Abstract
The impact of fining on the sensory and chemical properties of Washington State white wine was investigated. Unfined, commercially prepared Chardonnay and Gewürztraminer wines were treated with bentonite (1000 mg/L), isinglass (60 mg/L), Sparkalloid (360 mg/L), activated charcoal (450 mg/L), whole milk (500 mg/L), or wheat gluten (400 mg/L). Ethyl dodecanoate was the only volatile compound to significantly differ among Chardonnay treatments, which was highest in the control (0.031 mg/L) and lowest in Chardonnay treated with bentonite (0.017 mg/L). Conversely, a number of volatile compounds varied significantly among Gewürztraminer treatments. Ethyl acetate was significantly highest in the activated charcoal treatment (25.4 mg/L), while lowest in the Sparkalloid treatment (22.1 mg/L). In addition, Gewürztraminer treated with activated charcoal contained high concentrations of higher alcohols. Wheat gluten significantly decreased the concentrations of 1-hexanol, 3-methyl-1-butanol acetate, and 2-methyl-1-butanol. Benzeneethanol was significantly lower in the Sparkalloid, wheat gluten, and bentonite treatments. Conversely, benzeneethanol was highest in the isinglass (85.2 mg/L) and activated charcoal (74.7 mg/L) treatments. 2-Phenylethyl acetate and linalool were lowest in Gewürztraminer fined with bentonite. No significant differences were found among treatments for either varietal when the wines were subjected to difference testing (duo-trio) by an untrained panel (p ≥ 0.05). No differences were found among Gewürztraminer treatments evaluated by a trained panel, whereas differences in spicy aroma and floral/honey flavor were observed among Chardonnay treatments (p ≤ 0.05). This study demonstrated the impact fining can have on the chemical and sensory properties of wine and confirmed the importance of selecting the appropriate type of fining agent in order to maintain wine quality.
Fining is critical to consumer acceptance of white wines, as bottle haze may eventually lead to consumer rejection and economic loss to the winery (Lopez et al. 2001). Together with racking and filtration, fining agents improve clarity and increase shelf life. Fining alters the chemical composition (i.e., protein or polyphenol) of wine (Zoecklein et al. 1995, Boulton et al. 1996, Gómez-Plaza et al. 2000, Castillo-Sánchez et al. 2006, Ribéreau-Gayon et al. 2006). Fining may also impact the sensory quality of wines, the extent depending on the selected agent and wine (Razmkhab et al. 2002).
Fining refers to the deliberate addition of materials to a wine followed by the precipitation of components (Boulton et al. 1996). One agent, bentonite, is commonly used to reduce protein content and aids in achieving a heat-stable wine. Proteinaceous fining agents help reduce browning and astringency by removing tannins and polymeric phenols. Synthetic substances, such as polyvinylpolypyrrolidone (PVPP), can be added to reduce polyphenols, whereas carbon agents decolorize and deodorize. For example, fining with activated charcoal reduced the color intensity of sherry wine (Lopéz et al. 2001), and fining with casein, potassium caseinate, isinglass, egg albumin, and gelatin reduced the color intensity of white wine (Cosme et al. 2008).
Fining can improve wine quality. In Champagne wine casein or casein plus bentonite improved the foaming properties (foam height and foam stability) compared to bentonite alone and a control (Dambrouck et al. 2005). In another study, gelatin-fined wines subjected to sensory evaluation were affected by protein fining and were significantly less astringent than unfined wine (p < 0.05) (Maury et al. 2001). A study on Sevyal blanc wine concluded that treatment with bentonite and PVPP was as effective as sulfur dioxide in preventing wine browning and could be used to maintain wine color and quality (Main and Morris 1991).
Conversely, fining agent performance can be unpredictable and may result in overfining, excessive lees production, and a loss in wine quality. Albariño wines fined with bentonite had a lower concentration of terpenes and C13-norisoprenoids (13%) and C6 alcohols (33%) (Armada and Falqué 2007). These volatile compounds are responsible for the varietal aroma of Albariño wines. Similarly, fining agents have reduced the concentration of aromatic compounds such as total flavonoids, ethyl esters, acetates, and alcohols in various wine varietals (Voilley et al. 1990, Moio et al. 2004).
Red wine color is sensitive to fining. Red Vinhão wine fined with PVPP had a greater color loss than gelatin, egg albumin, and casein after 20 months of storage (Castillo-Sanchez et al. 2006). The PVPP also caused a greater loss in anthocyanin content in the wines exposed to rotation by a rotary vat.
Therefore, it is essential to select the correct agent and concentration when fining to maintain wine integrity and quality. The primary objective of this study was to evaluate the impact of fining on the sensory and chemical properties of white wines, specifically Washington State Chardonnay and Gewürztraminer. Bentonite, isinglass, Sparkalloid, and activated carbon were selected as fining agents based on industry demand and lack of published data concerning sensory impact on wine. Additionally, the efficacy of wheat gluten and whole milk—less researched agents—was investigated. It was hypothesized (1) that fining would affect the chemical properties of protein concentration, turbidity, and volatile compound composition as well as sensory properties such as aroma, taste, and flavor of white wine and (2) that fining agents would behave differently from one another.
Materials and Methods
Wine.
Unfined Gewürztraminer (2.9% w/v residual sugar, 0.8 g/L titratable acidity [TA], pH 3.07, 11.4% v/v ethanol) and Chardonnay (0.1% residual sugar, 0.6 g/L TA, pH 3.51, 14.4% v/v ethanol) wines were obtained from Chateau Ste. Michelle winery (Paterson, WA) at the end of alcoholic fermentation. Neither wine had undergone malolactic fermentation. Both wines were held in sterilized, food-grade, 120 L plastic drums at ~8°C until fined. Alcohol content was measured using an ebulliometer (Presque Isle Wine Cellars, North East, PA); pH was measured using an Accumet AB15 Plus pH meter (Fisher Scientific, Pittsburg, PA); and TA was analyzed using a TitroLine Easy autotitrator (Schott Instruments, Deutschland, Germany); residual sugar values were provided by the winery.
Enological products.
Sodium bentonite was obtained from Crosby-Baker (Westport, MA), and a 5% (w/v) slurry was prepared in deionized water and allowed to settle for 24 hr at room temperature. Sparkalloid powder (Cellar Pro, Steinbart Wholesale, Portland, OR) was added to deionized water at a 3% (w/v) concentration and boiled for 10 min before wine addition. Two hours before wine addition, a 20% (w/v) slurry of isinglass (Ichtyocolle, Scott Labs, Walla Walla, WA) was prepared. A 5% (w/v) slurry of GemPro HiQ wheat gluten (Manildra Group, Shawnee Mission, KS) was also prepared and stirred overnight. Whole milk (Ferdinand’s Creamery, Pullman, WA) and activated carbon (hydrocarbon trap fill-adsorbent charcoal; Varian, Palo Alto, CA) required no prior preparation.
Wine fining.
Trials were conducted in replicate to determine appropriate concentrations of each fining agent. The treatments applied were an unfined control, bentonite (500, 750, 1000, 1250, and 1500 mg/L), isinglass (15, 60, 75, 90, and 105 mg/L), Sparkalloid (300, 360, 420, 480, and 540 mg/L), activated charcoal (100, 250, 350, 450, and 500 mg/L), whole milk (50, 250, 750, and 1000 mg/L), and wheat gluten (50, 100, 200, 300, and 400 mg/L). Proportions of the six fining agents were pipetted into 50 mL wine. Wines were then mixed and allowed to settle at 13°C. After seven days, turbidity levels were measured at room temperature. Samples were also subjected to a heat stability test (Pocock and Rankine 1973). The lowest concentration of fining agent that yielded the wine with the greatest stability was used in large-scale fining.
For the large-scale fining, each wine varietal was racked off into 20 L glass carboys and free SO2 levels were adjusted to 20 to 30 mg/L using potassium metabisulfite (JT Baker, Phillipsburg, NJ). Sulfur levels were monitored using the aeration-oxidation method for SO2 analysis (Buechsenstein and Ough 1978). The following concentrations of fining agent were added to the 20 L carboys: 1000 mg/L bentonite, 60 mg/L isinglass, 360 mg/L Sparkalloid, 500 mg/L whole milk, 400 mg/L wheat gluten, and 450 mg/L activated charcoal. The fining agents were stirred into the wines and allowed to settle at ~12.8°C for 7 days.
After seven days, turbidity levels were measured in triplicate before filtration through a 0.45 μm membrane filter (Vitipore Plus 0.45 μm, Gusmer Enterprises, Fresno, CA), bottled, and closed with natural corks (Tri-State, Moscow, ID). Wines were stored at ~4°C until chemical and sensory analysis.
Chemical analysis.
Chemical analysis was performed in triplicate on all wine treatments after four months storage at 4°C. All volatile standards, as well as potassium hydrogen phthalate, tartaric acid, and bovine serum albumin, were purchased from Sigma-Aldrich (St. Louis, MO). Sodium hydroxide (NaOH) and sodium chloride (NaCl) were purchased from JT Baker). Coomassie blue reagent was purchased from Bio-Rad Laboratories (Hercules, CA).
Turbidity measurements were made after fining and before filtration and after storage using a turbidimeter (Orbeco-Hellige, Farmingdale, NY) and expressed in nephelometric turbidity units (NTUs). After bottle storage, protein concentrations were measured using a modified Bradford assay using Coomassie Brilliant Blue (Bradford 1976) with a Genesys 10 UV spectrophotometer (Thermo Electron Corp., Waltham, MA).
Volatile analysis.
Volatile compounds were extracted from the wines through solid-phase microextraction (SPME). As a sample preparation technique, SPME provides advantages over conventional preparation techniques. Specifically, SPME takes less than one hour to complete, is less expensive, does not require solvent extraction, and produces fewer artifacts during sample preparation. SPME has also been used extensively for the characterization of flavor compounds in wines (Howard et al. 2005, Rocha et al. 2001). Prior to use, a polydimethylsiloxane/divinylbenzene fiber was preconditioned at 250°C for 30 min. The optimized SPME parameters were as follows: 2 mL sample was placed into a 4 mL amber glass vial with 0.65 g NaCl (6 M) and a magnetic stir bar (Whiton and Zoecklein 2000, Howard et al. 2005). The vial was securely capped with a Teflon-coated silicon septum and allowed to equilibrate for 5 min while stirred magnetically at ambient temperature (~22°C). After equilibration, the sample headspace was extracted for 45 min at ambient temperature while being magnetically stirred and introduced into the injection port of the gas chromatograph. For each fining treatment, wines were prepared and evaluated in triplicate.
Gas chromatography-mass spectrometry (GC-MS) analyses were carried out using an HP 5890 gas chromatograph coupled to an HP-5973 mass selective detector (Hewlett-Packard, Palo Alto, CA). The GC was equipped with a 0.75 mm i.d. deactivated injection liner (Supelco, Bellefonte, PA). Chromatographic separations were achieved using a 60 m length, 0.32 mm i.d., 0.25 μm film thickness, DB-1 column (J&W Scientific, Folsom, CA). The injector temperature was maintained at 200°C. The injection was made in splitless mode at 200°C for 5 min. Helium was used as the carrier gas with a constant flow rate of 1 mL/min. The oven temperature settings were 33°C for 5 min; 5°C/min ramp to 50°C; 2°C/min ramp to 225°C; held at 225°C for 13 min. The mass spectrometer was operated in electron impact (EI) mode at 70 eV. The temperature of the detector was maintained at 230°C. Data were collected in SCAN mode from the mass range 35 to 550 m/z. Identification of volatiles was confirmed using retention time of spectra match of standard compounds. Secondary confirmation was conducted using NIST mass spectra library (National Institute of Standards and Technology, Gaithersburg, MD).
Prior to analysis, internal standards 1-pentanol and 1-dodecanol were added to each calibration standard and sample. These standards were selected based on their response recoveries and retention time when compared to target volatile compounds. Both internal standards were added to each calibration standard and sample to yield a concentration of 10 mg/L 1-pentanol and 0.5 mg/L 1-dodecanol in solution. For quantification, the peak area of each analyte was normalized using the area of the internal standard. The relative areas were then interpolated using the standard curve generated for each specific analyte.
Sensory evaluation.
Forced choice duo-trio test.
Bottled wines were stored for four months at 4°C prior to difference testing by an untrained panel. A forced choice duo-trio test (constant reference) was used to determine whether the fined wines differed from the unfined (control) wine. Each varietal was evaluated over a two-day period, with three fining treatments (and a control) evaluated each day, for a total of four evaluation days. Thirty panelists participated on each evaluation day and received a nonmonetary incentive for their participation. Assigning an α risk of 0.05 and a β risk of 0.5, the power of the test was 0.5 (Meilgaard et al. 1999). Panels were conducted in individual sensory booths in the Sensory Facility, Food Science and Human Nutrition, Washington State University (WSU), Pullman. Booths were equipped with yellow colored lights to mask visual differences in the wines. Demographic data were collected from panelists, who were recruited through email, Internet announcements, and bulletins posted throughout the WSU Food Science and Human Nutrition building.
Each panelist was presented with water, crackers, a napkin, and a cuspidor. Panelists were presented with three samples per flight, for a total of three flights. Per flight, the control sample (labeled reference) was presented along with two samples (one control sample, one treatment sample) individually marked with a three-digit code. Twenty-five milliliters of each sample was served at 10°C in clear ISO/INAO wineglasses and covered with plastic petri dishes. Wines were maintained at 10°C in a water bath (VWR1155 Refrigerated Constant Temperature Circulator, PolyScience, Niles, IL). Panelists were asked to examine, sniff, and/or taste each sample and then determine which of the two coded samples differed from the reference sample. Evaluations were recorded on laptop computers and analyzed using Compusense five release 4.6 software (Guelph, Ontario).
Trained panel.
Twelve WSU students and staff (11 females and 1 male, ages 23–70) participated in the trained panel. Panelists were recruited through a campuswide email and online WSU announcements and received a nonmonetary incentive at the end of each session. All sessions were conducted in the WSU Sensory Facility.
Panelists were trained over nine one-hour sessions to recognize specific taste and aroma attributes as well as intensities of each attribute. In the first session, panelists were trained using basic taste and aroma standards (Table 1⇓) prepared in a commercial base wine (Franzia Refreshing White Wine, Ripon, CA). These attributes were found to represent taste, aroma, and flavor attributes in Chardonnay and Gewürztraminer wines by a small focus group of experienced wine tasters. Panelists were also instructed on the use of the 15-cm unstructured line scale (with high and low anchors). Aroma standards were individually evaluated by sniffing each standard, after which different attribute intensities were discussed as a group.
In subsequent sessions, panelists reviewed standards and evaluated the experimental wines. Panelists also practiced sample evaluations in the sensory booths and were oriented with the sensory data collection system (Compusense five release 4.6). Following each training session, data were collected from each panelist to monitor individual panelist performance and reproducibility as well as overall group performance. Results were discussed during the next training session and were used as guidelines for future training sessions.
Bottled wines were held for seven months at 4°C before evaluation by the trained panel. After training, panelists participated in four days of evaluations (two days per varietal) in individual testing booths equipped with white lighting. A complete balanced block experimental design was used and each panelist evaluated each fining treatment twice. Taste and aroma standards used during training were available to the panelists before evaluations. Wine bottles were held at ~18°C in a water bath (PolyScience) and were poured immediately before serving. Twenty-five mL sample aliquots were presented in ISO/INAO clear wineglasses and covered with plastic petri dishes. Samples were labeled with three-digit codes and were served in random order. Panelists were asked to evaluate each sample for attribute intensity using a 15-cm unstructured line scale. Accompanying each sample was a reference sample (unfined control treatment) and its attribute intensities, which had been determined by the panel during training.
Statistical analysis.
A three-way fixed-effects ANOVA was performed on chemical and sensory data using XLStat (Addinsoft, Paris, France). For the chemical data, the main effects examined were fining treatment and replication, while for the sensory data, the main effects were fining treatment, replication, and panelist. Interactions were also evaluated. Differences between treatment means were compared using Fisher’s least significant difference. Compusense five release 4.6 was used to collect all sensory data and analyze data from the duo-trio test. Significance was defined as p ≤ 0.05.
Results and Discussion
Chemical analysis.
Turbidity.
The initial turbidities of unfined Chardonnay and Gewürztraminer wines were 27.8 and 211 NTU, respectively. Turbidity levels were again assessed after fining and before filtration and bottling (Table 2⇓). For Chardonnay, the control was the only treatment to exceed 10 NTU. Wines fined with whole milk (WM), wheat gluten (WG), Sparkalloid (SP), bentonite (BN), and isinglass (IS) were significantly lower in turbidity, with IS the lowest and activated charcoal (AC) the highest (p < 0.05). Wheat gluten and BN did not significantly differ in turbidity. Fining with WG (40 g/hL) has resulted in a more clarified Chardonnay than fining with BN, although neither clarified as well as IS or casein (Marchal et al. 2002). In the Gewürztraminer, only BN, IS, and SP achieved turbidity levels <10 NTU after seven days fining. The WM and WG were not significantly different than the control, indicating that these two agents were not as successful in clarifying as BN, IS, and SP.
Turbidity measurements were again taken after three months storage in the bottle (Table 2⇑). In the Chardonnay, the WG treatment was significantly higher than the other treatments, including the unfined control, which did not significantly differ from one another. Bentonite-treated Gewürztraminer had the lowest turbidity level (p ≤ 0.05). Whole milk had the greatest turbidity and was significantly higher than the control, suggesting that it induced haze. For either varietal, all treatments had turbidity levels <10 NTU after storage; however, more differences were observed between the Gewürztraminer treatments than the Chardonnay treatments.
At bottling, the Chardonnay control dropped from its initial turbidity level of 211 to 10.9 NTU, suggesting that the majority of haze-forming material naturally settled from the wine. Similarities in turbidity observed among AC, BN, IS, WM, SP, and control treatments after bottle storage were most likely attributed to natural settling rather than fining. Wheat gluten, however, caused an increase in haze in the Chardonnay. On the contrary, the Gewürztraminer wine had more haze at bottling, which resulted in greater differences in fining agent performance.
Protein.
Chardonnay fined with BN had the lowest concentration of protein (Table 2⇑). In addition, the application of SP or AC significantly reduced protein concentration when compared with the control (p < 0.05). The IS, WM, and WG treatments were greater than the control. These three fining agents are protein-based and thus their presence may have contributed to the increase in protein content.
In the Gewürztraminer, the AC, IS, and control treatments were highest in protein (Table 2⇑). Bentonite had the lowest protein concentration, a 57.71 mg/L reduction from the control (p < 0.05). This finding agrees with a report that BN (10 to 50 g/hL) was most successful at removing protein from Champagne (Dambrouck et al. 2005). Both WM and WG treatments were low in protein, but neither one was as low as BN. While the ability of IS to reduce protein levels in Gewürztraminer wine has not been extensively studied, it appeared to be ineffective, as did AC. Additionally, higher protein levels in AC-treated wine were not unexpected as carbon does not target proteins.
Volatile analysis.
Fourteen different volatile compounds were monitored in Chardonnay (Table 3⇓), with several other compounds detected but not quantified because of weak chromatographic signals. The analytes were selected based on preliminary volatile analysis of Chardonnay as well as those shown to be present in Chardonnay wine (Howard et al. 2005, Lee and Noble 2003, Whiton and Zoecklein 2000, Wondra and Berovič 2001). Ethyl acetate, 2-methyl-1-propanol, 3-methyl-1-butanol, and benzeneethanol had the highest concentrations (Table 3⇓). In a study of California Chardonnay, ethyl acetate, 2-methyl-1-propanol, and benzeneethanol were shown to be characteristic volatile compounds (Lee and Noble 2003). The only compound to significantly differ among treatments was ethyl dodecanoate, which was highest in the control and lowest in Chardonnay treated with IS, AC, WG, and BN. Stable linkages between bentonite and aroma compounds of must and wine (hexanol, ethyl hexanoate, and isoamyl acetate) result in a loss in compound concentration as they settle from solution with bentonite (Violley et al. 1990, Moio et al. 2004). Ethyl dodecanoate is a high molecular weight ethyl ester, formed during alcoholic fermentation, and is responsible for a leafy or soapy odor detected in wines (Acree and Arn 2004). For the most part, the volatile profile of the Chardonnay wine appeared unaffected by the fining agents applied.
In the Gewürztraminer wine, 19 volatile compounds were selected based on their significance to the varietal character of Gewürztraminer (Table 4⇓). These compounds included linalool, nerol, and l-α-terpinol, terpenes responsible for the floral notes common to Gewürztraminer wines (Flores et al. 1991, Guth 1997a, Reynolds and Wardle 1989).
Several differences were observed in the volatile composition of Gewürztraminer wine (Table 4⇑). All higher alcohols differed significantly among treatments. Overall, WG reduced the concentration of higher alcohols more so than the other fining agents. The control, AC, and BN treatments had the highest concentration of 2-methyl-1-propanol, whereas the IS, SP, WM, and WG treatments had the lowest. Benzeneethanol is known to enhance the roasted, toasty aroma in wine (Peinado et al. 2004). It was greatly affected by the use of SP, WG, and most significantly BN. The IS treatment had the highest concentration. The WG treatment had the lowest concentration of 1-hexanol (p < 0.05).
The AC treatment led to significantly higher 3-methyl-1-butanol and 2-methyl-1-butanol than any other treatment. However, a study with unfined sherry wine and sherry fined with activated charcoal found no significant differences in aroma profile (Lopéz et al. 2001). The different results found in the two studies could be explained by the concentrations of activated charcoal applied. The study with sherry used 180 mg/L AC, whereas our study used 450 mg/L AC. The adsorption capacity of activated charcoal depends on its dosage, and a higher dose could have a greater impact on the aroma compounds than a lower dose (Lopéz et al. 2001).
There were significant variations in ester concentrations among treatments. Ethyl acetate, which contributes to the sweet, fruity aroma of Gewürztraminer wines (Guth 1997a, 1997b), was significantly higher in the AC and control treatments when compared with the IS, SP, WM, and WG treatments and was lowest in the SP treatment. 3-Methyl-1-butanol acetate and 2-methyl-1-butanol acetate were significantly decreased by wheat gluten. 2-Phenylethyl acetate, which adds floral, fruity, and honey notes to wine, was significantly reduced by BN. The highest concentration was in the SP treatment, which was higher than the control. The BN treatment had the lowest amount of ethyl decanoate, which contributes fruity and grape aromas to wine. The IS and BN treatments produced the lowest concentrations of ethyl dodecanoate. Both compounds were high in the SP treatment, and for ethyl dodecanoate, there was no significant difference between SP and the control.
Overall, wheat gluten reduced the concentration of the lower aliphatic esters (e.g., ethyl butanoate and ethyl hexanoate). The higher aliphatic ethyl esters (ethyl decanoate and ethyl dodecanoate) were most affected by bentonite, which led to the lowest concentrations of these compounds.
There were slight differences in the three terpene compounds. Linalool is a compound characteristic to Gewürztraminer (Amerine and Roessler 1976), and the lowest concentration was observed in the BN treated wine. This finding agrees with a study in which bentonite (60 g/hL) significantly reduced the total concentration of monoterpenes and C13-norisoprenoids (13%), specifically linalool, geraniol, β-pinene, and limonene in Albariño wine (Armada and Falqué 2007). In Falanghina wines fined with bentonite (80 g/hL), there were significant losses in linalool and geraniol (Moio et al. 2004).
Several factors affecting the interaction between fining agents and free or bound volatile compounds have been suggested (Voilley et al. 1990, Moio et al. 2004, Armada and Falqué 2007), including the chemical nature of the volatile compound (i.e., polarity, functional groups, structure) and physical properties of the fining agent (i.e., isoelectric point). For example, fining agents can directly adsorb aroma compounds. Strong linkages between bentonite and ethyl esters have been shown to reduce the concentration of these aroma compounds during bentonite fining (Voilley et al. 1990).
Additionally, volatile concentrations could indirectly decrease as a result of interactions between volatile compounds and macromolecules in the wine, the latter eventually adsorbed by the fining agent. Consequently, some or all of the volatile compound is eliminated together with the fining agent. For instance, interactions were found between aldehydes and amino acids in an aqueous medium (Voilley et al. 1990). Should bentonite be added to the medium, charge-charge interactions between the negatively charged bentonite and positively charged proteins would form and precipitate out of suspension. As a result, aldehydes (i.e., hexanal and hexenals, responsible for grassy flavors at low concentrations) are removed from the wine (Clarke and Bakker 2004). Moreover, yeast cells have been shown to retain ethyl esters in their cell walls (Voilley et al. 1990). Settling aids, such as Sparkalloid, are used to facilitate the settling of yeast and other macromolecules from wine. Together with the yeast, ethyl esters are removed from the wine, indirectly affecting wine aroma and flavor.
Furthermore, fining agents derived from animals (such as whole milk) may contain a certain amount of fat. In this situation, volatile compounds may be absorbed and removed from the wine with the fining agent. Conversely, the fat content reduces the adsorption capacity of the protein, preventing it from adsorbing compounds that may contribute to aroma and flavor (Ribéreau-Gayon et al. 2006).
Sensory analysis.
Difference test.
Using the duo-trio test, 20 of the 30 (67%) panelists were needed to distinguish between treatments to achieve significance (p ≤ 0.05) (Meilgaard et al. 1999). The untrained panel was unable to discriminate between unfined wines and wines fined using any of the fining agents (p > 0.05) (Table 5⇓). However, 63% of panelists were able to distinguish between the control and wines fined with BN, WM, or WG, hinting at treatment differences. Similar results were found in Gewürztraminer wine. While no significant differences were observed (p > 0.05), a high percentage of panelists (60%) were able to distinguish between the control and the Gewürztraminer fined with IS.
The temperature at which the wines were served during the panel may explain the lack of difference between treatments in both varietals. Ross and Weller (2008) found that white wines served at 18°C had higher aroma intensities than those served at 10°C and 12°C, suggesting that white wines should be served warmer than 4 to 10°C to achieve maximum aroma profile.
Trained panel.
In the Chardonnay wines, of the 13 attributes evaluated by a trained panel, only spicy aroma and floral/honey flavor significantly differed (Table 6⇓). Isinglass significantly lowered both spicy aroma and floral/honey flavor, the lowest mean intensities for either attribute. Wheat gluten produced the highest concentration of spicy aroma, whereas whole milk produced the highest intensity of floral/honey flavor. The results for the floral/honey flavor should be interpreted cautiously as a significant panelist effect was observed (Table 7⇓), indicating that individual panelists varied in how they rated that particular attribute. In fact, significant panelist effects were observed for many of the attributes in both varieties, perhaps due to differences in panelist sensitivity to certain attributes or to insufficient training with standards for particular attributes (those that had a panelist effect). The significant effects could also have resulted from variability in use of the scale between panelists or individual anatomical and physical differences; these factors are difficult to eliminate, even through training. The short period of training could also be a factor, as the panelists only received nine hours of training on the attributes evaluated. These findings are similar to a study that found no noticeable differences in the appearance, aroma, and flavor of Chardonnay wine fined with various concentrations of collagen, skim milk, and isinglass (Meunier 2003).
The odor thresholds of several important wine volatile compounds have been published (Guth 1997a, 1997b, Zea et al. 2001, Peinado et al. 2004). While odor activities were not investigated in this study, published values were used to compare volatile and sensory data. For example, no significant differences were observed among the lower aliphatic ethyl esters (i.e., ethyl butanoate, ethyl hexanoate, and ethyl octanoate) in Chardonnay, although several esters exceeded their odor threshold values and theoretically produced aromas capable of being detected by the panelists (Table 8⇓). Esters generally contribute to the fruity aroma of wine. Therefore, the volatile results concur with the sensory results, as no differences were found in fruity aroma.
No sensory differences were found in the Gewürztraminer wine (Table 9⇓), which was unexpected, considering the number of differences observed between the volatile compositions of wines treated with different fining agents. In addition, only three attributes did not have a significant panelist effect, suggesting that panelist variability (i.e., use of scale, differences in sensitivity to certain attributes, anatomical differences) may have prevented treatment differences from being observed. The degree of difference between compound concentrations could also explain the lack of sensory differences found among treatments. For example, ethyl acetate is responsible for sweet and fruity notes and had a high concentration in Gewürztraminer, suggesting that the compound contributed to its fruity characteristic found by the panelist. However, differences among treatments did not differ more than 5 mg/L, which may have been too small to be detected by the panel. Additionally, 3-methyl-1-butanol acetate, known to contribute banana, fruity, and sweet characteristics to wine (Acree and Arn 2004), has an odor threshold of ~1.5 mg/L (Peinado et al. 2004), which is lower than that found in the present study. 3-Methyl-1-butanol acetate has been identified as one of the highest odor active compounds in Gewürztraminer (Guth 1997a). Nevertheless, the panel did not detect differences in fruity aroma/flavor and sweet taste, suggesting that chemical differences were not significant enough to induce perceivable differences. In another study, Gewürztraminer fined with bentonite (30 g/hL) was significantly lower in cooked vegetative aroma and higher in chemical aroma than an unfined control when evaluated by a trained panel (Flores et al. 1991). In fact, other than chemical aroma, all attributes were scored higher in the control wine than in the fined wine. Panel training was similar between studies, as were the number of panelists. Conversely, significant panelist-by-replication interactions were observed in both studies, making it difficult to generalize the sensory implications of bentonite fining on Gewürztraminer wine.
Isinglass is said to enhance fruity aromas in wines. In Gewürztraminer, this was demonstrated in that the highest fruit aroma and flavor intensities were observed in the isinglass treatment. The opposite was observed in the Chardonnay, where isinglass had the lowest fruit aroma and flavor intensities. These results suggest that isinglass will behave differently depending on the wine variety and that winemakers should consider fining trials prior to fining to establish the sensory impact of isinglass on the fruity notes of specific wines.
In Gewürztraminer, nerol and l-α-terpinol concentrations were substantially lower than their odor thresholds (0.5 and 1 mg/L, respectively) (Acree and Arn 2004), and their contribution to the floral aroma and/or flavor was most likely minimal. However, the odor threshold for linalool is ~0.015 mg/L (Peinado et al. 2004), similar to the concentrations found in the Gewürztraminer, and may have been responsible for the high floral intensities observed by the panel. Surprisingly, the panel was unable to distinguish among fining treatments, especially considering the low concentration of linalool in the bentonite treatment, indicating that the chemical differences among treatments were not significant enough to generate sensory differences.
Conclusions
The chemical differences found among fining treatments in the Chardonnay and Gewürztraminer wines support the hypothesis of this study. Activated charcoal, wheat gluten, and whole milk clarified Chardonnay but failed to clarify Gewürztraminer as effectively as the other agents, whereas bentonite, isinglass, and Sparkalloid achieved wine clarity in both varietals. Bentonite was most effective at reducing protein in both varietals. Fining had more impact on the volatile profile of Gewürztraminer wine than Chardonnay wine, and few differences in sensory properties were observed in either wine. Bentonite significantly reduced linalool levels in Gewürztraminer, and while sensory differences were not significant, winemakers should take precautions when fining Gewürztraminer with bentonite. Differences in spicy aroma and floral/honey flavor of Chardonnay fined with wheat gluten and isinglass, respectively, should be interpreted with caution, as they were accompanied by significant panelist interactions. Based on the differences observed between varietals, bench trials are strongly encouraged when determining the type and concentration of fining agent. Turbidity, protein content, and sensory impact should be assessed throughout fining trials to ensure appropriate agent dosage.
Footnotes
Acknowledgments: The authors acknowledge the Washington Wine Commission for their financial support, and Columbia Crest Winery for their generous wine donation.
- Received April 2009.
- Revision received September 2009.
- Accepted October 2009.
- Published online March 2010
- Copyright © 2010 by the American Society for Enology and Viticulture