Quantitative gas chromatography olfactometry and chemical quantitative study of the aroma of four Madeira wines

Analytica Chimica Acta xxx (2005) xxx xxx Quantitative gas chromatography olfactometry and chemical quantitative study of the aroma of four Madeira wines Eva Campo a, Vicente Ferreira a,, Ana Escudero a, José C. Marqués b, Juan Cacho a a Laboratory for Flavor Analysis and Enology, Department of Analytical Chemistry, Faculty of Sciences, University of Zaragoza, 50009 Zaragoza, Spain b Department of Chemistry, University of Madeira, Funchal 9000-390, Portugal Received 28 July 2005; received in revised form 6 October 2005; accepted 17 October 2005 Abstract The aroma profile of four Madeira wines from the most emblematic grape varieties, Malvazia, Boal, Verdelho and Sercial, has been studied by sensory analysis, quantitative gas chromatography olfactometry (GC O) and gas chromatography mass spectrometry (GC MS). The aroma of the wines was characterized as candy, nutty, maderized, toasty, lacquer and dried fruit. The GC O was carried out on extracts obtained by a dynamic headspace technique. The GC O profiles of Madeira wines were compared to the GC O profiles obtained from three young white monovarietal wines made with the same varieties so that the identification of odorants related to the particular process of elaboration of Madeira wines was possible. The aroma profile of Madeira wines was extremely complex, and was rich in sotolon, phenylacetaldehyde, wood extractable aromas, and lacked of the most important varietal aromas, such as linalool, 3-mercaptohexyl acetate and methoxypyrazines. A large number of potentially important and unknown odorants, most of them specific to Madeira wines, was also detected by GC O. The GC MS results confirmed most of the results of the GC O study, which suggests that the proposed GC O strategy is a useful tool for screening the presence of active odorants in wine. 2005 Published by Elsevier B.V. Keywords: Madeira; Grape; Gas chromatography olfactometry; Quantitative analysis; Sotolon; Phenylacetaldehyde 1. Introduction Madeira wine is produced following some traditional and very specific processes. Fermentation is first stopped by the addition of natural grape spirit (containing 95% (v/v) of ethanol) to obtain a wine containing ca. 18 20% (v/v) of ethanol and different amounts of unfermented sugars, ranging from 25 to 110gL 1, depending on the variety. After this, the wine is submitted to a baking process known as estufagem, during which the wine is kept at rather high temperatures (45 50 C) for as long as 90 days or even more. After this treatment, the wine is allowed to undergo a normal maturation process in oak casks (647 L) for a minimum period of 3 years. The four basic types of Madeira are named after the corresponding grape variety from which they are made, Malvazia, Boal, Verdelho and Sercial. These traditional winemaking and maturation procedures lead to the formation of the typical and characteristic bouquet of Madeira wines. Corresponding author. Tel.: +34 976762067; fax: +34 976761292. E-mail addresses: vferre@unizar.es, vferre@posta.unizar.es (V. Ferreira). Despite being one of the world s most famous dessert and aperitif wines, not much research has been carried out into Madeira wines. One of the first papers was published in the late 1990s by Nogueira and Nascimento [1]. In this work the authors characterized Madeira wines in relation to some physicochemical and sensorial parameters. Camara and co-workers have studied the evolution of sotolon and other furanic compounds in relation to the sugar content and the age of wine [2]. This same group of authors have also characterized the varietal volatile composition of the musts from the four white grape varieties used to produced Madeira wine [3] and have recently published a paper dealing with the characterization of the volatile profile of Madeira wines by sorptive extraction techniques [4]. However, and to the best of our knowledge, there are not previous studies dealing with the aroma composition of Madeira wines, which means that the number and nature of the odorants responsible for the characteristic aroma nuances of these wines are not known. The main aims of the present work are to determine and classify according to their potential sensory role, the odorants most likely involved in the aroma profile of Madeira wines and to determine the existence of specific odorants that may be related 0003-2670/$ see front matter 2005 Published by Elsevier B.V. doi:10.1016/j.aca.2005.10.035 ACA-226924; No. of Pages 8

2 E. Campo et al. / Analytica Chimica Acta xxx (2005) xxx xxx to the particular processes followed in the production of Madeira wine. The approach used is based on a quantitative gas chromatography olfactometric technique [5] carried out on extracts obtained by a particular dynamic headspace technique, which has been satisfactorily used to screen the presence of important odorants in the aroma profile of white monovarietal wines [6]. This technique is complemented by the GC MS determination of a large number of odorants, so that a comparison between GC O scores and odor activity values (OAVs) can be carried out. A secondary objective will be, therefore, to evaluate the potentiality of the GC O approach to screen the presence of powerful odorants. 2. Materials and methods 2.1. Wines Four samples of Madeira wines (10-years-old) from Malvazia, Boal, Verdelho and Sercial monovarietal grapes were supplied by the Madeira Wine Company (September 2004). Three monovarietal young white wines from 2003 vintage were also selected for this study. The wines elaborated with Malvasia, Boal and Verdello grape varieties were taken directly from cellars to ensure that they consisted of a single grape variety. Wines were supplied by a winery from the vicinal Canary Islands (Tenerife). Sensory analysis was carried out by a panel of experts to verify the quality of the samples, and to define their most important aromatic descriptors. The representativeness of the aroma of the corresponding variety (young wines) was also evaluated. The sensory study, the GC O analysis and the quantitative determination were carried out in the 3 months after the selection of the wines. During this period, the bottles were stored at 4 Cinthe dark. 2.2. Reagents and standards The chemical standards were supplied by Aldrich (Gillingham, UK), Fluka (Buchs, Switzerland), Sigma (St. Louis, USA), Lancaster (Strasbourg, France), PolyScience (Niles, USA), Chemservice (West Chester, USA), Interchim (Monlucon, France), International Express Service (Allauch, France) and Firmenich (Geneva, Switzerland). LiChrolut EN resins and polypropylene cartridges were obtained from Merck (Darmstadt, Germany). Dichloromethane and methanol of LiChrosolv quality was from Merck (Darmstadt, Germany); absolute ethanol, pentane and ammonium sulfate were from Panreac (Barcelona, Spain) and all of them were ARG quality; pure water was obtained from a Milli-Q purification system (Millipore, USA). Semiautomated solid phase extraction was carried out with a VAC ELUT 20 station from Varian (Walnut Creek, USA). 2.3. Wine sensory analysis The sensory panel was composed of 6 females and 2 males, 23 40 years of age, all of them belonging to the laboratory staff Fig. 1. Graph of the mean sensory ratings MF(%) of the four wines studied. Notations * and ** indicate significance at p < 0.1 and p < 0.05, respectively. and with a long experience in sensory analysis. Five specific 1 h training sessions were carried out. In the first one, judges generated descriptive terms for the Madeira wines. In sessions two and three, different aroma standards were presented and discussed by the panel. From these discussions, the seven aroma terms shown in Fig. 1 were selected for further descriptive analysis. In training sessions fourth and fifth, panelists scored the intensity of each attribute using a seven-point scale (0 = nondetected, 1 = weak, hardly recognizable note, 2 = clear but not intense note, 3 = intense note), half values were allowed but did not bear any description. After the training period, wine samples were evaluated in duplicate along two formal sessions (four samples per session). In all cases, wines (20 ml at 20 C) were presented in coded, black tulip-shaped wine glasses covered by glass Petri dishes. Samples were presented in a random order. The data processed was a mixture of intensity and frequency of detection (what we labeled as modified frequency, MF), which was calculated with the formula proposed by Dravnieks [7]: MF(%) = F(%) I(%), where F(%) is the detection frequency of an aromatic attribute expressed as percentage, and I(%) is the average intensity expressed as percentage of the maximum intensity. Descriptive analysis data was analyzed by χ 2 analysis using Microsoft Excel for Windows. 2.4. GC O analysis Preparation of wine extracts: The volatiles of the wine were collected using a purge-and-trap system. The trap was formed by a standard polypropylene solid phase extraction (SPE) tube (0.8 cm internal diameter, 3 ml internal volume) packed with 400 mg of LiChrolut EN resins. Such resins were selected

E. Campo et al. / Analytica Chimica Acta xxx (2005) xxx xxx 3 because of their excellent ability to extract aroma compounds [8]. The bed was washed with 20 ml of dichloromethane and dried by letting air pass through (negative pressure of 0.6 bar, 10 min). The tube was placed on the top of a bubbler flask containing a mixture of 80 ml of wine and 20 ml of artificial saliva [9]. The mixture was continuously stirred with a magnetic stir bar and kept at a constant temperature of 37 C by immersion in a water bath. A controlled stream of nitrogen (100 ml min 1 ) was passed through the sample during 200 min. Volatile wine constituents released in the headspace were trapped in the cartridge containing the sorbent and were further eluted with 3.2 ml of dichloromethane. The extract was kept at 30 C for 2 h to eliminate any water content by freezing and further decantation. After this, the extract was concentrated under a stream of pure N 2 to a final volume of 200 L. Sniffings were carried out in a Thermo 8000 series GC equipped with a FID and a sniffing port (ODO-1 from SGE) connected by a flow splitter to the column exit. The column used was a DB-WAX from J&W (Folsom, CA, USA), 30 m 0.32 mm with 0.5 m film thickness. The carrier was H 2 at 3 ml min 1. One microlitre was injected in splitless mode, being 1 min the splitless time. Injector and detector were both kept at 250 C. The temperature program was the following: 40 C for 5 min, then raised at 4 C min 1 up to 100 C and at 6 C min 1 up to 200 C. To prevent condensation of high-boiling compounds on the sniffing port, this was heated sequentially using a laboratorymade rheostat. A panel of eight judges, six women and two men, carried out the sniffings of the extracts. Sniffing time was approximately 30 min and each judge carried out one session per day. The panelists were asked to rate the intensity of the eluted odor using a seven-point category scale (0 = not detected, 1 = weak, hardly recognizable odor, 2 = clear but not intense odor, 3 = intense odor), half values being allowed. The quantitative ability of this technique has been already proved [5,6]. On this occasion, as some of the odorants in these extracts were much diluted, the olfactometric signal finally processed was not the mean of the olfactometric scores given by the different sniffers, but the modified frequency (%MF), calculated with the formula previously given. The identification of the odorants was carried out by comparison of their odors, chromatographic retention index in both DB-WAX and DB-5 columns and MS spectra with those of pure reference compounds. 2.5. Quantitative analysis 2.5.1. Major compounds (microextraction and GC FID analysis) Quantitative analysis of major compounds was carried out using the method proposed and validated by Ortega et al. [10]. In accordance with this method, 3 ml of wine and 7 ml of water were salted with 4.5 g of ammonium sulfate and extracted with 0.2 ml of dichloromethane. The extract was then analyzed by GC with FID detection using the conditions described elsewhere [10]. Quantitative data were obtained by interpolation of relative peak areas in the calibration graphs built by the analysis of synthetic wines containing known amounts of the analytes. 2-Butanol, 4-methyl-2-pentanol, 4-hydroxy-4-methyl- 2-pentanone, and 2-octanol were used as internal standards and for quality control purposes. 2.5.2. Minor compounds (SPE and GC Ion Trap MS analysis) This analysis was carried out using the method proposed and validated by Lopez et al. [8]. In accordance with the method, 50 ml of wine, containing 25 L of BHA solution and 75 L of a surrogated standards solution (surrogates were isopropyl propanoate, 3-octanone, heptanoic acid and damascone), was passed through a 200 mg LiChrolut EN cartridge at about 2 ml min 1. The sorbent was dried by letting air pass through ( 0.6 bar, 10 min). Analytes were recovered by elution with 1.3 ml of dichloromethane. An internal standard solution (2-octanol and 4-hydroxy-4-methyl-2-pentanone in dichloromethane) was added to the eluted sample. The extract was then analyzed by GC with Ion Trap MS detection under the conditions described in Ref. [8]. 2.5.3. Sotolon (4,5-dimethyl-3-hydroxy-2(5H)-furanone) (SPE and GC Ion Trap MS analysis) This analysis was carried out using the method proposed and validated in Ref. [11]. In accordance with the method, 50 ml of wine (to which 7.5 g of ammonium sulfate have been previously added) were loaded into a SPE bed formed by 800 mg of LiChrolut EN resins packed in a 6 ml filtration tube from Supelco (Madrid, Spain). The bed was washed with 5 ml of water first, then dried, and finally washed with 15 ml of a mixture pentane/dichloromethane (20/1). Analytes were eluted with 6 ml of dichloromethane and this volume was spiked with 50 L of the internal standard solution (67 mg L 1 of 2-octanol in dichloromethane). This volume was concentrated to 100 L by evaporation in a centrifuge tube heated at 45 C and analyzed by GC Ion Trap MS under the conditions described in Ref. [11]. 3. Results and discussion 3.1. Wine sensory analysis The aromatic characteristics of the four Madeira wines considered in this study were described by a sensory panel using seven different previously agreed sensory descriptors. The results of the sensory analysis can be seen in Fig. 1. As shown in the figure, the aroma of these wines is described as maderized, toasty, spicy, lacquer, dried fruit, candy, and nutty. Spicy was the term with lowest scores, while candy, toasty, maderized and dried fruits reached the highest marks. The scores of the four different wines for the terms dried fruits and toasty were rather homogeneous, which suggests that are generic characteristics of Madeira wines. The most discriminative terms are maderized, candy and lacquer, as χ 2 analysis shows. The wine from Boal is the least candy and most lacquer; wine from Sercial is the opposite: the most candy and the least lacquer; wine from Malvazia is the least maderized; the wine from Verdelho has nearly average scores in all cases.

4 E. Campo et al. / Analytica Chimica Acta xxx (2005) xxx xxx 3.2. Specificities of the aroma profile of Madeira wine by quantitative GC O Two complementary objectives have been aimed in this section. The first one is to build a hierarchical list of the odorants that conform the aroma of Madeira wines, and the second one is to point out the elements on such list that are specific from Madeira wines and cannot be found on equivalent lists from normal dry wines. In order to meet such goals, a quantitative GC O study has been carried out on extracts obtained from the four Madeira wines and from three monovarietal young white wines elaborated with Malvasia, Boal and Verdello, three of the varietals used in the production of Madeira wines. These varietals could not be obtained from Madeira Island, but from the vicinal Canary Islands (Northern part of Tenerife). Although the climatic and soil conditions from both islands are very similar, it is possible that some specificity linked to the Madeira Island may be missing. However, the comparison between the profiles of the young wines, representatives of the varietal model, and their matured Madeira equivalents should make it possible to have a first idea of the chemical differences relevant to the aroma properties introduced by the traditional processes of estufagem and canteiro. Results from the GC O study are summarized in Table 1. The olfactometric experiment was performed on wine extracts obtained by dynamic headspace on LiChrolut EN resins. The olfactometric evaluation was carried out by a panel of eight trained assessors using a seven point quantitative scale. This strategy provides data of semiquantitative value and makes it possible to identify potentially important aroma compounds in wine [5,6]. More than 90 different odorants were detected in the GC O experiment. However, for the sake of simplicity, those odorants not reaching a maximum GC O score of 30% in any of the wines studied were eliminated from the present study and were considered as noise. After this operation the number of odorants was reduced to 53. Table 1 summarizes the 53 different odor notes detected in the GC O experiment and the mean odor modified frequency value given by the assessors. Data in the table are arranged according to the differences found between Madeira wines and their young white equivalents and have been roughly classified into four main categories: (i) Odorants present exclusively (or at highest levels) in Madeira wines. (ii) Odorants from young wines not present (or present at minimum levels) in Madeira wines. (iii) Odorants common to Madeira and young wines. (iv) Odorants showing miscellaneous behavior. 3.2.1. Odorants present exclusively (or at highest levels) in Madeira wines As shown in the table, wines from Madeira contain maximum levels of a large number of odorants. Up to 28 odorants are found in this category, and 22 out of the 28, were not even detected in the young wines. Some of the odorants present in the list come directly from the wood in which Madeira wines are aged. This is the case of (Z)-whiskylactone and of some volatile phenols (2-methoxyphenol, 2-methoxy-4-vinylphenol and m- cresol). However, the presence of many other odorants from different chemicals families reveals the existence of numerous processes of aroma generation. The presence of hexanal is most likely due to the direct oxidation of hexanol, but the presence of 1-octen-3-one and 3-nonen-2-one suggests that the oxidation of fatty acids is also an active route of aroma formation. Similarly, the presence of sotolon, methional, phenylacetaldehyde and of 2,3-pentanedione may be related to the degradation of amino acids and of sugars, as suggested by different authors [2,12 14]. Another group of odorants seems to be formed by the slow esterification of some of wine organic acids or of organic acids formed by some of the aforementioned processes. This could be the case of ethyl 2-methylpentanoate (temptatively reported by first time in wine) and of ethyl cinnamate and dihydrocinnamate. Some of the odorants listed above are very active odorants which, most likely, are going to play an outstanding role in the aroma of Madeira wine. In particular, compounds such as phenylacetaldehyde and sotolon has been previously reported to contribute, respectively, to the old-wood notes [15] and to the typical aged aroma [14,16 19] of different wines. A point that should be remarked is the fact that the list contains 13 unknown odorants, 11 of which seem to be exclusive of Madeira wines. These odorants could not be identified by the customary strategies followed in this study, and should require specific methods of isolation and characterization. 3.2.2. Odorants from young wines not present (or present at minimum levels) in Madeira wines This category groups the compounds which follow the opposite behavior, i.e., odorants present in the young wines which cannot be found in Madeira wines. Only five compounds are listed in the category, but all of them are very important odorants playing outstanding role in the aroma of young wines. The cases of linalool and of 3-mercaptohexyl acetate are particularly outstanding. Linalool is a very important odorant related to the floral character of some wines, while 3-mercaptohexyl acetate is related to the tropical fruit character of some wines, particularly varietals from Verdello [6]. None of the two odorants was even detected in Madeira wines. These two compounds and 2- methyl-3-furanthiol can be easily degraded by oxidation. The null or lowest levels of phenylethyl and isoamyl acetates may be due, on the contrary, to the natural hydrolysis of the esters. The presence of the latter in the Sercial sample may be related to the major amount of sugar fermented in this dry wine. 3.2.3. Common odorants to Madeira and young wines Here compounds with a GC O profile similar to both types of wines are compiled. Most of the compounds found in this group are generic wine odorants, mostly produced as by-products of alcoholic fermentation, such as fatty acid ethyl esters, fusel alcohols, fatty acids or diacetyl. 3.2.4. Odorants showing miscellaneous behavior This category groups compounds following unclear trends. In the case of methoxypyrazines, the isopropyl member was found

E. Campo et al. / Analytica Chimica Acta xxx (2005) xxx xxx 5 Table 1 Odorants found by GC O in Madeira wines elaborated with traditional white grape varieties (Malvazia, Verdelho, Boal and Sercial) and in young wines elaborated with Malvasia, Boal and Verdello); gas chromatographic retention data, olfactory description, chemical identity and modified frequency percentage (%MF) LRI DB-Wax Odor description Identity MAL F MAL Y BOA F BOA Y VER F VER Y SER F (i) Odorants present exclusively (or at highest levels) in Madeira wines 971 Fusel, rancid 3-Methylbutanal a 53 47 68 66 987 Solvent ni b 32 20 18 56 1057 Garlic ni b 65 38 38 35 1075 Butter 2,3-Pentanedione a 50 50 25 29 1092 Grass Hexanal a 56 10 14 1145 Fruity Ethyl 2-methylpentanoate a 65 26 29 24 47 29 1188 Fruity ni b 20 29 29 40 1280 Wet ni b 53 35 20 1307 Mushroom 1-octen-3-one a 53 38 47 14 6 32 1321 Fruity ni b 47 29 10 41 1377 Metallic ni b 69 10 1425 Fruity ni b 82 59 83 22 91 1460 Potato Methional a 18 35 48 25 1511 Rancid, wet 3-Nonen-2-one a 59 68 32 38 1524 Grass ni b 38 20 56 32 1544 Fruity ni b 50 14 1660 Green, honey Phenylacetaldehyde c 69 76 88 60 1708 Peppermint ni b 43 18 47 34 1737 Honey ni b 57 10 35 52 38 1794 Roses ni b 68 50 69 71 1815 Wet ni b 68 10 25 1872 Smoky 2-Methoxyphenol c 82 59 68 82 1895 Flowery Ethyl dihydrocinnamate c 43 26 61 17 67 72 1974 Coconut (Z)-whiskylactone c 59 43 80 61 2107 Leather m-cresol c 53 64 56 21 76 2150 Flowery Ethyl cinnamate c 20 35 35 2209 Bitumen 2-Methoxy-4-vinylphenol c 52 89 84 59 2225 Spicy Sotolon c 61 64 56 87 (ii) Odorants from young wines not present (or present at minimum levels) in Madeira wines 1134 Banana Isoamyl acetate c 88 19 83 35 84 61 1317 Meaty 2-Methyl-3-furanthiol a 42 1560 Flowery Linalool c 66 31 1735 Box tree 3-Mercaptohexyl acetate a 12 62 1825 Roses Phenylethyl acetate c 43 31 28 (iii) Odorants common to Madeira and young wines 965 Fruity ni b 38 36 71 73 79 48 79 1003 Butter 2,3-Butanodione c 89 71 79 65 87 78 91 1029 Solvent Isobutyl acetate c 56 49 65 36 48 52 35 1050 Fruity Ethyl butyrate c 87 76 89 83 78 79 89 1064 Fruity Ethyl 2-methylbutyrate c 68 71 79 75 79 60 87 1079 Fruity Ethyl 3-methylbutyrate c 73 73 89 82 84 75 89 1117 Bitter Isobutanol c 16 53 32 47 41 35 71 1201 Fruity ni b 76 20 59 41 60 36 32 1224 Fusel Isoamyl alcohol c 91 90 96 93 91 80 89 1244 Fruity Ethyl hexanoate c 79 88 76 87 75 82 82 1397 Grass (Z)-3-hexenol c 76 58 47 41 61 31 18 1455 Vinegar Acetic acid c 61 68 88 47 84 54 89 1638 Cheese Butyric acid c 41 19 32 33 53 7 60 1679 Cheese 2-/3-Methylbutyric acid c 84 61 79 65 83 53 83 1829 Baked apple -Damascenone c 66 66 60 78 60 68 61 1929 Roses -Phenylethanol c 59 38 43 36 64 44 64 (iv) Odorants showing miscellaneous behavior 1440 Pepper 3-Isopropyl-2-methoxypyrazine a 65 40 18 55 29 1537 Pepper 3-Isobutyl-2-methoxypyrazine a 53 1627 Toasty 2-Acetylpyrazine a 12 34 49 1438 Toasty 2-Furfurylthiol a 76 : not detected odorant. Abbreviations LRI: linear retention index, MAL: Malvasia, VER: Verdelho, SER: Sercial, BOA: Boal, F : fortified Madeira wines, Y : young dry wines. a Identification based on coincidence of chromatographic retention data and on the similarity of odor with standards. The compound did not produce any clear signal in the mass spectrometer because of its low concentration. b Not identified compounds. c Identification based on coincidence of gas chromatographic retention and mass spectrometric data with those of the pure compounds available in the lab.

6 E. Campo et al. / Analytica Chimica Acta xxx (2005) xxx xxx in Madeira wines, but not the isobutyl. The powerful odorant 2- furfurylthiol was detected only in the Madeira wine made with Sercial. A fifth category, grouping compounds whose presence would have been expected, should be also considered in this discussion. Furfural, 5-methylfurfural, 5-hydroxymethylfurfural and 5-ethoxymethylfurfural were not detected in the GC O experiment, in spite of the fact that these compounds are formed to a large extent in the maturation of sweet wines [2,12,20]. This result suggest that these compounds, even if are quantitatively important, are not relevant to the aroma of Madeira wine. 3.3. Ranking of odorants by GC O Table 2 compares the hierarchical lists of odorants obtained for Madeira wines and for the set of equivalent young white wines. Odorants are ranked according to the maximum %MF reached in the experiment. In order to simplify, only odorants with %MF higher than 50% are listed. A first and obvious difference is the number of odorants present in each group. While 41 odorants are found in Madeira wines, only 17 main odorants form the aroma of the young wines. A second question is that while all the odorants from young wines are known, there are quite a large number of unknowns in the Madeira wines. Some of these unknowns may play an outstanding role on the aroma of these wines, particularly the unknown 1425. A third question is that, leaving aside pure varietal compounds, such as linalool, 3-mercaptohexyl acetate and methoxypyrazines, the rest of odorants in the list of young wines are common to both lists. All these compounds have been extensively reported as the base of wine aroma by different authors [21,22]. The table points out that the aroma profile of Madeira wine is much more complex than and very different to that of young white wines. Main differences are the absence of varietal compounds in Madeira wines, and the presence of many other intense odorants with diverse aroma characteristics. 3.4. Chemical quantitation: odor active odorants Table 3 shows the concentrations, normalized by their corresponding threshold values, of 68 odorants found the fortified Madeira wines. Compounds in the table are ranked according to the maximum odor activity values (OAVs) reached in Madeira wines. Altogether there are 33 odorants that can reach concentrations above their odor thresholds in this set of wines, although given the number of important unidentified odorants, this figure will be probably much higher. Data in the table confirm most of the results obtained in the olfactometric study, and in fact support the usefulness and validity of the quantitative GC O approach used in this work. It can be seen that nearly all the compounds with high OAV had also high GC O scores (bold type in Table 3). The single exceptions to this observation are isobutanol, m-cresol and isobutyl acetate. Due to some reason, these compounds seem to be more easily detected by GC O than by the normal orthonasal olfaction from hydroalcoholic media (standard procedure for estimating odor thresholds). It can be hypothesized that the presence of ethanol difficult the detection Table 2 Ranking of odorants in fortified Madeira wines and in young wines (only those odorants reaching a value of %MF higher than 50, in at least one wine of each family, are considered) Identity Fortified wines Isoamyl alcohol 96 2,3-Butanedione 91 ni 1425 91 Ethyl butyrate 89 Ethyl 3-methylbutyrate 89 Acetic acid 89 2-Methoxy-4-vinylphenol 89 Phenylacetaldehyde 88 Ethyl 2-methylbutyrate 87 Sotolon 87 2-/3-Methylbutyric acid 84 Ethyl hexanoate 82 2-Methoxyphenol 82 (Z)-whiskylactone 80 ni 965 79 m-cresol 76 ni 1201 76 (Z)-3-hexenol 76 2-Furfurylthiol 76 Ethyl dihydrocinnamate 72 ni 1794 71 Isobutanol 71 ni 1377 69 3-Methylbutanal 68 3-Nonen-2-one 68 ni 1815 68 -Damascenone 66 ni 1057 65 Isobutyl acetate 65 Ethyl 2-methylpentanoate 65 3-Isopropyl-2-methoxypyrazine 65 -Phenylethanol 64 Isoamyl acetate 61 Butyric acid 60 ni 1737 57 ni 1524 56 ni 987 56 Hexanal 56 1-Octen-3-one 53 ni 1280 53 2,3-Pentanedione 50 Young wines Isoamyl alcohol 93 Isoamyl acetate 88 Ethyl hexanoate 88 Ethyl butyrate 83 Ethyl 3-methylbutyrate 82 -Damascenone 78 2,3-Butanedione 78 Ethyl 2-methylbutyrate 75 Acetic acid 68 Linalool 66 2-/3-Methylbutyric acid 65 3-Mercaptohexyl acetate 62 (Z)-3-hexenol 58 Isobutyl acetate 54 3-Isopropyl-2-methoxypyrazine 54 3-Isobutyl-2-methoxypyrazine 53 Isobutanol 53 Maximum %MF

E. Campo et al. / Analytica Chimica Acta xxx (2005) xxx xxx 7 Table 3 Odor activity values (OAVs) of odorants found in the four Madeira wines studied and odor thresholds Identity OAV max MAL BOA VER SER Odor threshold a ( gl 1 ) Phenylacetaldehyde 133 101 126 133 45 1 b 3-Methylbutyric acid 49 37 29 34 49 33 [23] Ethyl 3-methylbutyrate 48 35 27 31 48 3 [23] Acetaldehyde 37 27 30 36 16 500 [24] Butyric acid 32 20 18 11 32 173 [23] Decanoic acid 31 15 6.1 8.5 31 1000 [23] -Damascenone 26 19 26 16 14 0.05 [24] Ethyl hexanoate 23 15 10 12 23 14 [23] (Z)-whiskylactone 7.9 6.85 5.3 7.4 7.9 67 [25] 2-Methoxy-4-vinylphenol 7.9 6.2 5.9 4.7 7.9 40 [24] Isoamyl acetate 7.3 4.3 2.9 2.8 7.3 30 [23] -Phenylethanol 6.7 3.6 1.7 2.5 6.7 14000 [23] Ethyl lactate 6.7 3.5 1.5 2.3 6.7 154636 [26] Ethyl butyrate 6.7 4.9 6.7 3.6 4.5 20 [23] Hexanoic acid 4.8 3.2 2.0 2.9 4.8 420 [23] Ethyl 2-methylbutyrate 4.7 3.2 2.3 2.4 4.7 18 [23] 2,3-Butanedione 2.3 1.7 1.9 1.7 1.5 100 [23] Ethyl dihydrocinnamate 2.3 1.6 2.0 1.4 1.5 1.6 [23] Ethyl octanoate 2.2 1.9 1.58 1.9 2.2 580 [26] Isoeugenol 2.1 1.8 2.2 1.6 1.8 6 b Eugenol 2.1 1.7 1.6 1.3 2.1 6 b Sotolon 2.1 1.8 1.6 1.6 2.1 9 b 2-Methoxyphenol 1.9 1.5 1.4 1.3 1.7 9.5 [23] 2-Methylbutyric acid 1.8 1.1 0.73 0.81 1.8 50 [27] Isoamyl alcohol 1.5 0.96 0.70 0.69 1.5 30000 [24] Propanoic acid 1.4 1.32 1.35 1.46 1.16 8100 [23] Vanillin 1.2 0.84 0.88 1.22 0.89 995 b Ethyl cinnamate 1.09 0.62 0.63 0.65 0.58 1.1 [23] Benzoic acid 1.05 0.91 0.88 1.0 0.80 1000 b Isobutyric acid 1.05 0.58 0.40 0.30 1.05 2300 [23] 4-Ethyl-2-methoxyphenol 1.03 0.82 0.68 0.74 1.01 33 [23] -Butyrolactone 1.02 0.48 0.22 0.22 1.02 35000 b (Z)-3-hexenol 1.01 0.45 0.41 0.46 0.48 400 [23] 3-(Methylthio)propanol 0.88 0.45 0.27 0.20 0.88 1000 [23] 4-Vinylphenol 0.79 0.62 0.66 0.42 0.79 180 [25] -Nonalactone 0.66 0.53 0.43 0.48 0.66 30 [23] Diethyl succinate 0.53 0.31 0.14 0.25 0.53 200000 [26] Hexyl acetate 0.47 0.15 0.17 0.16 0.13 1500 [26] Isobutanol 0.43 0.27 0.20 0.19 0.43 40000 [23] Acetovanillone 0.38 0.25 0.22 0.19 0.35 1000 b 4-Propyl-2-methoxyphenol 0.30 0.26 0.25 0.30 0.23 10 b Phenylethyl acetate 0.30 0.13 0.09 0.01 0.30 250 [24] Ethyl decanoate 0.27 0.11 0.12 0.11 0.10 200 [23] Phenylacetic acid 0.24 0.20 0.17 0.20 0.24 1000 [28] o-cresol 0.19 0.15 0.19 0.13 0.14 31 [8] 1-Hexanol 0.19 0.16 0.15 0.14 0.19 8000 [23] Ethyl 3-hydroxybutyrate 0.17 0.15 0.16 0.17 0.11 20000 b Ethyl vanillate 0.16 0.13 0.11 0.13 0.16 990 [8] m-cresol 0.15 0.13 0.15 0.11 0.14 68 [23] 4-Ethylphenol 0.14 0.12 0.10 0.14 0.13 440 [25] Furfural 0.12 0.07 0.08 0.08 0.07 14100 [23] Acetoine 0.08 0.06 0.08 0.08 0.02 150000 [26] -Decalactone 0.07 0.05 0.04 0.04 0.07 88 [23] Isobutyl acetate 0.07 0.03 0.01 0.02 0.07 1600 [21] -Decalactone 0.06 0.04 0.03 0.03 0.06 386 [23] 2,6-Dimethoxyphenol 0.05 0.03 0.03 0.03 0.04 570 [8] -Terpineol 0.04 0.02 0.04 0.00 0.03 250 [23] Butyl acetate 0.03 0.02 0.01 0.03 0.01 1800 [26] Siringaldehyde 0.02 0.01 0.01 0.02 0.02 50000 [27] (E)-whiskylactone 0.02 0.01 0.00 0.02 0.00 790 [25] -Octalactone 0.01 0.01 0.01 0.01 0.01 400 [27] 4-Allyl-2,6-dimethoxyphenol 0.01 0.01 0.01 0.01 0.00 1200 [27] Methyl vanillate 0.01 0.01 0.01 0.01 0.01 3000 [8] Ethyl furoate 0.01 0.01 0.01 0.01 0.01 16000 [23]

8 E. Campo et al. / Analytica Chimica Acta xxx (2005) xxx xxx Table 3 (Continued ) Identity OAV max MAL BOA VER SER Odor threshold a ( gl 1 ) 5-Methylfurfural 0.01 0.01 0.01 0.01 0.01 20000 [26] Linalool 0.00 0.00 0.00 0.00 0.00 25 [23] Geraniol 0.00 0.00 0.00 0.00 0.00 20 [23] -Citronelol 0.00 0.00 0.00 0.00 0.00 100 [26] Bold letters indicate compounds present in the GC O ranking shown in Table 2. a Reference from which the value has been taken is given in parentheses. In Refs. [21,8] the matrix was a 10% water/ethanol solution at ph 3.2; in Ref. [23] the matrix was a 11% water/ethanol solution containing 7 g L 1 glycerol, 5 g L 1 tartaric acid, ph adjusted to 3.4 with 1 M NaOH; in Ref. [26] thresholds were calculated in wine. In Ref. [24] the mixture was 10% in ethanol, in Ref. [25] the matrix was a synthetic wine containing 12% ethanol, 8 g L 1 glycerol and different salts. In Ref. [27] the matrix was water. b Calculated in the laboratory; orthonasal thresholds were calculated in a 10% water/ethanol mixture containing 5 g L 1 of tartaric acid at ph 3.2. of these compounds. Something similar could happen to isoamyl alcohol, which is ranked quite low in Table 3, while is always ranked in the first places in GC O lists. Another important difference between lists in Tables 2 and 3 is that acetaldehyde, decanoic acid and ethyl lactate, ranked quite high in Table 3, were not detected by GC O. The case of acetaldehyde and ethyl lactate, both extremely polar and small odorants, is probably due to a lack of retention of these compounds in the trap. Apart from these observations, Table 3 confirms that monoterpenols are definitively not important odorants in Madeira wines and that phenylacetaldehyde, sotolon, (Z)- whiskylactone and some volatile phenols from wood are important odor active compounds in Madeira wines. 4. Conclusions The study presented here has shown that Madeira wines elaborated with the traditional white grape varieties; Malvazia, Boal, Verdelho and Sercial, possess a large number of odorants detectable in the olfactometric studies. The GC O profile of Madeira wines lacks varietal compounds such as terpenols or cystein-derived thiols, is rich in sotolon, phenylacetaldehyde and wood extractable aromas, and contains a large number of intense odorants not identified which were not even detected in the corresponding young wines. In spite of some limitations, the quantitative GC O approach used in the study arises as a valid tool for determining the existence of intense odorants of wine. Acknowledgements This work has been funded by the Spanish government, project AGL 2001-2486, AGL-2004-06060ALI and by the European project Interreg IIIB, CARVINMAC. The authors gratefully thank Madeira Wine Company (Funchal-Madeira, Portugal) and Bodegas Viñátigo (Tenerife, Spain) who kindly provided wine samples. References [1] J.M.F. Nogueira, A.M.D. Nascimento, J. Agric. Food Chem. 47 (1999) 566. [2] J.S. Camara, J.C. Marques, M.A. Alves, A.C.S. Ferreira, J. Agric. Food Chem. 52 (2004) 6765. [3] J.S. Camara, P. Herbert, J.C. Marques, M.A. Alves, Anal. Chim. Acta 513 (2004) 203. [4] R.F. Alves, A.M.D. Nascimento, J.M.F. Nogueira, Anal. Chim. Acta 546 (2005) 11. [5] V. Ferreira, J. Pet ka, M. Aznar, J. Cacho, J. Chromatogr. A 1002 (2003) 169. [6] E. Campo, V. Ferreira, J. Cacho, J. Agric. Food Chem. 53 (2005) 5682. [7] A. Dravnieks, Atlas of Odor Character Profiles, ASTM Data Series 61, Philadelphia, 1985. [8] R. Lopez, M. Aznar, J. Cacho, V. Ferreira, J. Chromatogr. A 966 (2002) 167. [9] D.D. Roberts, T.E. Acree, J. Agric. Food Chem. 43 (1995) 2179. [10] C. Ortega, R. Lopez, J. Cacho, V. Ferreira, J. Chromatogr. A 923 (2001) 205. [11] V. Ferreira, I. Jarauta, R. Lopez, J. Cacho, J. Chromatogr. A 1010 (2003) 95. [12] I. Cutzach, P. Chatonnet, D. Dubourdieu, J. Agric. Food Chem. 47 (1999) 2837. [13] A. Escudero, P. Hernandez-Orte, J. Cacho, V. Ferreira, J. Agric. Food Chem. 48 (2000) 4268. [14] A.C.S. Ferreira, J.C. Barbe, A. Bertrand, J. Agric. Food Chem. 51 (2003) 4356. [15] M. Aznar, R. Lopez, J. Cacho, V. Ferreira, J. Agric. Food Chem. 51 (2003) 2700. [16] T.T. Pham, E. Guichard, P. Schlich, C. Charpentier, J. Agric. Food Chem. 43 (1995) 2616. [17] B. Martin, P.X. Etievant, J.L. Le Quere, P. Schlich, J. Agric. Food Chem. 40 (1992) 475. [18] E. Guichard, P.X. Etievant, R. Henry, A. Mosandl, Z. Lebensm.-Unters. Forsch. 195 (1992) 540. [19] M. Masuda, E. Okawa, K. Nishimura, H. Yunome, Agric. Biol. Chem. 48 (1984) 2707. [20] I. Cutzach, P. Chatonnet, D. Dubourdieu, J. Agric. Food Chem. 48 (2000) 2340. [21] V. Ferreira, N. Ortin, A. Escudero, R. Lopez, J. Cacho, J. Agric. Food Chem. 50 (2002) 4048. [22] A. Escudero, B. Gogorza, M.A. Melus, N. Ortin, J. Cacho, V. Ferreira, J. Agric. Food Chem. 52 (2004) 3516. [23] V. Ferreira, R. Lopez, J. Cacho, J. Sci. Food Agric. 80 (11) (2000) 1659. [24] H. Guth, J. Agric. Food Chem. 45 (1997) 3027. [25] J.N. Boidron, P. Chatonnet, M. Pons, Conn. Vigne Vin. 22 (4) (1988) 275. [26] P.X. Etievant, Wine, in: H. Maarse (Ed.), Volatile Compounds in Foods and Beverages, Marcel Dekker, 1991, p. 483. [27] L.J. van Gemer, A.H. Nettenbreijer, Compilation of Odour Threshold Values in Air and Water, TNO, 1977. [28] J.A. Maga, Cereal Sci. Today 18 (1973) 326.