Characterization of Odor-Active Compounds in Californian Chardonnay Wines Using GC-Olfactometry and...

Characterization of Odor-Active Compounds in CalifornianChardonnay Wines Using GC-Olfactometry and GC-Mass

Spectrometry

SEUNG-JOO LEE* ,† AND ANN C. NOBLE

Department of Viticulture and Enology, University of California, Davis, California 95616

Nineteen commercial Californian Chardonnay wines were analyzed by gas chromatography-massspectrometry (GC-MS). Freon extracts of wines were separated by silica gel chromatography intothree fractions. Volatiles were quantified by GC analysis of each fraction using internal standardsadded to the wine prior to Freon extraction. Twelve of the 19 wines were evaluated by GC-Olfactometry(GC-O). Of the 81 compounds shown to be odor-active (OA) by GC/O, 74 were quantified and 61were tentatively identified, all of which had been previously reported in grapes or wines. Overallconcentrations of compounds with floral or oak-related aromas were higher in wines shown bydescriptive analysis to be high in intensity of either floral or oak notes, respectively. The relationshipbetween sensory intensity ratings from a previous descriptive analysis of the wines and 74 OAcompounds was modeled by partial least-squares regression (PLS) analysis. This PLS model onlyexplained 17% of the variation in the OA variables, whereas a PLS using a subset of 16 OA peaksexplained 64 and 47% of variance in the sensory and GC data, respectively. Fruity wines high inpeach, citrus, and floral terms were separated from those high in oak-related sensory attributes (oak,vanilla, caramel, spice, and butter). In both PLS models, the fruity and floral terms were associatedwith isoamyl acetate, 2-phenylethyl acetate, linalool and two unknowns exhibiting minty and bandaid-caramel odors; the oaky attributes were associated with vanillin, oak-lactones, 4-ethyl guaiacol,γ-nonalactone, 2-acetyl furan, eugenol, 2-methoxy phenol, and two unknowns with plastic and smokyodors.

KEYWORDS: Wine; odor-active compounds; Chardonnay; gas-chromatography- olfactometry; partial

least-squares regression analysis

INTRODUCTION

In a recent consumer study, the flavor of wine was found tobe one of the most important attributes to consumers whenbuying wine (1). Over eight hundred volatiles have beenidentified in wine aroma including alcohols, esters, organicacids, aldehydes, ketones, and monoterpenes (2-5). Thesecompounds come from the grapes and are produced duringfermentation and post fermentation treatments such as oakstorage and bottle aging (2-4).

Volatiles of Chardonnay grapes and wines have been studiedusing a variety of techniques. Principal component analysis(PCA) of headspace volatiles of three white wine varieties(Riesling, Chardonnay, and French Colombard) clustered winesby grape varieties (6). Chardonnay wines were higher in estersand vitispirane, while the Rieslings were higher in terpenecomponents, with French Colombard wines falling in between.

In studies of seven Chardonnay wines from Burgundy,dichloromethane extracts were analyzed by GC-O (7) by one

judge using Charm analysis (8). Of the 32 odor-active com-pounds detected, the 11 compounds which had the highest“Charm values” were identified as contributing to the distinc-tiveness of the wines' aromas. They were vanillin (vanilla),diacetyl (butter), 4-vinylguaiacol (curry-smoked), ethyl cin-namate (cherry pits), ethyl hexanoate (green-grassy), ethyl2-methyl butanoate (apple), ethyl butanoate (fruity), guaiacol(smoked-spicey), and three unidentified compounds, which hadaromas of burnt sugar, wet ashes, and honey. In a preliminarystudy of white Burgundies, cyclotene, maltol, guaiacol (2-methoxy phenol), and ethyl cinnamate were claimed to be themost potent odorants (9).

Using a different approach to study Chardonnay aromas,volatiles that had been hydrolyzed from the glycosides ofChardonnay juice were analyzed by GC-MS (10). Compounds(n ) 181) were identified, of which more than 70% of themetabolites were C13 compounds, norisoprenoids. Benzenederivatives accounted for 20% of the total volatile concentration,while monoterpenes made up only 5%. The potential of theseChardonnay glycosides to serve as flavor precursors wasconfirmed, as the intensity of tea, floral, lime, honey, oak, talc,

* To whom correspondence should be addressed. E-mail: [email protected].: + 82-16-869-6026. Fax 82-31-709-9876).

† Current address: Korea Food Research Institute, Sungnam, Korea.

8036 J. Agric. Food Chem. 2003, 51, 8036−8044

10.1021/jf034747v CCC: $25.00 © 2003 American Chemical SocietyPublished on Web 12/03/2003

and pineapple aromas was increased by addition of their acidhydrolysates to base wines (11).

To investigate the relationships between sensory profiles andwine volatiles, multivariate statistical methods have been used,including principal component analysis of instrumental variables(PCA-IV) (12), generalized procrustes analysis (13), and partialleast squares regression (PLS) (14, 15), and have been compared(16). In previous studies of Chardonnay wines, including thosecited above, hundreds of volatiles have been identified. How-ever, with the exception of studies of the aroma of whiteBurgundy wines by Moio et al. (7) and Le Fur et al. (9,13),neither the odor activity of volatiles nor the sensory significanceof compounds identified in Chardonnay wines have beensystematically investigated. The objectives of the present studywere to identify and quantify odor-active compounds in Char-donnay wines using GC-O and GC-MS and relate the OAcompounds to the sensory properties of the wines using partialleast squares regression analysis.

MATERIALS AND METHODS

Wines.Nineteen 1997 Californian Chardonnay wines were analyzedin 2000, all of which had been profiled by descriptive analysis (DA)6-10 months before this study (17). Details about the wines, whichwere held at 10°C during the studies, are shown inTable 1.

Chemicals.Diethyl ether, pentane, and silica gel 60 (particle size0.063-0.200 mm, 70-230 mesh) were purchased from EM Science,a division of EM Industries, Inc (New Jersey). Trichlorofluoromethane(Freon 11), absolute ethanol, and compounds used as internal standards(IS) (methyl octanoate, 2-methyl-1-pentanol, and 3-methyl-3-hydroxy-2-butanone) were obtained from Sigma-Aldrich Chemical Co. (St.Louis, MO). The IS stock solution was prepared by adding 5µg ofeach internal standard to 100 mL of absolute ethanol.

Extraction and Fractionation of Wine Volatile Compounds.Volatiles were extracted using a modification of a procedure describedelsewhere (18). Before extraction, 45 g of NaCl and 3 mL of IS stocksolution were added to 150 mL of wine, which was then extracted threetimes with 50 mL of trichlorofluoromethane (Freon 11) using a liquid-liquid extractor at 28∼30 °C. The extract was concentrated to∼2 mLby distilling off the solvent on a Vigreux column (40× 2 cm). Thesolvent was further removed under a purified nitrogen stream until thevolume was reduced to 1 mL. The aroma extracts were fractionatedby silica gel chromatography to provide better GC resolution, using amodification of Guth’s method (19). The Freon extract (1 mL) wasplaced in a glass column (30× 1.9 cm i.d.) packed with silica gel 60.

The sample was fractionated by elution with 200 mL of pentane anddiethyl ether (Fraction 1, 85/15; Fraction 2, 70/30) and 200 mL ofdiethyl ether (Fraction 3). The eluates were dried over sodium sulfateovernight and concentrated to a final volume of 1 mL, as describedabove, and stored at-5 °C for subsequent analyses.

The recovery of internal standards after sample preparation (extrac-tion, fractionation, and GC analysis) was evaluated for five wines (JL,CDB-C, CAL, DEL and SH) to determine the recovery of the method.Recovery ranged from 82% for methyl octanoate in fraction 1 to 73%for 2-methyl-1-pentanol (fraction 2) and to 61% for 3-methyl-3-hydroxy-2-butanone in fraction 3. Reproducibility of the samplepreparation method was examined for one wine (SH), which wasextracted in duplicate, with each fraction’s extract analyzed in duplicateby GC. A two-way analysis of variance (extraction, injection) for eachpeak showed no significant differences due to extraction for all buttwo peaks and no significant differences due to injection for all buttwo peaks.

Gas Chromatography-Mass Spectrometry (GC-MS).A 1-µLsample of each concentrated fraction of the wines was analyzed induplicate on a Hewlett-Packard (HP) gas chromatograph model 6890equipped with a split/splitless injector and a DB-WAX bonded fusedcapillary column (30 m× 0.25 mm i.d., film thickness) 0.25 µm,J&W scientific Inc., Folsom, CA). The detector was a mass spectrometer(MS 6890 series Mass selective detector, Hewlett-Packard, Palo Alto,CA). Temperature of the inlet was 220°C. Splitless time was 1 min.Purge flow to split vent was 50 mL/min for 1 min. Column headpressure was 14.14 psi and the helium carrier gas flow rate was 1.3mL/min. Average helium gas velocity was 30 cm/s. The oventemperature was held at 40°C for 4 min and programmed at 4°C/minto 185°C and held for 20 min isothermally. Mass spectra in the electronimpact mode (MS-EI) were generated at 70 eV and ion sourcetemperature was 230°C. Mass spectra were taken over them/z range45-300. The total ion chromatogram (TIC) acquired by GC-MS wasused for peak area integration. HP MS chemstation software G1701BAver.B.01.00 was used for data acquisition.

To determine the reproducibility of the duplicate injections anddetermine which peaks varied across wines, two-way analyses ofvariance (wines, injection) were performed for each odor-active peak.All peaks varied highly significantly across wines, with only fourvarying significantly across replications.

GC-Olfactometry (GC-O). All analyses were performed at E & JGallo Winery, (Modesto, CA). For GC-O analysis, 1µL of theconcentrated fractions were injected on a HP GC model 6890 equippedwith a split/splitless injector. At the end of the capillary, the effluentwas split into the HP MS Mass selective detector described above anda sniffing port (Gerstel, Germany). The sniffing port was held at 250°C to prevent any condensation of volatile compounds. Humidified airwas added at 100 mL/min in the sniffing cone to reduce fatigue anddrying of the judge’s nasal passage. The column and operatingconditions were the same as those used for GC-MS.

For determination of odor-active (OA) compounds, four judges whohad previous experience with GC-O were used. Assessors were seatedin front of the sniffing port and asked to smell the effluent of thecolumn. An “olfactometer button” (Gerstel, Germany) was depressedwhen an aroma was detected. The initiation and termination of aromadetection was recorded by an HP Pascal workstation. Judges also gaveverbal descriptions of perceived odors that the experimenter recorded.Each fraction of the four wines (CAL, CDB-C, DEL, and JL) whichhad been shown to have the largest differences in aroma by sensorydescriptive analysis (17) was evaluated once by GC-O by each of thefour judges. For the remaining eight wines (identified inTable 1), twojudges evaluated each fraction once. Peaks were identified as odor-active using a modified definition of the detection frequency method(20). A peak was reported as OA, if it was detected by two or morejudges in the same wine.

Identification. Odor-active compounds screened by GC-O weretentatively identified by comparison of the Kovats retention index (KI)(21) and the MS fragmentation pattern with those of referencecompounds or with mass spectra in the Wiley 275 library and previouslyreported Kovats retention indices. The Kovats retention indices (KI)of unknown compounds were determined by injection of the sample

Table 1. Wines and Their Regions of Origin

code winery region of origin

CALa Callaway Riverside (Temecula)CDB-A Clos Du Bois SonomaCDB-Ca Clos Du Bois SonomaCDB-S Clos Du Bois SonomaCONa Concannon AlamedaDELa Delicato San JoaquinDUN Canandaigua Wine Co. Dunnewood, North CoastEBE Eberle Paso RoblesFET Fetzer MendocinoGP Geyser Peak SonomaJLa J. Lohr MontereyMERa Meridian Paso RoblesPR Pine Ridge Napa-CarnerosRHPa R. H. Phillips Esparto (Dunnigan Hills)SHa Sutter Home Napa ValleyTAFTa Taft Street Winery SonomaTESa Monterey Vineyards Monterey CountyVMEa Villa Mt. Eden Napa ValleyVMErsa Villa Mt. Eden (Grand Reserve) Napa Valley

a Wines analyzed by GC/O.

Odor-Active Compounds in Chardonnay Wines J. Agric. Food Chem., Vol. 51, No. 27, 2003 8037

with a homologous series of alkanes (C6-C28). GC/MS conditions werethe same as described above.

Quantification. The relative concentrations of the odor-activevolatiles in all 19 wines were determined by GC-MS (TIC) bycomparison with concentrations of internal standards, assuming aresponse factor of 1. Methyl octanoate, 2-methyl-1-pentanol, and3-methyl-3-hydroxy-2-butanone were used as the internal standards forfractions 1, 2, and 3, respectively.

Statistical Analysis.Analyses of variance were run on the GC datausing PROC GLM on Statistical Analysis Systems (SAS) for Windows,version 6.12 (Cary, NC). Principal component analysis (PCA) wasperformed on the mean ratings for eight sensory attributes using thecovariance matrix with no rotation on SAS. The sensory data setincluded eight aroma descriptors, which were peach/apricot, citrus,floral, caramel, butter, vanilla, spice and oak. The GC data set included74 OA volatiles. Of the 81 OA compounds quantified, five compounds(ethyl acetate, 1,1-diethoxy ethane, ethyl propionate, ethyl isobutyrate,and 2-pentanone) were eliminated from the instrumental data set becausethey could not be quantified in many of the wines, as wereâ-dama-scenone and unknown KI) 2157, which could not be quantified inany wine.

PLS is a “soft modeling” method, which extracts linear combinationsof variables from one data set (OA volatiles) that best predict variationin another data set (sensory ratings). To explore the relationship betweenthis sensory profile data and the OA volatiles for these 19 wines, partialleast squares regression analysis (PLS) was conducted using theUNSCRAMBLER ver.7.6 (CAMO A/S, Trondheim, Norway). Asecond PLS was run excluding all OA compounds for which less than50% variance was explained. The 16 odor-active compounds selectedfor the resulting PLS model are identified inTable 2. In all PLSanalyses, the odor-active volatiles were standardized (mean/standarddeviation), while the aroma intensity ratings were assigned a weightingof 1 (unstandardized). In both PLS models, the GC data were treatedas the independent variables (X matrix), with the sensory data used asthe dependent variables (Y matrix).

RESULTS AND DISCUSSION

GC-O. Eighty-one odor-active peaks were detected by twoor more judges in at least one fraction of one of the four winesevaluated by GC-O by four judges. With the exception of1-butanol and two unknown compounds (KI) 1762 and 2371),the same OA peaks were also detected in one or more of theother eight wines, in which no additional OA compounds werefound. In Table 2, the KI, tentative identification, fraction inwhich the compound was detected and aroma descriptions byGC-O are shown for the OA compounds. In addition, for thefour wines for which GC-O was performed by four judges, thenumber of times each peak was detected is reported.

With the exception of huge peaks such as fusel alcohols andorganic acids, which eluted in two fractions, the internalstandards and each volatile were detected in only one fraction.Although many of the tentatively identified OA compounds havenot been reported in Chardonnays, all have been previouslyreported in grapes or wines of other varieties ofVitis Vinifera(seeTable 2) or studies of oak flavor (22).

As shown in the PCA of the sensory data (Figure 1), thewines were primarily separated by intensity of “oak” terms (oak,caramel, butter, vanilla, and spice) versus “fruit” descriptors(peach/apricot, citrus, and apple) and floral. For example, winesCAL and DEL are high in the fruity and floral notes and lowin oak terms, whereas the converse is true for wines JL andCDC. The detection frequencies of the odorants were inspectedto see if there was any correspondence to these sensory notes(Table 2). Wines with higher intensity of oaky and spicy notes,such as wines JL and CDB-C, had fairly high detectionfrequencies for “oaky/spicy” compounds, most of which arisefrom oak-aging, such as (trans) oak lactone, eugenol, 4-vinyl

guaiacol, vanillin, and furfural. Similarly, linalool andR-terpi-neol had high detection frequencies in the wines high infruitiness and floral notes, such as CAL and DEL. Other volatilessuch as the fruity esters produced during fermentation did notshow appreciable differences in detection frequencies amongthe wines. Perhaps this is because the esters are ubiquitouslypresent at concentrations well above threshold levels.

Each of the eight identified compounds, previously reportedby Moio et al. (7) as contributing to the distinctiveness ofBurgundian Chardonnay were found in all wines in the presentstudy. In contrast, only two of the four compounds claimed byLe Fur et al. (9) to be potent odorants in French Chardonnayswere detected in these 19 wines (guaiacol (2-methoxy phenol)and ethyl cinnamate). Four norisoprenoids (vitispirane,â-dama-scenone, 3-oxo-R-ionol, and TDN (1,1,6-trimethyl-1,2-dihy-dronaphthalene)) have been suggested to be important to thearoma of Chardonnay (6, 10, 23). In the present study, onlyâ-damascenone and 3-oxo-R-ionol were found.â-Damascenoneelicited a strong honey/cooked apple note in six wines, but wasnot detected by GC/O in the other six. Upon elution of 3-oxo-R-ionol, a low intensity of spiciness was detected in wines JLand CDB-C but not detected by GC/O in the fruity CAL andDEL wines. However, both compounds were detected by GC-MS in all wines.

Composition of OA Compounds.The mean concentrationsof 79 OA compounds are given inTable 3 for the 19 wines.Of 81 OA compounds, two compounds (â-damascenone andunknown KI ) 2157) could not quantified because of weakchromatographic signals. Over 80% of the total volatile materialwas contributed by seven compounds: isoamyl alcohol, 2,3-butanediol, diethyl malate, acetic acid, hexanoic acid, octanoicacid, and 2-phenylethanol. Except for three compounds (2,3-butanediol, diethyl malate, and acetic acid), the concentrationof these compounds did not vary much across wines. Thus thesecompounds are speculated to contribute to the background orbase flavor of these wines rather than to differentiate amongthe wines.

To simplify inspection ofTable 3, compounds are groupedby their aromas as described by GC-O. The wines are arrangedfrom left to right to reflect the progression in aroma characterfrom very oaky (low fruity) to very fruity (low oaky) wines.The individual concentrations of the oak-associated OA com-pounds roughly correspond to intensity of oaky aromas of thesewines. Guaiacol, the oak lactone isomers, 4-ethyl guaiacol,eugenol, and vanillin, which mainly come from oak barrelcontact, were found in higher concentrations in oaky/spicy winesthan those in fruity and floral wines. JL, which was the mostintense in oak, vanilla, and spice aromas by descriptive analysis,had the highest concentrations of furfural, guaiacol, 4-ethylphenol, 4-ethyl guaiacol, (cis) oak-lactone, eugenol, and vanillin.In examining the pattern of distribution of the “floral” com-pounds, concentrations of three compounds (linalool,R-terpe-neol and a “minty” unknown (KI) 1688)) correlated signifi-cantly with the intensity of the floral aroma (r ) 0.80, 0.69,0.75 respectively). DEL wine had the highest concentration ofthese three “floral” compounds, while they were far lower inthe oaky/spicy wines (JL, CDB-C, VMEres, TES) than in otherfloral and fruity wines (CAL, SH, FET, VME). The higherconcentrations of these terpenes in DEL reflect its varietalcomposition. DEL had 2% (v/v) Muscat Canelli and 1% WhiteRiesling, whereas the other 18 wines were 100% Chardonnay.Unlike the floral compounds, the sums of the concentrations ofthe fruity odor compounds were similar across all 19 wines,even though there were large variations in fruitiness.

8038 J. Agric. Food Chem., Vol. 51, No. 27, 2003 Lee and Noble

Table 2. Odor-Active Compounds Found in Californian Chardonnay Wines. Frequency of Detection in GC−O Analyses and Odor Description by FourJudges

detection frequencyc

KIa Fb odorants JL CDB−C CAL DEL odor descriptiond refe

885 III ethyl acetatef 2 2 2 2 sweet, fruity c−e900 III 1,1-diethoxy ethaneg 4 4 2 1 buttery, creamy c−e950 I ethyl propionatef 1 1 1 0 juicy fruit c−e955 I ethyl isobutyratef 1 2 1 2 fruity c−f960 I 2-pentanoneg 0 3 2 1 fruity d,e

1028 I ethyl butanoatef 4 3 4 4 fruity, banana b−f1038 III 1-propanolg 2 2 0 2 musty c,d,e1053 I ethyl 3-methyl butanoatef 3 4 3 4 fruity, apple b−f1085 II, III 2-methyl-1-propanolg 3 2 2 2 glue, alcohol c−f1116 II 2-pentanolg 1 1 2 0 green-fruit, sweet d,e1118 I isoamyl acetatef,i 2 2 3 4 banana b−f1138 III 1-butanolg 1 1 0 0 medicinal c,d,e1206 II, III 2/3 methylbutanolg 4 4 4 4 balsamic, alcohol b−f1229 I ethyl hexanoatef 4 3 4 4 fruity-juicy a−f1272 III acetoing 4 4 4 2 butterscotch c,d,e1276 I unknowni 3 1 3 3 plastic1354 II 1-hexanolg 3 2 1 2 green grass a,c,d−f1379 II (trans) 3-hexen-1-olg 2 0 1 1 green a,c,d,e1427 I ethyl octanoatef 1 0 0 1 sweet, soapy, fresh a,c,d−f1434 III acetic acidg 4 4 4 4 vinegar d,e,f1458 II furfuralg 3 4 3 3 woody, almond c,d,e1487 II 2-ethyl-1-hexanolg 2 0 0 2 mild green, alcohol c,d1500 II 2-acetyl furanf,i 2 1 0 0 sweet caramel c,d1523 II propanoic acidg 0 0 1 3 soy d,e1542 III 2,3-butanediol (d,l)g 0 2 2 0 butter, creamy d1544 II linaloolg,i 2 1 1 4 floral a−f1557 III 2-methyl propanoic acidg 1 2 2 0 sweaty d,e1587 III unknown 0 0 2 1 glue, alcohol, thinner1614 II, III butanoic acidg 4 4 4 4 sweaty a, d−f1630 I ethyl decanoateg 2 0 1 1 fruity a,c,d,e1635 III butyrolactoneg 3 4 2 2 sweet, musty a,c,d,e1651 III unknown 2 0 2 2 perfumy rose1652 I unknown 3 0 0 0 fruity1660 II, III 2/3 methyl butanoic acid 4 3 4 4 stinky socks, sweaty c−f

+ furfuryl alcoholg,i

1687 II R-terpineolg 0 0 0 2 minty a,c,d1688 III unknowni 0 0 2 1 minty1700 I unknown 0 0 2 2 bread, smokey1714 III 3-methylthio-1-propanolg 4 3 4 3 potato b,d−f1722 II unknown 0 0 1 1 chemical, green1730 II pentanoic acidg 2 2 2 2 sweaty a,d1740 III unknown 2 0 1 0 musty1762 I unknown 0 0 0 3 honey1809 I 2-phenylethyl acetatef,i 4 4 4 4 honey, apple c−f1810 III unknown 3 4 4 4 cooked sugar, honey1813 II â-damascenoneg 1 0 4 4 honey, cooked apple a,c−f1840 II hexanoic acidg 4 4 4 4 sweaty a,d−f1853 II 2-methoxy phenolf,i 4 1 2 3 smoky, spicy b−f1886 III (cis) oak-lactonef,i 2 1 0 0 spicy b,-e1910 II, III 2-phenyl alcoholg 4 4 3 4 floral a−f1957 III (trans) oak-lactonef,i 3 3 3 3 spicy c−e1962 II (trans) 2-hexenoic acidg 0 0 1 2 fatty, musty g1972 II unknown 0 0 2 0 spicy, brown1973 I unknowni 0 0 3 1 bandaid, caramel2024 II 4-ethyl guiacolf, i 3 1 2 2 spicy, smokey e2033 III pantolactoneg 4 4 4 4 cotton candy d,e2053 III diethyl malateg 2 3 4 2 brown sugar d2058 II octanoic acidg 2 4 3 3 cheese a,d,f2079 III γ−nonalactoneg,i 2 3 3 0 sweet, creamy a,d,e2100 II unknown 0 0 0 2 cottoncandy2101 III homofuraneolg 0 0 3 3 cotton candy, jam f2130 III unknown 2 2 3 0 sweet, creamy2139 I ethyl cinnamateg 1 1 1 3 raisin b−f2156 III unknown 4 2 1 3 cooked sugar, apple sauce2157 II unknown 0 0 4 3 cooked sugar, spicy2164 II eugenolf,i 4 4 3 3 clove a,b,d−f2171 III diethyl-2-hydroxy- pentanedioateh 3 2 3 0 cotton candy d2195 II 4-ethyl phenolg 2 0 1 2 medicinal, phenolic b,d,e2200 III 4-vinyl-2-methoxy-phenolg 4 4 3 4 smokey, nutty a,b,d−f2220 III unknown i 0 2 0 3 smokey, woody2234 II unknown 0 0 2 1 spicy2241 III 4-ethoxycarbonyl-γ-butanolactoneh 2 2 2 3 smokey, toasted d,e2272 II decanoic acidg 0 4 4 2 dusty a,d,e


Relating Instrumental and Sensory Data.The PLS modelusing 74 GC peaks yielded a one-dimensional model whichexplained 67% of variance in the sensory data and 17% of theGC data in the first dimension. InFigure 2, the correlationloadings for the sensory terms (in capitals) and the OAcompounds are shown with the wine scores. Similar to thepattern seen in the PCA (Figure 1), there is a distinct separationbetween fruity/floral and oak related terms and between fruitywines (CAL, DEL) and oaky wines (JL, CDB-C, VMEres,CDB-A, TES) along the first dimension. Similarly, the firstdimension contrasts compounds with fruity/floral aromas versusthose that have oaky notes. Linalool and unknown 1688 (whichhad a minty odor), were associated with the “floral” sensoryattribute. Fruity esters such as isoamyl acetate and 2-phenylethylacetate were located close to the “peach” and “citrus” attributes.Similarly, oaky odor compounds (4-ethyl guaiacol, eugenol,2-acetyl furan, (cis) oak-lactone, 2-methoxy phenol, and vanillinwere associated with “oak” and “spice” sensory attributes.(trans) Oak-lactone andγ-nonalactone were more closelyassociated with the “vanilla” and “butter” terms.

The center ellipsoid inFigure 2 indicates 50% explainedvariation. This means that all the GC peaks located inside thiscircle were poorly modeled and mathematically do not explainvariation in the sensory data. The PLS was rerun after

eliminating these peaks to more clearly see the relationshipsbetween OA compounds that had a strong predictive relationshipto the sensory terms. The PLS model using 16 OA peaks(Figure 3) explained 64% variance in the sensory data and 47%for the GC variables. All 16 of the OA variables weresignificantly modeled in this PLS, as determined by theuncertainty test (24). As with the PLS model using 74 OA peaks,compounds with oaky/spicy notes are located close to the oakysensory terms, while those with fruity or floral notes are locatedclosely to citrus and peach and floral, respectively, indicatingtheir strong correlations. For example, vanillin is significantlycorrelated with “oak”, “vanilla” and “spice”, respectively,r )0.85, 0.84, and 0.78 (p < 0.001).

Although the PLS model shows a strong relationship betweenthese 16 OA volatiles and the sensory profile data, it does notestablish a causal relationship. Only by sensory evaluation ofwines or systems spiked with selected compounds can it bedetermined if OA compounds contribute significantly to thecharacteristic aromas of Chardonnay wines.

CONCLUSION

OA compounds were detected by GC-O of volatile fractionsisolated from 12 wines. The same compounds were found in

Table 2. (Continued)

detection frequencyc

KIa Fb odorants JL CDB−C CAL DEL odor descriptiond refe

2273 III 2,6-dimethoxyphenolg 4 3 3 0 nutty, smokey h2358 III diethyl tartarateg 4 3 1 2 earthy, musty a,d2371 II unknown 0 0 0 2 mushroom2400 III unknown 2 2 3 2 thinner2481 III unknown 2 0 2 2 mothball2512 III 5-hydroxymethyl-2-furfuralg 2 0 0 0 cardboard, paper d,e2561 III vanilinf,i 4 4 3 4 vanilla a,b,d−f2640 III acetovanilloneg 2 2 0 0 spicy, sweet a,d,e2650 III 3-oxo-R-ionolh 3 2 0 0 spicy a

a Kovats indices of unknown compounds on DB−WAX column. b Fraction in which most of compound appeared after column chromatography. c Detection frequency ofterm used over 4 GC/O runs. d Odor description usually reported by at least 2 judges. e Volatiles reported previously in wines or grapes. Letter corresponds to numberedreference. a,(10); b, (7); c, (23); d, (2); e, (5); f, (19); g, (25); h, (26). f Identified by comparison with MS spectra and KI of authentic references. g Identified by comparisonwith published MS spectra and KI. h Identified by comparison with MS spectra in Wiley 275 library. i Odor-active compounds used in second PLS model.

Figure 1. Principal Component Analysis of 19 wines from descriptive analysis using 8 aroma terms (17). Attribute loadings are shown as vectors; winescores are shown as capital letters. Wine codes are defined in Table 1.


Tabl

e3.

Mea

nC

once

ntra

tions

(mg/

L)of

Odo

r-act

ive

Com

poun

dsin

19W

ines

(n)

2)

KIa

code

bJL

VMEr

esC

DB-

AC

DB-

CTE

SPR

EBE

MER

GP

CD

B-S

RH

PD

UN

CO

NVM

ETA

FTFE

TSH

DEL

CAL

Frui

ty88

5et

ac1.

913

2.88

86.

805

3.10

71.

505

1.46

12.

540

2.39

62.

033

7.91

66.

805

7.22

695

0et

prop

0.06

60.

053

0.12

00.

083

0.08

10.

100

0.05

595

5et

ibut

0.06

70.

060

0.02

70.

036

0.06

30.

028

0.03

196

02p

ento

ne1.

312

0.18

01.

609

0.13

60.

492

0.44

00.

179

0.69

10.

415

0.31

60.

566

0.50

110

28et

but

1.04

00.

657

0.72

61.

086

0.69

50.

879

0.66

40.

823

1.13

60.

445

0.74

31.

081

0.73

40.

715

0.91

80.

563

1.36

00.

906

0.84

410

53et

ipen

t0.

041

0.05

90.

011

0.05

10.

056

0.00

90.

008

0.08

30.

007

0.01

50.

092

0.00

40.

049

0.02

80.

057

0.00

60.

038

0.02

90.

042

1116

2pen

tol

0.38

00.

033

0.13

20.

676

0.04

30.

189

0.09

70.

107

0.19

60.

119

0.04

40.

129

0.08

40.

313

0.19

90.

207

0.08

90.

088

0.14

011

18is

oam

ac0.

349

0.24

60.

307

0.41

70.

246

0.12

90.

220

0.47

60.

159

0.23

70.

715

0.24

70.

442

0.49

10.

530

0.30

70.

690

0.52

20.

519

1229

ethe

x0.

600

0.83

51.

243

1.34

10.

755

0.79

40.

774

0.78

70.

795

0.98

30.

952

0.66

40.

864

0.77

91.

178

0.82

20.

693

0.72

20.

843

1427

etoc

t0.

808

1.48

21.

761

1.87

71.

212

1.47

71.

249

1.28

51.

455

1.78

51.

387

1.10

91.

486

1.32

21.

939

1.38

81.

162

1.09

81.

284

1630

etde

c0.

128

0.22

40.

221

0.21

80.

182

0.25

70.

198

0.20

50.

195

0.29

30.

209

0.14

30.

284

0.22

40.

297

0.18

00.

268

0.17

70.

191

1652

unk1

652

0.39

91.

318

0.29

10.

070

0.37

90.

271

0.82

60.

143

0.11

80.

272

0.55

10.

058

0.26

30.

310

0.17

70.

083

0.00

80.

028

0.01

018

09ph

enyl

ac0.

056

0.09

10.

073

0.04

90.

047

0.03

40.

066

0.06

80.

041

0.08

40.

090

0.03

60.

128

0.08

10.

099

0.07

70.

121

0.06

80.

079

2139

etci

nna

0.00

60.

007

0.00

20.

006

0.00

70.

002

0.00

10.

005

0.00

40.

002

0.00

40.

003

0.00

60.

012

0.01

10.

002

0.03

30.

004

0.00

3

Flor

al15

44lin

aloo

l0.

031

0.01

90.

033

0.04

00.

027

0.01

40.

028

0.03

70.

024

0.05

40.

021

0.02

90.

070

0.08

80.

056

0.04

30.

109

0.14

20.

050

1651

unk1

651

0.05

80.

111

0.07

80.

016

0.09

60.

104

0.09

80.

091

0.08

50.

046

0.04

80.

120

0.12

50.

057

0.06

30.

059

0.03

90.

043

0.04

516

87at

erpn

ol0.

034

0.02

10.

106

0.10

90.

043

0.05

10.

078

0.02

60.

053

0.10

70.

016

0.04

40.

058

0.07

50.

066

0.08

60.

044

0.18

10.

111

1688

unk1

688

0.01

80.

018

0.06

00.

000

0.01

90.

022

0.02

80.

024

0.06

80.

035

0.01

80.

032

0.02

20.

024

0.01

80.

054

0.05

20.

110

0.07

119

10ph

enyl

ol11

6.85

92.7

890

.04

30.8

862

.15

66.7

374

.90

144.

2847

.99

54.3

827

.03

103.

7648

.86

101.

9260

.20

88.0

939

.90

52.4

915

3.84

Butte

r,cr

eam

y90

0di

etho

x0.

479

0.89

30.

214

0.73

20.

998

0.51

40.

743

1.39

50.

923

0.53

30.

649

1.30

312

72ac

etoi

n1.

675

3.23

44.

808

1.23

70.

743

2.65

11.

048

1.29

82.

334

4.57

20.

430

0.69

81.

917

2.02

02.

601

0.96

62.

464

0.24

73.

245

1542

btud

iol

1.19

628

.467

33.6

901.

075

33.7

1253

.677

89.8

5439

.568

38.4

0827

.807

25.4

7441

.842

41.4

5620

.245

16.9

8048

.756

0.30

417

.302

29.0

9520

79no

nlac

tn0.

614

0.30

90.

205

0.42

30.

655

0.52

90.

263

0.43

70.

325

0.22

50.

307

0.21

10.

389

0.17

50.

269

0.49

50.

351

0.08

40.

227

2130

unk2

130

0.40

71.

194

1.20

70.

366

1.23

02.

835

1.68

91.

171

1.25

11.

146

0.75

20.

491

1.79

40.

483

1.03

21.

426

0.21

30.

456

Oak

-rela

ted

1458

furfu

ral

21.1

91.

650.

554.

050.

950.

670.

420.

521.

410.

571.

820.

360.

540.

920.

890.

330.

230.

180.

2418

53m

etxp

henl

0.28

40.

105

0.06

70.

134

0.25

40.

037

0.04

10.

151

0.08

60.

038

0.11

00.

040

0.06

00.

091

0.05

10.

023

0.02

10.

049

0.02

518

86oa

klac

tnc

0.38

20.

208

0.19

40.

346

0.08

00.

232

0.16

60.

171

0.10

60.

087

0.06

50.

059

0.06

20.

084

0.07

80.

067

0.03

70.

053

0.03

319

57oa

klac

tnt

0.99

60.

960

1.35

50.

329

0.83

40.

783

0.50

20.

472

0.98

50.

724

0.79

60.

727

0.60

50.

492

0.34

30.

472

0.19

80.

149

0.17

319

72un

k197

20.

000

0.01

90.

015

0.00

00.

046

0.01

20.

010

0.04

30.

013

0.01

10.

013

0.01

10.

016

0.01

10.

011

0.01

20.

094

0.03

74.

043

2024

etgu

iacl

0.05

00.

021

0.00

80.

015

0.01

40.

007

0.00

50.

019

0.00

90.

004

0.00

40.

003

0.00

40.

006

0.00

70.

003

0.00

00.

003

0.00

021

64eu

geno

l0.

362

0.15

70.

249

0.32

70.

102

0.09

90.

081

0.11

40.

110

0.17

40.

101

0.07

80.

073

0.08

90.

076

0.03

10.

019

0.05

90.

021

2195

4etp

henl

0.12

90.

008

0.00

00.

089

0.03

90.

005

0.00

90.

020

0.00

30.

000

0.01

20.

006

0.01

10.

005

0.03

50.

006

0.02

00.

014

2200

vigu

iacl

0.38

01.

730

1.41

51.

024

3.25

90.

560

0.66

71.

105

0.85

31.

860

3.45

60.

981

2.61

21.

259

1.59

00.

896

1.34

21.

493

1.35

622

20un

k222

00.

108

0.17

10.

317

0.14

00.

521

0.27

90.

100

0.10

20.

224

0.28

60.

055

0.16

50.

088

0.12

70.

079

0.14

10.

050

0.09

10.

078

2234

unk2

234

0.04

20.

057

0.12

70.

044

0.13

90.

041

0.02

70.

059

0.05

80.

123

0.06

50.

041

0.09

00.

077

0.06

50.

054

0.08

20.

078

0.08

122

414e

cbul

ac1.

522

5.22

06.

106

1.65

36.

081

3.51

23.

968

4.27

06.

526

5.84

63.

872

5.25

05.

696

2.77

02.

434

4.48

02.

041

2.45

61.

417

2273

syrin

gol

0.24

40.

537

0.51

20.

086

0.68

40.

439

0.41

60.

454

0.49

90.

685

0.69

90.

314

0.62

80.

376

0.51

00.

378

0.15

50.

231

0.06

625

61va

nilin

1.22

30.

983

1.17

20.

345

0.78

70.

760

0.61

00.

634

0.57

20.

613

0.74

50.

239

0.28

80.

496

0.37

80.

390

0.25

40.

166

0.10

726

40ac

vani

ln0.

391

0.87

61.

068

0.26

21.

056

0.97

10.

897

0.85

31.

386

1.15

50.

523

1.59

40.

849

0.55

40.

516

0.97

90.

253

0.35

20.

233

2650

oxoi

onl

0.30

12.

674

1.72

10.

551

1.42

80.

451

0.92

20.

997

0.78

71.

426

0.98

70.

885

2.28

20.

426

0.54

00.

662

0.62

50.

866

0.47

8

Burn

tsug

ar,c

aram

el15

002a

cfyr

an0.

313

0.12

30.

214

0.22

10.

073

0.11

70.

120

0.07

00.

069

0.11

00.

079

0.06

90.

042

0.08

40.

086

0.04

90.

044

0.03

017

62un

k176

20.

010

0.00

90.

003

0.01

00.

005

0.00

10.

001

0.00

40.

003

0.00

10.

004

0.00

20.

006

0.00

60.

003

0.00

20.

037

0.00

90.

006

1810

unk1

810

2.72

36.

885

5.19

01.

592

7.24

24.

107

6.07

85.

263

6.69

15.

345

8.17

59.

052

8.19

96.

067

4.52

35.

693

4.72

14.

803

3.55

419

73un

k197

30.

000

0.00

60.

002

0.00

70.

006

0.00

00.

002

0.00

30.

003

0.00

10.

003

0.00

30.

005

0.00

40.

004

0.00

10.

028

0.00

90.

019

2033

panl

actn

0.19

10.

691

0.76

70.

163

0.70

20.

518

0.53

10.

407

0.53

20.

860

0.49

70.

535

0.58

20.

511

0.30

60.

626

0.15

30.

364

0.41

3


Tabl

e3.

(Con

tinue

d)

KIa

code

bJL

VMEr

esC

DB-

AC

DB-

CTE

SPR

EBE

MER

GP

CD

B-S

RH

PD

UN

CO

NVM

ETA

FTFE

TSH

DEL

CAL

Burn

tsug

ar,c

aram

el(C

ontin

ued)

2053

diet

mal

18.2

533

.02

43.3

44.

9952

.07

36.4

092

.74

11.2

214

.37

47.9

827

.84

20.7

326

.72

14.5

79.

5455

.93

10.0

38.

8218

.18

2100

unk2

100

0.25

10.

386

0.08

90.

446

0.17

40.

124

0.19

20.

072

0.05

50.

043

0.14

10.

014

0.02

80.

104

0.04

90.

024

0.11

30.

078

0.07

021

01ho

mof

ura

0.15

90.

206

0.12

60.

092

0.37

40.

345

0.29

80.

250

0.22

50.

182

0.29

30.

195

0.28

30.

139

0.20

60.

269

0.11

10.

136

0.24

221

56un

k215

60.

074

0.05

80.

067

0.07

20.

031

0.03

90.

052

0.05

90.

078

0.02

40.

041

0.14

00.

136

0.06

20.

059

0.14

20.

091

2171

det2

pen

2.93

37.

871

10.9

143.

670

8.98

96.

066

6.41

06.

884

10.4

629.

965

4.80

67.

772

6.34

84.

333

3.95

26.

344

3.19

72.

524

2.15

5Al

coho

l13

541h

exol

5.02

22.

182

3.59

81.

985

2.60

12.

740

1.74

72.

051

4.09

03.

777

2.71

54.

912

2.29

82.

881

1.89

93.

532

2.75

03.

051

2.32

513

793h

exen

ol0.

218

0.11

70.

114

0.20

30.

137

0.11

80.

063

0.19

00.

118

0.10

40.

082

0.13

70.

073

0.12

30.

152

0.10

70.

111

0.11

40.

062

1487

2eth

exol

0.10

80.

094

0.11

60.

200

0.18

30.

125

0.15

20.

248

0.14

80.

167

0.14

30.

153

0.12

00.

131

0.26

30.

143

68.3

30.

122

0.10

717

22un

k172

20.

032

0.04

70.

022

0.04

60.

042

0.00

70.

015

0.04

90.

006

0.01

80.

043

0.01

30.

012

0.05

50.

007

0.01

70.

035

0.02

60.

019

1038

1pro

nol

0.19

70.

144

0.09

20.

306

0.13

90.

039

0.09

60.

693

0.05

10.

072

0.10

90.

066

0.11

50.

199

0.19

40.

071

0.61

30.

364

0.21

910

852m

e1pr

op3.

057

0.41

00.

312

2.33

00.

450

0.21

10.

245

0.50

40.

287

0.29

50.

692

0.29

90.

645

0.82

30.

295

0.20

41.

017

1.19

80.

642

1138

1but

ol0.

430

0.06

40.

103

0.46

00.

089

0.18

00.

099

0.17

90.

103

0.09

00.

087

0.11

40.

121

0.34

60.

176

0.09

40.

095

0.27

60.

327

1206

isop

enol

57.4

436

.51

113.

4861

.26

42.0

410

3.55

81.0

348

.80

112.

1110

8.71

44.2

510

1.33

56.2

969

.09

52.3

892

.90

41.4

765

.10

63.5

617

14m

ethi

opro

0.45

92.

024

2.05

40.

410

3.02

71.

719

2.01

72.

210

3.57

92.

491

1.18

03.

428

3.71

01.

488

0.98

32.

369

1.27

31.

327

1.48

9

Acid

s14

34H

ac11

.29

11.9

325

.19

6.46

8.31

46.6

045

.75

12.0

931

.42

21.0

57.

8735

.35

7.96

9.96

7.19

23.3

310

.30

8.78

10.0

015

23pr

opic

0.61

40.

606

0.25

30.

680

0.57

10.

469

0.92

80.

749

0.38

10.

063

1.80

00.

559

0.42

40.

474

0.48

50.

160

1.78

00.

399

0.27

915

57m

epro

pic

0.57

80.

929

0.89

10.

451

0.68

80.

932

0.79

50.

878

1.06

40.

818

0.64

81.

008

1.29

00.

824

0.59

40.

774

0.61

50.

373

0.27

816

14bu

tic1.

505

1.65

24.

703

1.97

21.

626

2.00

52.

193

1.38

13.

841

4.55

31.

934

3.63

11.

866

1.53

51.

603

4.11

12.

330

1.76

91.

824

1660

me+

fur

4.03

82.

734

3.08

51.

912

2.50

73.

575

3.09

83.

897

3.07

13.

297

2.09

83.

025

1.64

55.

028

1.47

41.

785

1.52

22.

270

1.27

017

30pe

ntic

0.03

60.

066

0.03

20.

021

0.06

50.

016

0.02

30.

071

0.02

30.

023

0.03

50.

032

0.02

00.

047

0.01

50.

029

0.53

50.

036

0.02

018

40he

xic

6.66

67.

788

18.7

6217

.107

7.48

811

.581

10.2

457.

006

12.9

9116

.699

7.36

810

.319

7.56

67.

314

8.66

913

.828

8.47

36.

199

7.37

519

62he

xeic

0.00

00.

166

0.12

60.

000

0.23

00.

080

0.08

80.

150

0.10

90.

139

0.16

90.

109

0.04

90.

127

0.15

00.

152

0.00

00.

130

0.07

920

58oc

tic11

.82

14.4

530

.13

50.5

613

.46

19.0

716

.37

12.9

822

.11

31.5

112

.80

15.2

314

.35

13.7

316

.83

21.6

211

.32

10.1

512

.52

2272

deci

c3.

670

5.45

37.

139

14.1

494.

397

5.30

44.

557

5.26

35.

774

9.03

64.

329

3.34

45.

188

4.56

15.

658

5.15

44.

064

0.54

93.

533

Mis

cella

neou

s12

76un

k127

60.

143

0.19

70.

080

0.10

90.

164

0.18

10.

107

0.15

60.

106

0.06

80.

048

0.08

00.

028

0.13

80.

039

0.04

20.

107

0.02

80.

017

1587

unk1

587

0.12

70.

209

0.08

10.

129

0.15

30.

038

0.04

50.

245

0.06

00.

050

0.12

20.

072

0.22

90.

257

0.14

70.

074

0.16

60.

207

0.28

416

35bu

lact

n1.

660

5.18

02.

238

1.08

95.

395

2.76

03.

959

3.76

23.

034

2.49

23.

080

4.57

34.

648

3.84

32.

350

2.77

61.

708

2.59

72.

327

1700

unk1

700

0.00

30.

031

0.01

00.

007

0.02

80.

009

0.00

90.

015

0.01

60.

010

0.00

80.

012

0.03

20.

014

0.01

80.

005

0.01

00.

002

1740

unk1

740

0.39

00.

796

1.73

80.

339

1.11

70.

352

1.84

02.

303

0.96

61.

646

1.53

21.

176

1.16

11.

453

1.31

11.

786

1.37

91.

445

0.90

223

58di

etta

r0.

338

3.45

45.

204

1.12

23.

335

8.93

74.

921

0.96

87.

792

4.42

81.

313

1.70

82.

119

1.49

91.

221

3.44

90.

619

0.30

40.

691

2371

unk2

371

0.28

50.

024

0.06

90.

445

0.03

70.

012

0.01

60.

032

0.02

70.

043

0.03

10.

010

0.03

70.

021

0.02

80.

022

0.16

30.

094

0.09

124

00un

k240

069

.99

172.

0444

9.92

70.3

221

8.87

297.

2624

8.39

168.

8344

4.40

516.

1110

2.94

378.

6921

5.43

155.

3111

5.28

271.

9157

.33

82.6

086

.53

2481

unk2

481

0.26

50.

226

0.48

80.

283

1.50

60.

654

1.09

01.

341

2.13

70.

559

0.11

60.

541

0.26

10.

174

0.53

70.

832

0.36

40.

315

0.20

525

12hm

f0.

175

0.11

50.

198

0.03

60.

110

0.08

50.

085

0.24

50.

219

0.17

70.

082

0.00

00.

016

0.06

00.

053

0.12

00.

114

0.04

40.

058

aKo

vatis

indi

ces

ofun

know

nco

mpo

unds

onD

B-W

AXco

lum

n.b

See

Tabl

e2

for

full

nam

eof

com

poun

d.


all nineteen wines by GC-MS. PLS regression analysis yieldeda configuration similar to that found by PCA of the sensorydescriptive analysis data for the same wines. The largestdifference among the wines was a contrast between wines highin intensity of floral and fruity notes (and high in concentrationsof compounds with floral aromas, such as linalool andR-ter-pineol) and wines high in oaky and spicy notes (and high inconcentration of oak lactones, vanillin, 2-methoxy phenol, andeugenol). However, sensory evaluation of these OA compoundsmust be made to confirm that they contribute to the aroma ofChardonnay wines and determine which are most important inproducing the characteristic flavor of these wines.

ABBREVIATIONS USED

PCA, principal component analysis; PLS, partial least-squaresregression; GC, gas chromatography; GC-O, gas chromatogra-phy-olfactometry; GC-MS, gas chromatography-mass spectrom-etry; OA, odor-active

ACKNOWLEDGMENT

We gratefully acknowledge E&J Gallo winery for the supportand contribution for GC-O-MS analysis in their facility. Weare also thankful to Dr. Jane Yegge and Kevin Scott in our labfor their contribution in this research.

LITERATURE CITED

(1) Yegge, J. M.; Noble, A. C. Proceedings of the American Societyof Enology and Viticulture, 50th Anniversary Annual Meeting,Seattle, WA;Am. J. Enol. Vitic.2001 (special edition), p 28-31.

(2) Schreier, P. Flavor composition of wines: a review.CRC Crit.ReV. Food Sci. Nutr.1979, 12, 59-111.

(3) Rapp, A. Volatile flavour of wine: correlation between instru-mental analysis and sensory perception.Nahrung1998, 42, 351-63.

(4) Ebeler, S. E. Analytical chemistry: Unlocking the secrets of wineflavor. Food ReV. Intl. 2001, 17, 45-64.

Figure 2. Correlation loadings from PLS of 74 GC variables (small letters) and 8 sensory variables (capital letters). Bold capital letters are wine scores.Explained variance for X (GC data) is 7 and 23% for PC1 and PC2, respectively, and for Y (Sensory data) is 64 and 4%, respectively. Codes for winesand OA compounds are defined in Table 1 and Table 3, respectively.

Figure 3. Correlation loadings from PLS of 16 GC variables (small letters) and 8 sensory variables (capital letters). Bold capital letters are wine scores.Explained variance for X (GC data) is 47 and 8% for PC1 and PC2, respectively, and for Y (Sensory data) is 64 and 6%, respectively. Codes for winesand OA compounds are defined in Table 1 and Table 3, respectively.


(5) Etievant, P. X. Wine. InVolatile compounds of food andbeVerages; H. Maarse, Ed.; Dekker: New York, 1991; pp 483-546.

(6) Noble, A. C.; Flath, R. A.; Forrey, R. R. Wine HeadspaceAnalysis. Reproducibility and Application to Varietal Classifica-tion. J. Agric. Food Chem.1980, 28, 346-353.

(7) Moio, L.; Schlich, P.; Etievant, P. Acquisition et analysed′aromagrammes de vins de Bourgogne issus du cepage Char-donnay.Sci. Aliments1994, 14, 601-608.

(8) Acree, T. E.; Barnard, J.; Cunningham, D. G. The analysis ofodor-active volatiles in gas-chromatographic effluents. InAnaly-sis of Volatiles. Methods and Applications; P. Schreier, Ed.;Walter de Gruyter: Berlin, 1984; pp 251-268.

(9) Le Fur, Y.; Lesschaeve, I.; Etievant, P. X.4th InternationalSymposium on Cool Climate Viticulture and Enology, Rochester,NY; New York State Agricultural Experiment Station: Geneva,NY, 1996.

(10) Sefton, M. A.; Francis, I. L.; Williams, P. J. The volatilecomposition of chardonnay juices: A study by flavor precursoranalysis.Am. J. Enol. Vitic.1993, 44, 359-370.

(11) Francis, L.; Sefton, M. A.; Williams, P. J. Sensory DescriptiveAnalysis of the Aroma of Hydrolyzed Precursor Fractions fromSemillon, Chardonnay and Sauvignon blanc Grape Juices.J. Sci.Food Agric1992, 59, 511-520.

(12) de la Presa Owens, C. S.; P.; Wada, K.; Noble, A. C. UsingSensory and Instrumental Data to Interpret the Effect of Storageat Elevated Temperatures on Aroma of Chardonnay Wines.Annals NY Acad. Sci.1998, 855, 954-959.

(13) Le Fur, Y.; Mercurio, V.; Moio, L.; Blanquet, J.; Meunier, J.M. A new approach to examine the relationships between sensoryand gas chromatography-olfactometry data using generalizedprocrustes analysis applied to Six French Chardonnay wines.J.Agric. Food Chem.2003, 51, 443-452.

(14) Fischer, U.; Berger, R.; Hakansson, A.; Noble, A. C. The Impactof dealcoholization on the flavor of winesrelating concentrationof aroma compounds to sensory data using PLS analysis. InFlaVour Science: recent deVelopments.; Taylor, A. J., Mottram,D., Eds.; Royal Soc. Chem.: Cambridge, 1996; pp 335-338.

(15) Francis, I. L.; Kassara, S.; Noble, A. C.; Williams, P. J. TheContribution of Glycoside Precursors to Cabernet Sauvignon andMerlot Aroma: Sensory and Compositional Studies. InChemistryof Wine FlaVor; Waterhouse, A. L., Ebeler, S. E. Eds; AmericanChemical Society: Washington, DC, 1998; pp 13-30.

(16) Noble, A. C.; Ebeler, S. E. Use of multivariate statistics inunderstanding wine flavor.Food ReV. Intl. 2002, 18, 1-21.

(17) Yegge, J. Influence of sensory and nonsensory attributes ofChardonnay wine on acceptance and purchase intent, Ph.D.dissertation, University of California, Davis, 2001.

(18) Park, S. K.; Morrison, J. C.; Adams, D. O.; Noble, A. C.Distribution of Free and Glycosidically Bound Monoterpenesin the Skin and Mesocarp of Muscat of Alexandria GrapesDuring Development.J. Agric. Food Chem.1991, 39, 514-518.

(19) Guth, H. Identification of Character Impact Odorants of DifferentWhite Wine Varieties.J. Agric. Food Chem.1997, 45, 3022-3026.

(20) Pollien, P.; Ott, A.; Montigon, F.; Baumgartner, M.; Munoz-Box, R.; Chaintreau, A. Hyphenated Headspace-Gas Chromo-tography-Sniffing Technique: Screening of Impact Odorants andQuantitative Arogram Comparisons.J. Agric. Food Chem.1997,45, 2630-2637.

(21) Kovats, E. Gas chromatographic characterization of organicsubstances in the retention index system.AdV. Chromatogr.1965,1, 229-247.

(22) Sefton, M. A.; Francis, I. L.; Williams, P. J. Volatile noriso-prenoid compounds as constituents of oak woods used in wineand spirit maturation.J. Agric. Food Chem.1990, 38, 2045-2049.

(23) Simpson, R. F.; Miller, G. C. Aroma composition of Chardonnaywine. Vitis 1984, 23, 143-158.

(24) Martens, H.; Martens, M.MultiVariate Analysis of Quality: AnIntroduction; J. Wiley: Chichester England; New York, 2001;p 445.

(25) Wada, K.; Shibamoto, T. Isolation and identification of volatilecompounds from a wine using solid-phase extraction, gaschromatography, and gas chromatography/mass spectrometry.J.Agric. Food Chem.1997, 45, 4362-4366.

(26) Aznar, M.; Lopez, R.; Cacho, J. F.; Ferreira, V. Identificationand quantification of impact odorants of aged red wines fromRioja. GC-olfactometry, quantitative GC-MS, and odor evalu-ation of HPLC fractions.J. Agric. Food Chem.2001, 49, 2924-2929.

Received for review July 9, 2003. Revised manuscript received October30, 2003. Accepted November 2, 2003.

JF034747V


Characterization of Odor-Active Compounds in Californian Chardonnay Wines Using GC-Olfactometry and...

Documents

Transcript of Characterization of Odor-Active Compounds in Californian Chardonnay Wines Using GC-Olfactometry and...