 1994). An increase in titratable acidity (TA), which equates to an increase in hydrogen ions, also increases sourness intensity (Norris et al. 1984). However, sourness as it relates to titratable acidity is also dependent on the type of acid, as the anion may also bind, reducing 10

21 the net positive charge on the receptor membrane, and therefore reducing perceived acidity (Beidler 1978). The difference threshold for titratable acidity is very low, with only 0.02 to 0.05% differences in concentration required to cause a difference in perception (Amerine et al. 1965). Various wine constituents affect sourness. In winemaking, the balance between sourness and sweetness has always been a challenge. First and foremost, it is important to harvest grapes at the optimal point for balanced sweetness and acidity. However, if a wine is too sour, winemakers may add a sweetener to reduce the acidity. Scientifically, it has been reported that sourness can be suppressed by sweeteners (Bonnans and Noble 1993, Zamora et al. 2006). Ethanol may also affect sourness. Martin and Pangborn (1970) observed a decrease in the sourness of citric acid when ethanol increased from 4% to 24%. The results of Fischer and Noble (1994) were consistent with this: tartaric acid sourness decreased slightly with an increase in ethanol from 8% to 11% to 14% v/v, but the effect was only significant when the ph was 3.2. They also studied the interactions from different ph levels and found that the effect by ethanol on sourness was most significant at ph of 3.2, while no difference was found when the ph was 3.8. They attributed the decrease in sourness to a masking effect, and stated that ethanol may interact with ion-channel proteins, affecting sourness (Fischer and Noble 1994). The authors also found that sourness was not affected by catechins (100 to 1500 mg/l), nor did they find an interaction between ethanol and tannin on sourness. Bitterness Many compounds have been found to contribute to bitterness. These include amino acids, peptides, sulfimides, ureas and thioureas (such as PROP and phenylthiocarbamide [PTC]), esters and lactones, terpenoids, and phenols and polyphenols (Brieskorn 1990). Plant- 11

22 based bitter compounds, including phenols, flavonoids, isoflavones, terpenes, and glucosinolates, are all known to elicit bitter taste (Bravo 1998). In wine, phenolics are responsible for bitterness and astringency (Bravo 1998, Delcour et al. 1984), two sensations that are often confused. The distinction can be defined by the molecular weight of the phenolics. Plant tannins are generally greater than 500 Da (Bravo 1998), and low molecular weight compounds produce a bitter taste, while high molecular weight compounds evoke an astringent mouthfeel (Noble 1994). Although wine is expected to have some bitterness due to associations with ethanol (Guinard et al. 1996, Mattes 1994), the perception of bitterness is described as unpleasant, and can even evoke pain in some individuals. Bitterness ratings have also been correlated to mouth roughening and drying, especially when a compound is presented at higher concentrations (Kallithraka et al. 1997a). Bitterness has a low detection threshold in comparison to other food constituents (Hladik and Simmen 1996, McBurney 1978). Quinines have a threshold as low as 25 µmol/l, while sucrose can be as much as 10,000 µmol/l, (Hladik and Simmen 1996). Additionally, bitter taste has a longer duration than sweet, salty, or sour. For instance, the reaction time for sucrose is 0.55 sec, 0.37 sec for sodium chloride; 0.48 sec for citric acid, and 0.80 sec for quinine hydrochloride (McBurney 1978). The question of bitterness stimulation is widely disputed. McBurney hypothesized that there are 3 or more bitter receptors that can respond to different compounds: quinine, urea, and PTC or PROP (1978). Other mechanisms have been studied as well. For instance, the G protein-coupled receptor signaling pathway (the same mechanism for sweetness perception) is thought to transduce certain bitter compounds (Bartoshuk et al. 1994, Schiffman et al. 1995). 12

23 Other studies have proposed between 40 and 80 bitter taste receptors, called T2Rs, which can be expressed in circumvallate, foliate papillae, and fungiform papillae (Adler et al. 2000, Chandrashekar et al. 2000). In a study by Chandrashekar et al. (2000), fungiform papillae had the largest number of taste receptors, allowing for the perception of multiple bitter tastants on the same cell. This would account for why so many compounds can elicit the same bitter taste. Previous research in interactions among bitterness and macro-components of wine has been conducted. Bitterness can be reduced by the addition of sucrose (Noble 1994, Noble 1998, von Sydow et al. 1974), and enhanced by the presence of 4 to 24% ethanol (Martin and Pangborn 1970). Various studies have studied the interactions between multiple constituents and the effects they have on bitterness. For instance, Fischer and Noble (1994) found that bitterness was increased by interactions among ethanol, catechin, and a rise in ph, although ethanol had the largest influence. They also observed that perceived bitterness was increased by catechin, but not within the ph range of 3.2 to 3.8. They explained that the isoelectric point of certain proteins found in saliva might be close to 3.8, creating protein ionization. This allows the proteins to bind more frequently with catechin, preventing catechin compounds from binding to bitter taste receptors, and therefore reducing bitterness (Fischer and Noble 1994, Hagerman and Butler 1978). Finally, Singleton et al. (1975) found that high astringency can mask bitterness. Aromas Aroma is perhaps the most complex constituent of wine. It has been proposed that there are over 800 volatiles in wine (Etievant 1991) but that less than 100 of these are odoractive (Ferreira et al. 2000, Guth 1997a). Aroma compounds in wine are formed during and 13

24 vary depending on grape fruit development, processing techniques in the winery, and yeast fermentation of grape juice into wine. Aromas are detected when a volatile compound is in the vapor phase, mixed into the air, and passed through the nasal cavity (Clarke and Bakker 2004). Buck (2000) proposed a new model on aroma discrimination. Approximately 1000 olfactory receptors exist within the nasal cavity, each expressed by individual olfactory neurons. Neurons of the same receptor connect to the same set of glomeruli (Bozza and Mombaerts 2001, Mombaerts et al. 1996). Each receptor can accept multiple odorants, and each odorant can attach to multiple receptors (Malnic et al. 1999). The combined effect of the activation of different receptors allows for the brain to experience, remember, and distinguish thousands of patterns, each associated with a specific aroma (Rubin and Katz 1999). Theoretically, as the concentration of an odorant increases, the intensity also increases. However, there are multiple factors that change the perception and intensity of aromas. First, concentration itself affects perception of aromas. An odorant at low concentration might have a completely different description from the same odorant at high concentration (Amerine and Roessler 1975). Second, the presence of two or more odors can also alter how an aroma is perceived. These include masking, additive effects, synergistic effects, or no effect at all (Amerine and Roessler 1975). Masking occurs when an aroma is easily perceived when presented individually, but less intensely when another compound is also present. Additive effects occur when the intensities of two mixed compounds are the sum of the intensities of each compound when presented individually. Synergistic effects occur when a compound presented with another compound appears stronger than the theoretical intensity based on 14

25 concentration alone. These perception effects can occur psychologically, but may also occur due to chemical interactions in the wine. One of the most influential ideas in the chemistry of aromatic volatility lies in a compound s solubility. The ratio of the concentration of a volatile in the headspace to the concentration in the liquid is called the partition coefficient. This ratio, (and therefore the headspace concentration) is affected mainly by solubility, boiling point, and molecular weight of the aroma compound (Pozo-Bayon and Reineccius 2009). However, the solubility and volatility can be influenced by many interactions with macro-components in the wine, such as polysaccharides, proteins, and polyphenols. For instance, hydrophobic aromas interact with hydrophobic components in the wine, such as ethanol, proteins, and even other aroma compounds, resulting in higher odorant solubility in the liquid and a lower headspace concentration. In a study measuring esters, aldehydes, and alcohols, the sensory threshold of each component was reduced by interactions among the components (Conner et al. 1994). Various studies have researched the impact of other volatile and non-volatile constituents individually on aroma perception, and the results are conflicting. Gawel et al. (2007) found no significant differences in aroma or flavor intensity with increasing ethanol between 11.6 and 13.6%. Conner et al. (1994) found similar results with ethanol concentrations up to 17%, but found decreases in activity coefficients of esters at ethanol concentrations higher than 17%, relating inversely to acid chain length. They discussed how at concentrations less than 17%, ethanol is mono-dispersed in water, with similar properties as pure water, but at concentrations higher than 17%, ethanol molecules form hydrophobic aggregations, in which odorants are more soluble (Conner et al. 1994, Escalona et al. 1999). 15

26 Contrastingly, other studies have found differences in aroma perception with ethanol changes. For instance, Grosch (2001) found an increase in intensity of fruity and floral aromas when the ethanol concentration was reduced from 10% to 7%. In addition to differences due to a reduction in the partial pressure and, therefore, an increase in the partition coefficient, physiological reasoning may be applied. It is thought that ethanol increases the fluidity of cell membranes, allowing for easier transport of small and charged molecules: the efficiency of trans-membrane movement is greatly increased when the lipid bi-layer is disordered (Hunt 1985). Flavors Flavors emerge from a range of complex interactions between sample components, human physiology, and psychological factors. Flavor depends not only on concentration of volatiles, but also on interactions between volatiles, presence of non-volatile materials, and ethanol concentration (Goldner et al. 2009). Flavor includes tastes, retronasal olfactory perception, and trigeminal sensations. Small changes in these variables can change the flavor of a wine dramatically. Previous work has shown the cognitive integration that can occur when tastes and smells are combined. One study demonstrated that when an odor compound and a taste compound are presented together at subthreshold concentrations, the combination is still detectable (Dalton et al. 2000). Other studies have shown that when an odor compound is increased, the associated taste judgment increases, and vice versa (Bonnans and Noble 1993, Murphy et al. 1977, Murphy and Cain 1980). Also, presenting two compounds in a solution may not elicit an additive result. Instead, intensity ratings of the mixture are less than the added intensities of each individual compound (Murphy et al. 1977, Murphy and Cain 1980). 16

27 Another important factor is the mechanism by which odors are perceived as a flavor. The odors involved in flavor perception are experienced via retronasal olfactory perception. As the sample enters the oral cavity, the mouth rapidly brings the sample up to body temperature, and volatile odorants are released from the matrix through the back of the mouth into the nasal cavity. The environment in which these odors are experienced is unlike the environment in which aromas are experienced in that perception is internal, not external as with orthonasal perception. The difference in how these compounds are perceived influences intensity ratings. Previous research has shown that retronasal odors are less identifiable than their orthonasal counterpart, because of diffusion and subsequent absorption or adsorption of the volatile compound into the lungs and naso-oropharyngeal surfaces (Rozin 1982). Physiological factors People differ in their sensitivity and therefore in their perception of all attributes discussed previously. Perception of attributes can be influenced by physiological differences among individuals, in addition to wine matrix component interactions described previously. One of the important physiological differences is taster status. The inability of some humans to taste phenylthiocarbamide (PTC) was discovered in 1931 by Fox (1931). He attributed it to genetic variances between the populations. Since then, research has shown that differences are not dependent on genetic variation alone, but also based on gender and race. For instance, Fernberger (1932) found that females are more acute tasters than males, but Boyd and Boyd (1937) found that the gender effect was large in Wales but small in Cairo. Much work has been done to describe the difference between those who can taste PTC (tasters) and those who cannot (non-tasters). Another compound, 6-npropylthiouracil (PROP) was developed to replace PTC and its sulfurous odor (Fischer and 17

28 Kaelbling 1966), and even more effects were found, including personality type, food preferences, smoking habits (Fischer 1971), and even the identification of a subset of tasters: supertasters (Bartoshuk et al. 1992). Supertasters are able to perceive some bitter compounds as intensely bitter (Bartoshuk et al. 1992) and ethyl alcohol as more bitter and irritating (Bartoshuk et al. 1993), among other differences. The difference between nontasters, tasters, and supertasters lies in anatomically different taste buds. Supertasters have a significantly larger amount of fungiform papillae than tasters, who have more fungiform papillae than nontasters. In the same way, supertasters have a much larger taste pore density than tasters and nontasters (Bartoshuk et al. 1994). Today, a sample of PROP (0.032 M) and a sample of NaCl (0.1 M) is evaluated by each panelist. Those who rate NaCl as much higher in intensity than PROP are considered nontasters, those with similar ratings for both NaCl and PROP are considered tasters, and those where PROP intensity is rated much higher than NaCl are considered supertasters (Tepper et al. 2001). Taster status may affect perception and liking of many compounds. PROP tasters and supertasters tend to perceive caffeine, quinine, and other bitter compounds, as well as sweet-tasting compounds such as sucrose as more intense (Bartoshuk et al. 1994). PROP tasters and supertasters also have higher sensitivity to oral irritation from compounds such as capsaicin (Karrer et al. 1992) and benzyl alcohol (Prescott and Swain-Campbell 2000). All of these differences are challenges to overcome when performing sensory research, as small differences among panelists can lead to ambiguous results. Wine Matrix: Volatile and Non-Volatile Components The sensory attributes previously reviewed have been studied extensively by those interested in cognitive interactions between matrix components and human perception. An 18

29 additional research topic, lies in the chemistry of the components. Chemical interactions between matrix components can influence which components are actually available to be sensed by humans, imipacting the consumer perception of the wine. Tannin Tannins refer to a group of compounds that elicit astringency in the mouth by interacting with salivary proteins. Tannins in wine are made up of catechin and epicatechin as monomers, dimers, and oligomers, and are also known as flavanols, flavan-3-ols, condensed tannins, procyanidins, proanthocyanins, or proanthocyanidins (Cheynier et al. 2006). Grape seed tannins are procyanidins formed of catechin, epicatechin, and epicatechin 3-gallate units. Tannins from the skin reach approximately 30 mean degrees of polymerization (mdp) (Souquet 1996), compared with about 10 in the proanthocyanidins from seeds (Prieur et al. 1994) and stems (Souquet et al. 2000). Tannins are naturally found in grains (sorghum, millet, and barley), peas, carobs, dry beans and legumes, fruit, tea, and wine (Chung et al. 1998). In wine grapes, phenolic compounds are found in the solid parts of the grape, including skins, pulp, and seeds, and can be extracted by maceration during winemaking (Jackson 2000). Tannins in wine are found as flavans [catechin (Kallithraka et al. 1997a) and epicatechin (Kallithraka et al. 1997b)], flavonols [quercitin (Trock et al. 1990)], and phenolic flavonoids (catechin mono- and polymers). Skins, pulp, and seeds determine the potential concentrations of tannin, but winemaking techniques can change the final composition in wine (Katalinić 1997, Katalinić 1999). Crushing and pressing alter the phenolic composition in wines (Lamuela-Raventos and Waterhouse 1994). However, fermentation of juice on the skins influences the phenol levels 19

30 of the must, but it is dependent upon skin contact time. The total phenol concentration in finished red wines is usually between 1000 and 3500 mg/l (Blanco et al. 1998, Dufour and Bayonove 1999a, Noble 1998). After fermentation is complete, tannins are still unstable. With aging, tannins undergo enzymatic and chemical changes, such as polymerization and precipitation, (Cheynier et al. 2006, Noble 1998). While oxidation and aggregation with anthocyanins can occur, yielding higher molecular weight molecules, cleavage reactions are also possible (Haslam 1980, Vidal et al. 2002). The interaction among tannins and other molecules is dependent upon a number of factors. These include molecular size, flexibility, solubility of the tannin, ph, and characteristics of the other molecule (Haslam and Lilley 1988). Solubility is highly influential in tannin interactions. For instance, the solubility of a tannin changes based on its isoelectric point and the ph of the wine. Solubilized (ionized) tannins are unable to bind with other molecules, reducing the interactions between the two compounds. Solubility of tannins is also affected by ethanol, due to hydrophobic interactions. Tannins, which are large molecules, are largely non-polar, increasing their affinity to and interactions with other non-polar compounds, including ethanol. In addition to ethanol, tannins also have the ability to aggregate with themselves, forming larger complexes (Fulcrand et al. 1996). This aggregation increases protein precipitation and interaction with other molecules. However, polysaccharides can interfere with tannin-tannin aggregation (Riou et al. 2002), increasing solubility of the tannin, and decreasing interaction with other wine constituents. 20

31 Fructose Fructose is a monosaccharide with 6 carbons. In solution, it reacts reversibly with a hydroxyl group to form either a chain, furanose (5-carbon ring), or pyranose (6-carbon ring) (Sanz and Martinez-Castro 2009). It is the most water-soluble of all sugars, and is also soluble in polar solvents such as alcohol (Sanz and Martinez-Castro 2009). Fructose is the sweetest naturally-occurring sugar, as it is about 1.65 times as sweet as glucose, and 1.14 times as sweet as sucrose (Sanz and Martinez-Castro 2009). In red wines, fructose concentrations of less than 1.5 g/l are considered dry and the sweetness due to these sugars is not detectable on the palate (Jackson 2000). Sweetness can begin to be detected around 2 g/l, although most people require 10 g/l to detect distinct sweetness (Jackson 2000). Fructose concentration in red wines can range from not detectable to 2.5 g/l (Restani 2007). Fructose is metabolized by yeast during fermentation as an energy source. The byproducts of this fermentation are ethanol and carbon dioxide (Zamora 2009). After fermentation by S. cerevisiae is complete, the unfermented glucose and fructose are termed residual sugar (Constantini et al. 2009). Although both glucose and fructose occur in high levels in the must and decrease during fermentation, the ratio of fructose to glucose increases dramatically, because glucose is strongly preferred for fermentation by yeasts over fructose (Sanz and Martinez-Castro 2009). Ethanol Ethanol is the main volatile compound found in alcoholic beverages. It is composed of a polar hydroxyl group attached to a non-polar combination of a methylene and methyl group, giving it the ability to become miscible in both water and organic compounds, and to interact with many types of molecules, including aroma compounds. Factors affecting the volatility of 21

32 ethanol (and other volatile compounds) include temperature, pressure, and non-covalent bonding interactions with other compounds, volatile or non-volatile. An increase in both temperature and pressure tends to increase volatility. Non-covalent bonding between volatiles and non-volatiles decreases the volatility of the volatile component, while non-covalent bonding between volatiles and other volatiles can either increase or decrease volatility, depending on solubility of each compound. Ethanol is produced by the transformation of reducing sugars into ethanol by S. cerevisiae. Because the initial concentration of sugar varies between wines, the ethanol concentration in the final wine also varies. Generally, ethanol content ranges from between 10% and 15% (Pozo-Bayon and Reineccius 2009). Aromatic volatile compounds According to the literature, 3-methyl-1-butanol, also known as isoamyl alcohol, is a common aroma component in many foods, including Merlot wines (Figure 1). It has been described as fusel (Escuadero et al. 2007), malty (Gürbüz et al. 2006), pungent (Abraham and Berger 1994), and caramel (Villamor 2012). 3-Methyl-1-butanol has a published odor threshold of 30 mg/l in a 10% w/w ethanol in water solution (Guth 1997b). 3-methyl-1- butanol has a molecular weight of g/mol, and boils at 130 C. It is miscible in ethanol and soluble in water, up to 54 mg/ml. Buttery et al. (1988) found 3-methyl-1-butanol in cooked rice, and determined the odor threshold to be 300 µg/l in water. In wine, several studies have identified 3-methyl-1-butanol. Escuadero et al. (2007) attempted to determine which odor compounds were present in the most important five wine varieties from Spain, that is, those that contribute the most to the aromatic profiles. Among the many compounds detected was 3-methyl-1-butanol. They described the odor as fusel, and 22

33 a) b) c) Figure 1. Chemical structure of a) 3-methyl-1-butanol, b) 2-phenylethanol, and c) eugenol, adapted from 23