Abstract
Frontenac (Vitis spp. MN 1047) is a recently introduced, cold-hardy red winegrape that is currently the most-planted cultivar in much of the Upper Midwest. Through descriptive analysis, a set of aroma attributes common to red Frontenac table wines has been described, but the volatile compounds responsible for the characteristic sensory notes of the product have not been investigated. In order to identify these odor active compounds, eight Frontenac table wines were evaluated using stir bar sorptive extraction (SBSE) combined with concurrent gas chromatography/olfactometry-mass spectrometry (GC/O-MS). Eight panelists evaluated GC/O effluent using qualitative detection frequency analysis. Twenty-four volatiles were identified in odor regions perceived by panelists, including five alcohols, 14 esters, one lactone, two acids, and two volatile phenols. Twenty-three of these were confirmed by linear retention index data in separate GC-MS analyses, and 23 were quantified in runs using a known concentration of internal standard. Similar analyses of wines produced from V. riparia clone #89, a parent of Frontenac, found 16 volatiles common to Frontenac wines. A brief study of Frontenac juice with two days of skin contact suggested that four volatiles found in the wine may originate in the fruit.
- Frontenac
- cold-hardy winegrape
- cold-climate winemaking
- gas chromatography/olfactometry-mass spectroscopy
In 1996, the University of Minnesota breeding program released Frontenac (Vitis spp. MN 1047), a red winegrape arising from a cross of the direct producer cultivar Landot 4511 (Landal L.244 X Villard blanc) and Vitis riparia clone #89, found growing wild near Jordan, MN (Luby et al. 2006). Because of its extreme cold-hardiness and suitability for wine production, Frontenac is used to produce rosés, port-style dessert wines, and red table wines in the Upper Midwest and other cold-climate growing areas. Frontenac is currently the most-planted grape cultivar in Minnesota, accounting for 20% of the total vineyard plantings in 2007 (Tuck and Gartner 2007).
As an interspecific hybrid and F1 progeny of V. riparia, Frontenac produces wines with unique sensory characteristics. While key volatile compounds in fruit from one V. riparia cultivar have been identified (Schreier and Paroschy 1980), the vast number and variation among V. riparia cultivars (Pierquet 1977) make any correlations tenuous. Descriptive analysis has established a set of 13 aroma attributes common to red Frontenac table wines (Mansfield and Vickers 2009), but little work has been done to characterize the volatile compounds responsible for these characteristics.
From a quantitative standpoint, fermentation-derived volatiles such as fusel alcohols, fatty acids, and esters constitute the largest part of wine aroma (Montedoro and Bertuccioli 1986, Scharpf et al. 1986). As a means of expressing varietal character or cultivar typicity, however, this group of odor-active compounds has little impact, as these components are similar to all fermented beverages and contribute little to differentiate between wine types (Ebeler 2001). While representing a significantly smaller proportion of the overall pool, volatiles and volatile precursors derived from the grape are thought to produce distinctive differences among wines produced from different grape cultivars (Noble 1990, Nykänen, 1986). In a few cases, varietal character is defined by a single compound, or a small collection of them; the concentration of a handful of monoterpenes in aromatic grapes, for instance, has been found to correlate directly with flavor intensity in wine and can be used to distinguish between six aromatic winegrape cultivars (Strauss et al. 1986). For many cultivars, however, varietal character is more elusive, dictated by larger groups of compounds occurring in ratios unique to each grape. Delineating these interactions is more difficult than identifying one or a few distinctive compounds, and subsequently much less is known about the relationships between volatile compounds that define the so-called neutral cultivars. Further, processing parameters such as harvest date, skin contact time, yeast strain, secondary fermentation, and aging may alter or mask grape-derived components (Ebeler 2001). Thus, defining varietal character for most winegrape cultivars is difficult.
As volatile constituents account for only about 0.1% of the total wine matrix, and individual compounds are often present at concentrations lower than 1 μg/L (Rapp 1998), identification of perceived impact odorants in wine requires use of the most sensitive analytical methods. In addition, preliminary extraction and concentration of volatiles may be required. Stir bar sorptive extraction (SBSE), a recent development in sorbent methodology, uses a magnetic stir bar coated with polydimethylsiloxane (PDMS) to adsorb and later release volatile compounds from complex matrices for analysis. SBSE has been successfully coupled with gas chromatography-mass spectrometry (GC-MS) to analyze various volatile compounds in wine (Alves et al. 2005, Diez et al. 2004, Hayasaka et al. 2002, Sandra et al. 2001, Salinas et al. 2004, Zalacain et al. 2004) and has proven especially useful in the analysis and identification of trace wine compounds with low sensory thresholds, such as C13-norisoprenoids and Maillard reaction products (Alves et al. 2005). For this work, volatile extraction with SBSE has been coupled with concurrent GC-MS and gas chromatography/olfactometry (GC/O) to identify perceived odorants in red Frontenac table wines. This combined approach allows easy comparison of perceived aromas with volatile compounds, reducing the time necessary for peak identification and comparison and decreasing the risk of improper compound/odor correlation by eliminating the need for separate GC-MS and GC/O runs.
Materials and Methods
Wines
To determine volatile compounds common to Frontenac wines, a selection of eight wines, from various years and producers, was obtained from commercial vineyards and from the University of Minnesota (UM) Enology Project’s Research Winery (Table 1). Since the object of this study was to determine volatile compounds and sensory descriptors that are common to wines produced from regionally-grown Frontenac grapes, an effort was made to acquire wines grown and produced by different wineries and using different production methods, rather than controlling processing variables.
In addition to the Frontenac table wines, two wines produced from V. riparia clone #89 (vintages 2004 and 2005), and one Frontenac juice sample (vintage 2006) were sourced from the UM Research Winery. To produce the V. riparia wines, grapes were harvested, crushed, and destemmed and the must inoculated with Pasteur Red yeast (Red Star, Cedar Rapids, IA) and fermented on the skins for five days at ambient temperature before pressing. Wine production then proceeded as described elsewhere (Luby et al. 2006). For the Frontenac juice sample, grapes were crushed and destemmed, then underwent cold soak on the skins for 48 hr at 2°C prior to pressing and sampling.
Commercial wines were purchased in 750-mL glass bottles sealed with natural cork closures. Research V. riparia wines were procured in 375-mL glass bottles sealed with natural cork closures. Before analysis, wines were divided into 20-mL aliquots, placed into clean, nitrogen-sparged 20-mL screwcap vials, and held at −20°C until needed, ~1 to 2 months. Frontenac juice was similarly divided into clean, nitrogen-sparged 20-mL screwcap vials directly after pressing and stored under the same conditions.
Extraction method
Wine or juice samples were removed from cold storage and allowed to rest in sealed storage vials until reaching ambient temperature, ~21°C. A 10-mm magnetic stir bar coated with polydimethylsiloxane (PDMS) (Gerstel, Baltimore, MD), termed a Twister, was added to each sample vial. The vial was capped to exclude additional air and placed on a magnetic stir plate. The sample was stirred at 200 rpm for 90 min, allowing volatiles to reach equilibrium with the Twister absorbent, as determined by preliminary analysis. The Twister was then rinsed with distilled water to remove any juice or wine, patted dry with lint-free tissue, and used for volatile analysis.
Gas chromatography/olfactometry-mass spectrometry
GC/O-MS analysis was performed on an HP 5890 Series II gas chromatograph (Hewlett-Packard, Palo Alto, CA) modified to allow concurrent olfactometric and mass spectrometric analysis on an HP 5970 mass spectrometer (Hewlett-Packard). A Twister, loaded with wine volatiles as described above, was inserted into a split/splitless GC liner plugged with glass wool and sealed with a flip-top injection port lid (model 5188-2717; Agilent, Santa Clara, CA). The Twister was desorbed at 250°C for 5 min with cryofocusing via a loop of column submerged in a Dewar of liquid nitrogen. Volatiles were thus loaded into a DB5-MS column, 30m x 0.25 mm x 25 mm (J&W, Folsom, CA) with a 30:1 split flow. The oven temperature started at 50°C, was held for 2 min, ramped at 10°C/min to 250°C, and held for 5 min.
Via a splitter at the end of the column, equal portions of the GC effluent were routed to an MS detector and a sniff port, held at 270°C and 250°C, respectively. The mass spectrometer was operated at 70 eV in scan mode, scanning the range m/z 29 to 300 at 2.9 scans/sec. In the sniff port, the GC column effluent was combined with humidified air to increase olfactory perception by reducing nasal passage dryness and fatigue. Panelists were seated in front of the sniff port while evaluating and were asked to write descriptors of aromas perceived while simultaneously pressing a button to record the presence of a perceived odor. All evaluations were performed during May and June 2007.
GCxGC-time of flight mass spectrometry
In an attempt to identify coeluting peaks occurring from linear retention index (LRI) 1101 to 1118 on the GC-MS chromatogram, GCxGC-TOF was performed on wine sample D, which showed the greatest peak separation in that range. Analysis was performed on an Agilent 6890 gas chromatograph using the LECO GCxGC-TOF option. A Twister, loaded with wine volatiles as described above, was inserted into a split/splitless GC liner plugged with glass wool. Volatiles were then loaded onto an Rtx-5 column (9.5 m x 0.20 mm x 180 μm) (Restek, Bellefonte, PA) with a gas flow of 1 mL/min. The first oven temperature started at 50°C, was held for 2 min, and ramped at 10°C/min to 270°C and held 8 min. In the secondary oven, a DB-17 column (0.7 m x 0.10 mm x 100 μm) (Agilent) was started at 65°C, held for 2 min, and ramped at 10°C/min to 285°C and held 8 min. A second DB-17 column (0.2 m x 0.10 mm x 100 μm) (Agilent) was used in the detector oven. Temperature between the first and secondary oven was modulated in 3-sec intervals, with a hot pulse time of 0.4 sec and a 1.1-sec cool time. The mass spectrometer was operated at 70 eV in scan mode, scanning the range m/z 29 to 300 at 200 spectra/sec in total ion mode.
Statistical determination of significance
To achieve results significant at a P′ of 0.90, optimal sample size was calculated using,
where p1 = 0.75, and p2 = 0.5. A sample size of 64, or eight olfactory panelists evaluating each eight wines in a randomized complete block, resulted in a p value between 0.05 and 0.01. Panelists did not perform wine analyses in duplicate, as the discontinuous nature of human respiration makes true replication of GC/O analysis improbable, at best (Hanaoka et al. 2001). Each sniff run was considered an experimental unit, and a positive or negative binary response (i.e., each compound was perceived or not perceived at a certain time) was returned for each compound of interest.
Panelist selection
Eight panelists, five females and three males, between 22 and 30 years of age, were selected from students and staff of the UM Flavor Laboratory. Panelists were selected based on availability, previous experience with GC/O analysis, and lack of known anosmia.
Identification of odor-active compounds
Compounds perceived in at least 50% of the GC/O evaluations were identified when possible. Tentative identifications were made through comparison of MS fragmentation patterns with the Wiley 275 Mass Spectra Library. To verify identification, Frontenac wine samples A and G were spiked with a hydrocarbon ladder (C5 to C16 in pentane) at 1 mg/L, and extracted and analyzed via GC-MS in the same manner as the wine samples above. LRIs were calculated and compared with previously reported Kovats index (KI) values (Kovats 1965). As reported above, the compounds involved in coeluting peaks occurring from LRI 1101 to 1118 on the GC-MS chromatogram were tentatively identified via GCxGC-TOF through examination of the area between the closest identified peaks.
Quantification of odor-active compounds
Concentrations of selected volatile compounds were estimated based on GC peak area comparisons with the area of an internal standard (2-heptanone). A solution of 2-heptanone in ethanol was added to each of the eight sample wines at a concentration of 6.5 mg/L, and the samples were analyzed via GC-MS as described above.
Results
Eighteen odor regions, that is, those perceived by panelists in more than 50% of the runs, were identified in all Frontenac wines (Table 2). In 16 of these regions, 24 volatile compounds were tentatively identified by comparing MS-derived spectral data with spectra libraries. These included five alcohols, 14 esters, one lactone, two acids, and two phenols (Table 3). LRI comparisons confirmed the identity of 23 of the 24 volatile compounds tentatively identified via MS. Estimated volatile concentrations, based on the internal standard, were calculated (Table 4).
Nine odor regions contained single aroma compounds. An additional three regions were comprised of structurally similar pairs of closely eluting peaks: 2- and 3-methyl-1-butanol (LRI 746 to 754), ethyl-2- and ethyl-3-methylbutyrate (LRI 849 to 856), and 2- and 3-methylbutyl acetate (LRI 878 to 885). Four additional regions encompassed nonrelated but closely eluting compounds pairs with differing aroma characteristics. Aromas described in two odor regions could not be assigned to volatile compounds. In the region between LRI 1080 and 1089, no peaks were isolated. Two odor regions, those occurring between LRI 1101 and 1107 and between 1107 and 1118, were found to be one large region, consisting of three to four large, overlapping peaks, which made separation and identification of individual compounds difficult. Linalool was identified in this region in wines D and H and was thought to be responsible for the strong floral aroma perceived by panelists in the middle of the region. Analysis by GCxGC-TOF resulted in the tentative identification of 2,3-octen-1-ol, nonanol, and 2-methoxyphenol as compounds likely responsible for the sequence of earthy/mushroomy, green, and distinct smoky, burnt wood aromas perceived by panelists.
Fifteen volatiles identified in Frontenac wines were also found during GC/O-MS analysis in wines produced from two vintages of V. riparia #89 (Table 4); compound identification was confirmed with LRI calculations. Four volatile compounds identified in Frontenac wines were tentatively identified in the Frontenac juice sample via GC-MS screening (Table 4). Nonanol, mentioned in the GCxGC-TOF analysis above, was also tentatively identified in the juice sample.
Discussion
The primary objective of this work was to identify the chemical compounds responsible for the aroma of red Frontenac table wines. This goal is complex, as the aroma profile in the finished wine is dictated not only by the quantity and concentration of individual volatiles but also by the synergistic relationships that exist between them and between volatiles and nonvolatile compounds in the wine matrix (Ferreira et al. 2000). This analysis was designed to identify those compounds that can be perceived sensorially and subsequently did not attempt to monitor the glycoconjugates, S-cysteineconjugates, and S-methylmethionine and other compounds that exist as nonvolatile aroma precursors in many grapes (Baumes 2009). Regardless, determining which volatile compounds panelists can perceive is a first step in developing a clear picture of key impact odorants in Frontenac.
Identifying volatile compounds in odor-active regions
In practical terms, the compounds of interest to Frontenac varietal character are those that can be perceived by most wine drinkers. In this study, it was assumed that volatiles that could be perceived by at least 50% of the GC/O panelists would also be perceived by at least 50% of wine drinkers. As explained above, panel size was calculated to ensure that these results were significant at a P′ of 0.90.
One difficulty was in determining how to tabulate the number of responses for a given odor; slight variability in compound elution time and panelist response meant that the recorded time of individual odor perception was not perfectly aligned. To compensate for these differences, response times for each wine were normalized to the odor region around isoamyl acetate, which eluted consistently at ~6.60 min and was the compound most uniformly described by panelists as some variant of banana. This realignment revealed that panelists were signaling perception of what appeared to be the same odor at times that varied as much as 0.15 min, while examination of chromatograms showed that peak elution time varied by as much as 0.10 min. Subsequently, it was decided that percent odor perception would be tabulated based on a span of ~0.20 min, depending on whether compound descriptors could be grouped in one, or at most two, distinctive general character families such as fruity or herbaceous. In cases where significant odor regions were later determined to encompass two or three aroma compounds, all were retained as significant.
Ultimately, 23 volatile compounds could be positively identified in 17 odor regions. Of these, six potentially originated from the grape, surviving fermentation to exist in the finished wine. An additional eight compounds may be derived from grape-synthesized precursors, the final concentrations of which may be dictated, in part, by cultivar.
Grape-derived volatiles
Assuming that grape-derived volatiles are of greatest importance to varietal character, one part of defining typicity is identifying odor-active compounds that originate in the fruit and survive fermentation intact. A comparison of odor-active volatiles identified in Frontenac wines and those tentatively identified in Frontenac juice indicates that four alcohols, 2- and 3-methyl-1-butanol, 1-hexanol, and phenethyl alcohol, may originate in the grape (Table 4). Fusel alcohols 2- and 3-methyl-1-butanol have been identified in the wines of several different V. vinifera cultivars, where they are thought to arise largely through fermentation mechanisms (Schreier 1979), but have also been identified in grape berries (Schreier et al. 1976). Both compounds have been shown to develop in fruit through amino acid conversions (Drawert 1975), although this mechanism has not been identified specifically in grape berries.
Phenethyl alcohol is commonly found in wines and can be formed in grapes from the precursor phenylethyl-α-d-glucopyranoside, but is usually only present in small quantities (Winterhalter et al. 1999, Garcia et al. 2003). The greater part of the phenylethyl alcohol is thought to form during yeast fermentation, where 2-phenylalanine serves as a precursor (Laminkanra et al. 1996). As the phenylethyl alcohol content was not quantified in the Frontenac juice, it is not known whether phenethyl alcohol exists in concentrations comparable to those found in the finished wine or whether the bulk of the final concentration was formed during fermentation.
The six-carbon alcohol, 1-hexanol, is produced during crushing, when linolenic and linoleic acids are released from the grape skin and enter reactions catalyzed by lipoxygenase, peroxidase, and alcohol dehydrogenase (Iglesias et al. 1991, Garcia et al. 2003). The concentration of 1-hexanol in the finished wine is thus dependent on must aeration (Rocha 2004) and the length of juice-skin contact. Both 1-hexanol and phenethyl alcohol are known to survive fermentation (Schreier 1979).
In a finished wine, 2-methyl-1-butanol, 3-methyl-1-butanol, and phenethyl alcohol contribute subtle complexity at levels <3000 mg/L but become penetrating and unpleasant at higher concentrations (Ebeler 2001). In the six commercial Frontenacs examined, the estimated combined concentrations of the closely spaced 2- and 3-methyl-1-butanol ranged from 4.8 to 8.4 mg/L and phenethyl alcohol ranged from 1.1 to 2.1 mg/L. While well below the concentration at which the wine sensory profile would be negatively impacted, it must be emphasized that these values are only approximations. Panelists described regions corresponding to these compounds as chemical, green, or oxidized and of moderate intensity. 1-Hexanol eluted very close to isoamyl acetate and is likely to be the origin of fainter notes of fresh nuts, vitamin, and grainy aromas perceived immediately before the strong banana note typical of the latter compound. At levels above the sensory threshold, 1-hexanol has been found to contribute green, herbaceous off-odors in wines (Kotseridis and Baumes 2000).
When the monoterpene alcohol linalool occurs in wine, it is generally thought to originate in the grape (Ebeler 2001). Though present in Frontenac wine, this alcohol was not identified in the Frontenac juice. The importance of this compound is difficult to determine in the wines studied. In the odor region between LRI 1101 and 1118, the coelution of three or four compounds made unambiguous identification of all volatiles impossible. In wines A, D, F, and H, however, peaks were slightly separated in this area, allowing identification of linalool. Linalool may contribute floral notes to a wine and may be responsible for the panelist descriptors of floral, jasmine, and rose perfume perceived in this odor region. Linalool has been identified in several V. vinifera grapes and juices, in higher concentrations in aromatic whites, and at lower levels in neutral varieties (Strauss et al. 1986). Concentrations of linalool in Frontenac could not be calculated because of the multitude of coeluting peaks.
The volatile phenol, methyl salicylate, which was positively identified in all eight Frontenac wines, may also originate in the grape. While this compound was not identified in the Frontenac juice, it was identified in the V. riparia #89 wine and has been identified in earlier studies of V. riparia grapes (Schreier and Paroschy 1980). Methyl salicylate is rarely seen as a component of wine, although it has been identified in the V. vinifera sp. Emir grape of Turkey (Cabaroglu et al. 1997) and Huxelrebe (Caven-Quantrill and Buglass 2006). The sensory impact of this compound was not determined for either cultivar. Methyl salicylate is described in the literature as having an odor of wintergreen, mint, or a fresh green character, which correlates well with the waxy, minty, and fresh leaf descriptors assigned by panelists. If present at perceivable levels, this compound may contribute the fresh green descriptor identified by a descriptive analysis of Frontenac table wines (Mansfield and Vickers 2009).
Volatiles evolving from grape-derived precursors
As in most wines, a majority of odor-active compounds in Frontenac were generated during fermentation. While not directly derived from the grape, many so-called fermentation products are produced from precursors found in the fruit, such that varietal differences in precursor concentration result in varietal variation in wine aroma profile. Of these precursor-dependent volatiles, the compounds produced from amino acid metabolism, namely the isoacids and their ethyl esters and the fusel alcohols and their acetate esters, have the greatest sensory impact (Ferreira et al. 2000). This suggests that the ratios of five odor active compounds of Frontenac—the isoacid esters (ethyl isobutyrate, ethyl-2-methylbutyrate, and ethyl-3-methylbutyrate) and the fusel alcohol esters (isoamyl acetate and 2-methylbutyl acetate)—may potentially be unique to the cultivar and could serve as characterizing volatile components.
Three other identified compounds are of uncertain origin. Eugenol is a volatile phenol often found in barrel-aged wines. It is formed in oak staves as a result of lignin breakdown during stave toasting and can be transferred into the wine with other oak products (Chatonnet et al. 1990). Glycosidically bound precursors of eugenol are also found in the grape, and varietal differences have been observed in the finished wine (Ferreira et al. 2000). Two related compounds, the cinnamic esters ethyl dihydrocinnamate and ethyl cinnamate, arise from various precursors found in the grape or in oak (Jackson 2000). In this work, Frontenac samples with and without oak aging were used, so the effect of grape precursors on final eugenol and cinnamic ester concentration is unknown. As all wines showed some amount of these compounds, however, it can be deduced that precursors exist in the Frontenac grape.
Volatiles evolving from fermentation
The two odor active acids detected, octanoic and decanoic acid, are fermentation derived and may be produced via anabolic or catabolic mechanisms (Nykänen 1986, Schreier et al. 1976). These compounds have been reported to contribute fruity, fatty, or rancid notes to wine (Gómez-Miguez et al. 2007), which agrees with panelist descriptors (Table 3). Octanoic and decanoic acids (Table 4) were estimated to be present in some Frontenac wines at concentrations above their reported sensory threshold levels of 10 mg/L and 6 mg/L, respectively (Swiegers and Pretorius 2005).
Fatty acid esters, including the ethyl butanoate, ethyl hexanoate, ethyl octanoate, and ethyl decanoate identified in Frontenac, are also yeast-produced, synthesized through a process very similar to the anabolic production of fatty acids. The relative concentration ranges of these compounds found in Frontenac generally agreed with those found in studies of both red and white wines, with concentrations in the following descending order: ethyl octanoate, ethyl hexanoate, and ethyl butanoate (Baumes et al. 1986). These apolar esters have been found, as a group, to be present in smaller concentrations than polar esters like ethyl acetate but to contribute more to overall wine aroma (Baumes et al. 1986). The fatty acid methyl ester, methyl hexanoate, likely developed during fermentation as well. Although this compound has been found in grapes (Schreier et al. 1976), it was not identified in V. riparia fruit (Schreier and Paroschy 1980) and was not found in Frontenac juice (Table 4). As a class, the aroma contribution of fatty acid esters is generally fruity and is largely considered unimportant as a tool for distinguishing between wines (Ebeler 2001).
The remaining volatiles identified in Frontenac contribute little toward distinguishing varietal characteristics. Both ethyl lactate and diethyl succinate are fermentation-derived esters common to wines and other fermented beverages. Both may be synthesized by yeast and are also produced by strains of lactic acid bacteria, so secondary malolactic fermentation can result in increased concentration in the final wine. γ-Butyrolactone is also commonly found in wines and is thought to arise from the metabolism of glutamic acid during fermentation (Rocha et al. 2004). While this compound has been identified in the skins of some grape cultivars (Garcia et al. 2003), it was not found in Frontenac juice. Concentrations of γ-butyrolactone have been found to be higher in red wines, likely as a result of extended skin-wine contact performed in most red wine fermentations, but its sensory impact in overall wine aroma is considered minimal (Baumes et al. 1986).
As this work represents only the initial steps in identifying volatile compounds important to Frontenac wine aroma, there are several issues that require further investigation. The odor region between LRI 1080 and 1089, for which no peaks could be isolated, may result from a compound present at concentrations below the instrument detection level. Greater sample extraction and concentration may be useful in identifying the compounds responsible for the caramel, fresh bread, and toasted notes described in this region. Additional work is needed to determine whether the volatiles identified are actually responsible for the odors panelists perceived. It would also be of interest to further explore the volatiles present in Frontenac juice. The sample used for this work received only two days of skin contact, less than the 5 to 8 days generally used in Frontenac wine production. Since several compounds discussed earlier may be extracted from fruit skin, it is possible that analysis of Frontenac juice with longer skin contact time would reveal additional compounds of interest.
Relationship to sensory analysis
In previous work, descriptive analysis established a list of 13 descriptors to describe aroma attributes common to red table wines produced from Frontenac (Mansfield and Vickers 2009). These attributes are classed in seven first-tier terms in the wine aroma wheel (Noble et al. 1987: fruity (blackberry, black currant, cherry, jammy), floral (geranium, floral), vegetative (cooked vegetable, fresh green), spicy (black pepper, spice), woody (cedar), earthy (earthy), and caramelized (tamari). While the relationship between specific wine compounds and aromas perceived in the final matrix cannot be delineated without reconstitution studies, some broad comparisons can be made. Esters are likely responsible for fruity and floral notes in wines, and of the 14 identified as perceivable by GCO, 10 were described as either floral or fruity. Linalool was also described as having a strong floral aroma and could impart this note to the final wine because of its very low sensory threshold. Several compounds, including ethyl-2-methylbutyrate, methyl salicylate, and decanoic acid, were described as earthy or green, and eugenol could potentially impart the spicy or cedar aromas noted by panelists. Rotundone, a sesquiterpene generally thought to impart black pepper aromas (Wood et al. 2008), was not identified in Frontenac, despite the panel’s use of black pepper as a descriptor. While it is impossible to predict the aroma attributes that individual volatile compounds will impart to a complex matrix like a red wine, this rough comparison suggests that the volatile compounds identified in Frontenac at least have the potential to produce the attributes described.
Conclusions
Twenty-four odor-active volatile compounds, defined as those perceived in at least 50% of the 64 GC/O analyses, were identified in red Frontenac table wine. All volatiles have been previously reported in other wine types. Six compounds of sensory importance may be derived directly from the grape and an additional six may develop from grape-derived precursors; these compounds are most likely to be responsible for the varietal character of Frontenac. All other volatiles identified were the product of yeast or bacterial metabolism, and as such are common to most wines and many other fermented beverages, contributing little to varietal character. Further work is needed to unambiguously identify volatiles important to Frontenac wine typicity.
Acknowledgments
Acknowledgments: The Minnesota Agricultural Experiment Station supported this research. Commercial wines were donated by Alexis Bailly Vineyard, Falconer Vineyards, Fieldstone Vineyards, Morgan Creek Vineyards, Northern Vineyards Winery, and Saint Croix Vineyards.
Footnotes
-
↵4 Assistant Professor, Department of Food Science & Technology, Food Research Laboratory, NYSAES, Cornell University, Geneva, NY 14456.
- Received June 1, 2010.
- Revision received December 1, 2010.
- Accepted January 1, 2011.
- Published online December 1969
- © 2011 by the American Society for Enology and Viticulture