The role of potent thiols in Chardonnay wine aroma

Transcription

1 Capone 38 et Potent al. thiols in Chardonnay wine Australian Journal of Potent Grape thiols and Wine Chardonnay Research 24, wine 38 50, The role of potent thiols in Chardonnay wine aroma D.L. CAPONE, A. BARKER, P.O. WILLIAMSON and I.L. FRANCIS The Australian Wine Research Institute (AWRI), Urrbrae, SA 5064, Australia Corresponding author: Dr Dimitra L. Capone, dimitra.capone@awri.com.au Abstract Background and Aims: Polyfunctional thiols are key aroma compounds in many Sauvignon Blanc wines, but their role in other white cultivars is not clear. Methods and Results: A survey of 106 commercial Australian Chardonnay wines found a high concentration of 3 mercaptohexan 1 ol, 3 mercaptohexyl acetate, benzyl mercaptan and 4 mercapto 4 methylpentan 2 one, with nearly all wines having a concentration of all compounds well above their reported sensory detection threshold, and some having a concentration comparable to that found in highly fruity Sauvignon Blanc wines. Wines were made on a research scale from a set of Chardonnay juices sourced from 16 vineyards across Australia. Sensory descriptive analysis combined with quantitative aroma volatile data revealed that several aroma and flavour attributes were related to the concentration of the thiols. Conclusions: This study provided evidence that substantial flavour in Chardonnay can be contributed by these thiols. Additionally, consumer acceptance data showed that wines with a higher thiol concentration were liked by most consumers. Significance of the Study: This study revealed the importance of polyfunctional thiols in Chardonnay wine. Keywords: Chardonnay, chemical analysis, consumer preference, sensory analysis, varietal thiols Introduction Chardonnay is one of the most commercially important white cultivars grown around the world, being the second most planted in France and the most planted white cultivar in both California and Australia (Robinson et al. 2012, Wine Australia 2016). The Chardonnay wine style can vary widely depending on region of origin and winemaking practices, from strongly flavoured and oak influenced to light and acidic types (Saliba et al. 2013a,b, Gambetta et al. 2014). There have been several studies involving the combination of sensory and chemical compositional data that have assessed the most important aroma compounds in Chardonnay wine (Noble and Ebeler 2002, Lee and Noble 2003, Smyth 2005, Lorrain et al. 2006, Louw et al. 2010, Jaffré et al. 2011). Lee and Noble (2003) found 81 compounds to be odour active in 19 California Chardonnay wines, and partial least squares (PLS) regression revealed the fruity terms were related to 3 methylbutyl acetate, 2 phenylethyl acetate and linalool (Lee and Noble 2006). A further comprehensive study involving the use of two dimensional GC combined with time of flight MS tentatively identified 243 compounds (Welke et al. 2014). Most of the previous studies involved first identifying the volatile aroma compounds followed by quantification, but a limitation of this type of approach is that compounds with a low sensory threshold that are present at low or trace concentration may be missed as it is almost impossible for these compounds to be detected in GC/MS scan runs. The polyfunctional thiols: 3 mercaptohexan 1 ol (3 MH), 3 mercaptohexyl acetate (3 MHA) and 4 mercapto 4 methylpentan 2 one (4 MMP) are one such group of compounds that have an extremely low aroma detection threshold and are present in the nanogram per litre concentration range (Tominaga et al. 1998a). They are present in a wide array of fruits and vegetables; are known as varietal impact compounds in Sauvignon Blanc wines, and their aromas have been described as tropical, box hedge, passionfruit and citrus (Tominaga et al. 1998a). These thiols have previously been difficult to quantify, with one widely employed method involving the use of highly toxic mercury complexes to bind the thiols for the extraction (Darriet et al. 1995) and running the extracts in negative chemical ionisation mode on a GC/MS system. More recently, a simplified method has been developed where the free thiols, including phenylmethanethiol also known as benzyl mercaptan (BM) and (furan 2 yl) methanethiol also referred to as 2 furfurylthiol (FFT), are derivatised and then analysed using HPLC MS/MS (Capone et al. 2015). The thiols 3 MH and 4 MMP are known to be derived from cysteine and glutathione conjugate precursors in the juice, which break down during fermentation due to lyase activity in the yeast to release the free thiols, with 3 MHA transformed from 3 MH by acetyl transferase activity in the yeast (Swiegers et al. 2006). The concentration of these thiols in Sauvignon Blanc had been found to be particularly high, up to ng/l for 3 MH in some New Zealand wines (Lund et al. 2009). In other cultivars, their concentration has generally been lower, for 3 MH up to approximately 970 ng/l in five Riesling wines (Tominaga et al. 2000) and up to 1200 ng/l in 34 Pinot Noir wines (Capone et al. 2015). There have been limited polyfunctional thiol data published for Chardonnay wines, with a set of nine Spanish wines and 12 Australian wines studied (Mateo Vivaracho et al. 2010, Capone et al. 2015). For the Spanish wines, 4 MMP and 3 MHA were determined to be well above aroma threshold values in most of the samples, while 3 MH was present only above the sensory threshold in some wines. From sensory experiments, it was concluded by these authors that 3 MH can be considered unimportant to Chardonnay and that 4 MMP and 3 MHA were probably contributing fruity aroma rather than distinct box hedge or tropical fruit character (Mateo Vivaracho et al. 2010). In contrast, for the Australian Chardonnay wine data (Capone et al. 2015), both 4 MMP and 3 MHA were below sensory threshold, while 3 MH was generally well above. For doi: /ajgw.12294

2 2Capone Potent et al. thiols in Chardonnay wine AustralianPotent Journal thiols of Grape in Chardonnay and Wine Research wine both studies, BM was present at a sensorily important level in most wines. In sensory studies, some Australian Chardonnay wines have been described as having tropical fruit aroma (Soden et al. 2000, Skouroumounis et al. 2005, Torrea et al. 2011), which has been well accepted by consumers (Saliba et al. 2013b), but the compounds responsible for the tropical fruit, citrus/grapefruit or box hedge characters have not been previously defined. The aim of the present study was to assess the sensory significance of the polyfunctional thiols to Chardonnay wines, through a large survey of commercial wines, as well as through a sensory compositional study for wines made using standardised conditions from juices sourced from a wide range of vineyards across Australia. Materials and methods Chemicals and materials Analytical reagents were purchased from Sigma Aldrich (Castle Hill, NSW, Australia) unless otherwise specified. Unlabelled and isotopically labelled standards previously synthesised included 3 MH, [ 2 H 10 ] 3 mercaptohexan 1 ol (d 10 3 MH), 3 MHA, [ 2 H 5 ] 3 mercaptohexyl acetate (d 5 3 MHA), 4 MMP, [ 2 H 10 ] 4 mercapto 4 methylpentan 2 one (d 10 4 MMP) (Howell et al. 2004, Swiegers et al. 2007, Pardon et al. 2008), [ 2 H 5 ] 2 furfurylthiol (d 5 FFT) and [ 2 H 5 ] benzyl mercaptan (d 5 BM) (Capone et al. 2015). [ 2 H 9 ] (R/S) 3 S Glutathionylhexan 1 ol (d 9 Glut 3 MH), [ 2 H 8 ] (R/S) 3 S cysteinylhexan 1 ol (d 8 Cys 3 MH) and 3 S cysteinylglycinehexan 1 ol (CysGly 3 MH) were synthesised and characterised as described by Capone et al. (2011a), Grant Preece et al. (2010) and Pardon et al. (2008), respectively. Juice and wine samples Wine samples for the survey were commercially available and obtained from retail outlets. The majority of the wines purchased were chosen based on retail sales data of the top 50 Chardonnay wines sold in Australia across multiple price points and regions in Australia; the wines were obtained and thiol analysis was completed in February July Juice samples were obtained from 16 different sources across Australia in 2014: Hunter Valley, NSW (HV); two samples from Margaret River, WA (MR1 and MR2); Riverland, SA (RL); McLaren Vale, SA (MV); Padthaway, SA (P); Rutherglen, Vic. (R); two samples from Great Southern, WA (GS1 and GS2); Adelaide Hills, SA (AH); Yarra Valley, Vic. (YV); Orange, NSW (O); Mornington Peninsula, Vic. (MP); Great Western, Vic. (GW); Tumbarumba, NSW (T); and Coal River Valley, Tas. (TAS). A set of standardised criteria was followed for the collection of the Chardonnay juices (~30 L) as follows: 50 kg of hand harvested fruit was collected at a TSS value between 20.2 and 23.4 Brix, the grapes were then crushed/de stemmed soon after being harvested, followed by light pressing. No enzyme addition was used, and mg/l of free sulfur dioxide was added to each juice. The juice samples were transferred into 30 L stainless steel kegs with minimal ullage and were kept at 4 C during transport to the Wine Innovation Cluster winemaking facility located in Adelaide, SA. The standardised winemaking involved the addition of ~250 mg/l Maurivin PDM yeast (AB Mauri, Sydney, NSW, Australia) and fermented at C. Once fermentation was complete, the wines were racked off gross lees, sulfur dioxide added to 30 mg/l and cold stabilised at 0 C for 21 days. The wines were racked to remove fine lees, sulfur dioxide adjusted as required to at least 30 mg/l and filtered through a Z6 pad membrane prior to bottling into screwcap sealed 375 ml bottles. The wines were stored at 15 C until analysis. Both the volatile composition and the sensory analyses were carried out within a month of each other, approximately 2 months after bottling. The basic compositional data and the codes for the juice and wine samples determined at bottling are provided in Table 1. Chemical analysis of wine volatiles All volatiles were quantified by stable isotope dilution analysis using previously published methods as briefly detailed below. Thiols. Samples were prepared as described by Capone et al. (2015), by adding to an aliquot of wine (20 ml) the labelled standards, d 10 4 MMP, d 10 3 MH, d 5 3 MHA, d 5 2 furfurylthiol Table 1. Basic compositional data for the 16 Chardonnay juice samples collected from the 2014 vintage across Australia along with that of the finished wines. Region Juice Wine Code ph TA (g/l) TSS ( Brix) Alc. (% v/v) ph TA (g/l) G+F (g/l) VA (g/l) FSO 2 (mg/l) TSO 2 (mg/l) Hunter Valley, NSW HV Margaret River, WA MR Margaret River, WA MR Riverland, SA RL McLaren Vale, SA MV < Padthaway, SA P Rutherglen, VIC R Great Southern, WA GS Great Southern, WA GS Adelaide Hills, SA AH < Yarra Valley, Vic. YV Orange, NSW O Mornington Peninsula, Vic. MP Great Western, Vic. GW Tumbarumba, NSW T Coal River Valley, Tas. TAS <

3 Capone 40 et Potent al. thiols in Chardonnay wine Australian Journal of Potent Grape thiols and Wine Chardonnay Research 24, wine 38 50, and d 5 BM,eachwithafinal concentration of 500 ng/l. An addition of ethylenediaminetetra acetic acid disodium salt (20 mg), 50% acetaldehyde (80 μl) and freshly thawed 4,4 dithiodipyridine reagent (10 mmol, 200 μl) was then added. After 30 min, the sample was passed through a Bond Elut C18 cartridge (6 ml, 500 mg, Agilent Technologies, Forest Hill, Vic., Australia), with conditions as previously detailed. The eluate was collected, concentrated and reconstituted with 10% ethanol (200 μl) and run on an Agilent 1200 HPLC (Agilent Technologies) using a mm i.d., 5 μm, 100 Å Alltima C18 column (Grace Davison Discovery Sciences, Rowville, Vic., Australia) coupled to an Applied Biosystems 4000 QTrap hybrid tandem mass spectrometer (Applied Biosystems, Foster City, CA, USA) in electrospray ionisation mode as described in Capone et al. (2015). Thiol precursors. Samples were prepared for the analysis of thiol precursors using a method as described previously (Capone et al. 2010). In summary, the method involved taking an aliquot of 9.9 ml of sample with the addition of deuteriumlabelled d 8 Cys 3 MH and d 9 Glut 3 MH (50 μg/l final concentration) and passing the sample through a 6 ml, 500 mg Strata SDBL cartridge (Phenomenex, Lane Cove, NSW, Australia) with conditions described in the paper. The extract was run on an HPLC MS/MS using the same column as for the analysis of varietal thiols. Oxidation related compounds. Benzaldehyde, eugenol, furaneol, furfural, (E) 2 heptenal, hexanal, (E) 2 hexenal, homofuraneol, maltol, methional, methionol, 5 methylfurfural, 2 methylpropanal, 3 methylbutanal, (E) 2 nonenal, (E) 2 octenal, 2 phenylacetaldehyde and sotolon were analysed as described previously (Mayr et al. 2015) using GC combined with a triple quadrupole MS (GC/MS/MS) (Agilent Technologies). Fermentation volatiles. Fermentation volatiles, including various ethyl esters, alcohols, acids and acetates, were analysed with solid phase microextraction (SPME) using GC/MS (Agilent Technologies) as described previously (Siebert et al. 2005). C6 compounds. Hexan 1 ol, (E) 2 hexenal, (Z) 3 hexen 1 ol and (E) 2 hexen 1 ol were quantified as described by Capone et al. (2012a) also using SPME GC/MS. Monoterpenes. Linalool, α terpineol, nerol and geraniol were analysed as described by Pedersen et al. (2003). Norisoprenoids. β Damascenone and β ionone were analysed as described by Ugliano et al. (2008). Both of these methods involved solvent extraction, and the extracts were run on a GC/MS. A SPME GC/MS method was employed for the analysis of trans ethyl cinnamate (Smyth 2005). 4 Alkyl substituted γ lactones. γ Octalactone, γ nonalactone, γ decalactone and γ dodecalactone were quantified as described by Cooke et al. (2009a,b). This method employed solid phase extraction, and the subsequent concentrated solvent extract was injected in liquid mode on a GC/MS. Sensory descriptive analysis A panel of 11 assessors (three males and eight females) with an average age of 54 years (SD = 8.9) was convened for this study, all of whom are part of the trained descriptive analysis panel of The Australian Wine Research Institute (AWRI) with extensive wine sensory descriptive analysis experience. A consensusbased descriptive methodology was used as described by Bindon et al. (2014). The attributes, definitions/synonyms and reference standards used in the study are shown in Table 2. Samples were assessed over 4 days of formal sessions, in which wines were presented to panellists in 30 ml aliquots in three digit coded, covered, ISO standard wine glasses at C. Samples were assessed in isolated booths under daylight type lighting, with randomised block design and a randomised presentation order across assessors. Assessors were presented with five trays of three wines for the first two sessions and four trays of three wines for the last two sessions in a modified Williams Latin square incomplete random block design generated by Fizz sensory acquisition software (version 2.47B, Biosystèmes, Couternon, France). All samples were expectorated. The assessors were forced to have a 45 s rest between samples and a minimum of 10 min rest between trays. The wines were assessed in triplicate. The intensity of each attribute was rated using an unstructured 15 cm line scale (coded from 0 to 10, with indented anchor points of low and high placed at 10 and 90%, respectively). Data were acquired using Fizz sensory software. Panel performance was assessed using Fizz, Senstools (OP&P, Utrecht, the Netherlands) and PanelCheck open source software (panelcheck.com), to determine the degree of agreement with the panel mean and degree of discrimination across samples for individual assessors. All assessors were found to be performing to an acceptable standard according to limits based on long standing experience with panel performance data. Consumer test The consumer hedonic test was carried out in Adelaide, Australia, with 156 regular white wine consumers (males and females from the local community who drink white wine at least once every 2 weeks and are over 18 years of age) (see Table 3 for consumer demographics information). Consumers attended a single session to evaluate six Chardonnay wines, selected from the descriptive analysis as having a different level of passionfruit, pineapple and box hedge characters, and were given an incentive voucher on their completion. Upon arrival, consumers signed an informed consent form, completed a demographic questionnaire and were briefed on the tasks that they were to complete during a 1 h session. Respondents evaluated the wines one at a time with a forced 2 min break between wines and were encouraged to drink water between samples. Respondents could choose if they wanted to drink the samples or expectorate, a sink was provided in each booth. Wines (30 ml) were presented at 8 12 C in a randomised complete block design in three digit coded ISO wine tasting glasses in isolated sensory booths under daylight type lighting. Consumers rated each wine for overall liking on a nine point hedonic scale (dislike extremely to like extremely) and purchase intent on a five point scale (definitely would not buy to definitely would buy). After the tasting, consumers answered several questions regarding wine usage and attitudes. Data analysis The sensory data were subjected to ANOVA using Fizz. The effects of wine, judge (random effect), replicate and the corresponding two way interactions were assessed. Principal component analysis (PCA) was conducted using XLSTAT 2015 (version 2.02, Addinsoft, New York, NY, USA) on the mean values averaged over panellists and replicates, using the correlation matrix.

4 4Capone Potent et al. thiols in Chardonnay wine AustralianPotent Journal thiols of Grape in Chardonnay and Wine Research wine Table 2. Attributes, definitions and reference standards evaluated by panellists for the sensory descriptive analysis of the 16 Chardonnay wines. Attribute Definition/synonyms Standard composition Appearance Yellow colour intensity Intensity of the yellow colour Aroma Overall fruit intensity aroma Intensity of the fruit aromas in the sample Passionfruit Intensity of the aroma of passionfruit and guava 1 tsp Canned passionfruit pulp (John West) Pineapple Intensity of the aroma of pineapple 4 2 cm Cubes fresh pineapple and 1 tsp canned pineapple juice (Golden Circle) Melon Intensity of the aroma of melon 3 1 cm Pieces of fresh rock melon and honeydew melon Stonefruit Intensity of the aroma of stonefruits: peach, apricot and nectarine, both fresh and dried 1 tbsp Canned apricots (Goulburn Valley), 3 1 cm pieces of fresh nectarine Lemon Intensity of the aroma of lemon 1 2 cm Piece of lemon and lemon rind Lime Intensity of the aroma of lime 1 2 cm Piece of lime and lime rind Confection Intensity of the aroma of confection: red lolly, banana lolly, musk, caramel lollies 1 Red lolly (Allens), 1 musk lolly (Life Saver), and 1/2 banana lolly (Foodland), no wine Floral Intensity of the aroma of flowers: violets, blossoms, jasmine, talcum powder and lychee 40 μl of 100 mg/l Linalool, 25 μl of 200 mg/l 2 phenylethanol Green Vegetal Intensity of the aroma of green grass, green leaves, stalks, green beans green capsicum and geranium Intensity of the aroma of various vegetables: cooked vegetables such as onions, asparagus and green beans, water vegetables have been cooked in, drain Freshly picked grass, cm pieces of chopped green beans and 1 cm piece green capsicum, chopped, no wine 0.5 tbsp Water from canned mixed vegetables (Edgell s) and 0.5 tbsp water from canned green bean (Edgell s) Box hedge Intensity of the aroma of box hedge and cat pee Fresh box hedge leaves, no wine Flint Intensity of the aroma of flint, wet stones, metals, 20 μl of 1 mg/l Benzyl mercaptan toast Sweaty/cheesy Intensity of the aroma of sweat, cheese, blue cheese, cheddar cheese, body odour, sour milk, raw meat 100 μl mix of Hexanoic acid and 3 methylbutanoic acid Chemical Intensity of the aroma of chemical, plastic, fly spray, 20 μl of 2,6 Dichlorophenol varnish, shoe polish Pungent Intensity of the aroma and effect of alcohol 4 ml Ethanol Palate Overall fruit intensity flavour Intensity of the fruit flavours in the sample Tropical fruit Intensity of the flavour of tropical fruits: pineapple, passionfruit, melon, mango, guava and lychee Stonefruit Intensity of the flavour of stonefruits: peach, apricot and nectarine Citrus Intensity of the flavour of citrus fruits: lemon, lime, orange and grapefruit Green apple Intensity of the flavour of green apple Green Intensity of the flavour of green stalks, green leaves, grass, green vegetables and herbs such as mint and oregano Sweet Intensity of the taste of sucrose Viscosity Perception of the body, weight, or thickness of the wine in the mouth; low = watery, thin mouth feel; high = oily, thick mouth feel Acid Intensity of acid taste in the mouth including aftertaste Hotness Intensity of alcohol hotness perceived in the mouth, after expectoration and the associated burning sensation; low = warm; high = hot Astringency Drying and mouth puckering sensation in the mouth; low = coating teeth; medium = mouth coating and drying; high = puckering, lasting astringency Bitter Intensity of bitter taste perceived in the mouth, or after expectoration Fruit aftertaste Lingering fruit flavour perceived in the mouth after expectorating

5 Capone 42 et Potent al. thiols in Chardonnay wine Australian Journal of Potent Grape thiols and Wine Chardonnay Research 24, wine 38 50, Table 3. Summary of the demographic data of the consumers who participated in the consumer studies. Overview of consumers Consumer sample, n = 156 (%) Male 55 Female 45 Age >65 14 Education High school diploma 12 College diploma 22 Bachelor degree 31 Postgraduate degree 33 Other 2 Household income per year ($) Less than > How long drinking wine (years) Less than >20 65 Frequency of white wine consumption per week <Once times times 18 >5 times 22 Results and discussion Chardonnay survey The quantification of thiol compounds at near threshold concentration in wine is challenging, due to the exceptionally low concentration of these volatile reactive compounds found. A recently developed LC MS/MS method was applied, where the free thiols are derivatised, allowing quantification of these compounds in a relatively large sample set. One hundred and six commercial Australian Chardonnay wines were analysed for thiol compounds and 3 MH precursors. The vintage of the wines ranged from 2003 to 2013 with 2012 being the median value. The price of the wines ranged from A$4 to A$120 with a median of A$19, with 50% of the wines between A$14 and A$35. The distribution of the thiol data is shown in Figure 1. The concentration of 3 MH ranged from approximately 142 to 2636 ng/l with a median value of 650 ng/l. The sample set studied had a median concentration of 3 MH higher than that on the much smaller number of wines reported by Capone et al. (2015) and by Mateo Vivaracho et al. (2010) (median of 400 ng/l and approximately 30 ng/l, respectively). Compared to the previously reported data, there was a higher minimum value in the present study well above the sensory detection threshold (Tominaga et al. 1998b). It was notable that 11 of the wines examined had a concentration above 1500 ng/l, which is much higher than that previously reported and which resembled the concentration found in Australian and New Zealand Sauvignon Blanc wines (Mateo Vivaracho et al. 2010, Benkwitz et al. 2012, Capone et al. 2012b, 2015). In a few of the samples, 3 MHA was found at high concentration, generally similar to that previously reported (Mateo Vivaracho et al. 2010, Capone et al. 2015), and for this compound, the concentration measured in all of the samples was also present above the aroma detection threshold of 4 ng/l (Tominaga et al. 1996). In all but 22 of the wines, 4 MMP was at a concentration above its aroma detection threshold (0.8 ng/l) (Darriet et al. 1995). A previous sensory study involving addition of these compounds at a specific concentration (Mateo Vivaracho et al. 2010) indicated that a concentration of 3 MH or 3 MHA Two factor ANOVA using XLSTAT 2015 was applied for consumer and wine plus internal preference mapping with the descriptive analysis data as supplementary variables. The raw liking data were subjected to agglomerative hierarchical cluster analysis, with the similarity measured by Pearson correlation coefficient and agglomeration by the unweighted pair group average linkage method. The composition of the clusters was investigated with a chi squared test, testing the independence between the variables and overall liking (P = 0.05) for each cluster. Partial least squares (PLS) regression was conducted using The Unscrambler X (version 10.3, CAMO Process AS, Oslo, Norway), and for the consumer data, the sensory descriptive analysis data were used as x variables and mean liking scores for the consumer clusters as y variables, so that sensory attributes were used as predictors for the liking scores of the total population and each individual cluster. For relating the sensory attributes with the chemical composition, the chemical compositional data were treated as x variables and the sensory data were y variables. All data were standardised with full cross validation. Figure 1. Box plots showing the distribution of 3 mercaptohexan 1 ol (3 MH), 3 mercaptohexyl acetate (3 MHA) and 4 mercapto 4 methylpentan 2 one (4 MMP) (n = 106) and benzyl mercaptan (BM) (n = 80), 3 S cysteinylhexan 1 ol (Cys 3 MH) and 3 S glutathionylhexan 1 ol (Glut 3 MH) (n = 100) from a survey of Australian commercial Chardonnay wines. The median and the range where 50% of the data occur (the box) are shown, with the whiskers corresponding to the lowest and highest values within a range limit. The wines outside the whisker range are marked as outliers*.

6 6Capone Potent et al. thiols in Chardonnay wine AustralianPotent Journal thiols of Grape in Chardonnay and Wine Research wine above 148 or 6.4 ng/l, respectively, contributed fruity aroma to a white wine, while 4 MMP was noted as contributing a fruity and green aroma character from 1 ng/l. Overall, the results show that for Australian commercial Chardonnay wines these three polyfunctional thiols are likely to contribute fruity flavour to most wines. All wines except six had a concentration of BM well above the aroma detection threshold of 0.3 ng/l (Tominaga et al. 2003a), with seven wines being unusually high, above 29 ng/l, similar to the concentration reported for a small set of Chardonnay wines from France (Tominaga et al. 2003a) and above that measured by Capone et al. (2015). This compound is considered to be a contributor to smoky, struck match or struck flint characters in white wine, and in a sensory addition study (Mateo Vivaracho et al. 2010), an addition even as low as 0.7 ng/l gave a toasty, burnt and empyreumatic aroma. The 3 MH precursors (Cys 3 MH and Glut 3 MH) remaining in the Chardonnay wines were also quantified (Figure 1). The concentration of the precursors ranged from not detected to 40 μg/l and up to 180 μg/l for Cys 3 MH and Glut 3 MH, respectively. These concentration values were comparable to those found in Sauvignon Blanc commercial wines (Capone et al. 2010). The substantial quantity remaining in a wine is of interest as the work of Starkenmann et al. (2008) demonstrated that 3 MH can be released in the mouth from precursors through enzymes present in saliva. There were few significant correlations among the compounds, with no significant relationship between the concentration of 4 MMP, 3 MH and 3 MHA; a weak but significant correlation of 3MHA and BM (r = 0.30, P < 0.01, n = 80); and a stronger correlation between Cys 3 MH and Glut 3 MH (r = 0.67, P < 0.001, n = 103), with no significant correlation of the concentration of the precursors in either the wine or the juice with 3 MH as previously reported (Capone et al. 2011b, Pinu et al. 2012). With the relatively large number of wines analysed, links between the vintage of the wine, the retail price and the temperature of the region of origin during the growing season [mean January temperature (MJT)] were evaluated using multiple linear regression of the log transformed concentration. There was a weak but significant association of MJT with 3 MH concentration (P = 0.022, negative coefficient, R 2 adj = 0.03, error df = 102). For 3 MHA, both MJT and vintage were significant variables (P = and 0.031, respectively, MJT negative coefficient and vintage positive, R 2 adj = 0.13, error df = 102). The influence of MJT indicates that the cooler the viticultural region, the more likely the wines will have a higher value of these compounds, while for 3 MHA, the older vintages had generally a lower concentration, as expected based on previous studies that have shown this compound to be susceptible to acid hydrolysis over time in bottle (Herbst Johnstone et al. 2011). A similar model generated for 4 MMP showed vintage and MJT were significant variables, although for 4 MMP MJT was positive and vintage was negative (P = and 0.023, respectively, R 2 adj = 0.05, df = 102), suggesting the warmer regions and older vintages tended to have a higher concentration of this compound. The price and vintage for BM were significant contributors to a model (P = and 0.049, respectively, R 2 adj = 0.51, df = 83). The older vintages and higher priced wines were more likely to have a higher concentration of BM, with MJT tending to be negatively related (P = 0.073). As an addition to this study, thiols in a small number of wines from France were also quantified to compare thiol concentration between Old and New World producers. Three wines were obtained from the Chablis Appellation Contrôlée (AC) (2011, 2014 and 2013, all from different producers), as well as one from Marsannay AC (2010) and a single wine from Hautes Côtes de Nuits AC from Burgundy. The concentration of 3 MH in these wines ranged from 560 to 2229 ng/l, indicating that wines from premium producers in France can also have a substantial concentration of this compound. Similar to the Australian wines, the concentrations of 4 MMP and BM in the French wines were low, up to 2.2 and 8.7 ng/l, respectively, whereas 3 MHA was not detected in any of the French wines, which might reflect the relatively low number of samples analysed. Thiols in Chardonnay wine made using standardised winemaking The thiol compounds in Chardonnay wine were investigated in a second study, in which the wines were made under standardised conditions. This was to ensure that 100% Chardonnay grapes were used, as label integrity regulations in Australia, as well as in other countries, allow the addition of up to 15% of another grape cultivar without this being stated on the bottle label. Accordingly, there may have been a proportion of Sauvignon Blanc or other high thiol cultivars blended into some of the commercial wines examined in the survey. In addition, the commercial wines will have been produced with variation in viticultural and winemaking practices, such as yeast strain choice, oak treatment and use of malolactic fermentation. It was considered important to study wines where winemaking variables were as controlled as possible, and to allow the sensory properties of the wines to be related to the volatile composition. Quantification of the thiol compounds, as well as other targeted volatile aroma compounds, and a sensory descriptive analysis and consumer preference study were carried out on these wines. Juice samples were obtained from hand harvested Chardonnay fruit as previous studies had shown that an increase in the concentration of thiol precursors was observed when the fruit was machine harvested (Allen et al. 2011, Capone and Jeffery 2011). Fruit ripeness is also an important parameter as shown in previous studies (Peyrot des Gachons et al. 2000, Capone et al. 2011b, 2012b, Cerreti et al. 2015); therefore, the Chardonnay grapes were all picked at similar ripeness. Previous studies have shown that press rates and enzyme additions can affect the concentration of polyfunctional thiols (Allen et al. 2011, Capone and Jeffery 2011, Capone et al. 2012a, Maggu et al. 2007). Winemaking was standardised at the same winery and by the same winemaker. For the 16 young wines made from juices sourced from multiple Australian viticultural regions, 71 volatile aroma compounds were quantified using stable isotope dilution analysis methods. Apart from the thiols, fermentation derived esters, acids and acetates; C6 compounds; aliphatic γ lactones; norisoprenoids; monoterpenes; ethyl cinnamate; and a group of compounds related to oxidation were determined (Table 4). The minimum, maximum and median concentration of the volatile aroma compounds is summarised in Table 4, together with the aroma detection threshold and the concentration of these compounds that has been previously reported in Chardonnay wine. The focus of this research study was mainly on the thiol compounds, and it was notable that the compounds 3 MH and 3 MHA (Table 4) had a concentration much higher than that previously reported (Mateo Vivaracho et al. 2010, Capone et al. 2015). Surprisingly, the concentration of these

7 Capone 44 et Potent al. thiols in Chardonnay wine Australian Journal of Potent Grape thiols and Wine Chardonnay Research 24, wine 38 50, Table 4. Summary of the concentration of volatiles in the wines made from 16 Chardonnay juices sourced from across Australia. Compound Code Minimum (μg/ L) Maximum (μg/ L) Median (μg/ L) Aroma detection threshold (μg/l) Previously reported in Chardonnay wine (μg/l) 4 Mercapto 4 methylpentan 2 one (ng/l) 4 MMP (Gambetta et al. 2014) 0 (Capone et al. 2015) 23 (Mateo Vivaracho et al. 2010) 3 Mercaptohexan 1 ol (ng/l) 3 MH (Gambetta et al. 2014) 10 (Mateo Vivaracho et al. 2010) 1368 (Capone et al. 2015) 3 Mercaptohexyl acetate (ng/l) 3 MHA (Gambetta et al. 2014) 0 (Capone et al. 2015) 100 (Mateo Vivaracho et al. 2010) Benzyl mercaptan (ng/l) BM (Gambetta et al. 2014) (Gambetta et al. 2014) 2 Furfurylthiol (ng/l) FFT nd (Gambetta et al. 2014) 14 (Gambetta et al. 2014) Methionol MeTOH (Mayr et al. 2015) (Welke et al. 2014) 3710 (Lee and Noble 2003) Homofuraneol HMF (Mayr et al. 2015) (Lee and Noble 2003) 2 Methylpropanal 2 MPAL (Mayr et al. 2015) na Sotolon SO (Gambetta et al. 2014) 1.1 (Gambetta et al. 2014) 9.11 (Gabrielli et al. 2015) 2 Methylbutanal 2 MBAL (Mayr et al. 2015) nd (Buettner 2004) 3 Methylbutanal 3 MBAL (Mayr et al. 2015) 0.01 (Welke et al. 2014) Hexanal HEXAL (Mayr et al. 2015) 9.71 (Welke et al. 2014) (Flamini et al. 2002) (E) 2 Hexenal 2 HEXEAL (Mayr et al. 2015) (Flamini et al. 2002) Furfural FAL (Gambetta et al. 2014) (Gambetta et al. 2014) 5 Methylfurfural 5 MFAL (Gambetta et al. 2014) Trace 37.5 (Gambetta et al. 2014) Methional METAL (Gambetta et al. 2014) na (E) 2 Octenal OCTAL (Mayr et al. 2015) na Benzaldehyde BENZAL (Mayr et al. 2015) 3.7 (Garcia et al. 2003) (Welke et al. 2014) 2 Phenylacetaldehyde 2 PAA (Mayr et al. 2015) 4 28 (Gambetta et al. 2014) (E) 2 Nonenal 2 NONAL (Mayr et al. 2015) (Flamini et al. 2002) Ethyl acetate EA (Siebert et al. 2005) 38.6 (Lee and Noble 2003) (Louw et al. 2010) Ethyl propanoate EtPro (Gambetta et al. 2014) (Gambetta et al. 2014) Ethyl 2 methylpropanoate Et 2 MePr (Siebert et al. 2005) (Lee and Noble 2003) 2 Methylpropyl acetate 2 MeProAcet (Siebert et al. 2005) (Simpson and Miller 1984) Ethyl butanoate EtBu (Gambetta et al. 2014) (Louw et al. 2010) Ethyl 2 methylbutanoate Et 2 MeBu (Gambetta et al. 2014) (Gambetta et al. 2014) Ethyl 3 methylbutanoate Et 3 MeBu (Gambetta et al. 2014) (Gambetta et al. 2014) 2 Methylpropanol 2 MePrOH (Gambetta et al. 2014) (Gambetta et al. 2014) 2&3 Methylbutyl acetate 2 & 3- MeBuAc (Siebert et al. 2005) and 30 (Siebert et al. 2005) 3MeBuAc 129 (Lee and Noble 2003) (Louw et al. 2010) Butanol BuOH (Siebert et al. 2005) 64 (Lee and Noble 2003) 2130 (Louw et al. 2010) nd (Buettner 2004) (Lee and Noble 2003) 2&3 Methylbutanol 2 & 3 MeBuOH (Siebert et al. 2005) and (Siebert et al. 2005) Ethyl hexanoate EtHex (Gambetta et al. 2014) (Gambetta et al. 2014) Hexyl acetate HA (Gambetta et al. 2014) 24 (Gambetta et al. 2014) 1700 (Louw et al. 2010) Ethyl octanoate EtOct (Gambetta et al. 2014) (Gambetta et al. 2014)

8 8Capone Potent et al. thiols in Chardonnay wine AustralianPotent Journal thiols of Grape in Chardonnay and Wine Research wine Table 4. (continued) Compound Code Minimum (μg/ L) Maximum (μg/ L) Median (μg/ L) Aroma detection threshold (μg/l) Previously reported in Chardonnay wine (μg/l) Ethyl decanoate EtDec 726 > (Gambetta et al. 2014) 17 (Gambetta et al. 2014) 610 (Louw et al. 2010) Acetic acid AcetAcid (Gambetta et al. 2014) 6460 (Gambetta et al. 2014) (Louw et al. 2010) Propanoic acid PrAcid (Siebert et al. 2005) 63 (Lee and Noble 2003) (Louw et al. 2010) 2 Methylpropanoic acid 2 MePrAcid (Siebert et al. 2005) 200 (Louw et al. 2010) (Welke et al. 2014) Butanoic acid BuAcid (Siebert et al. 2005) (Welke et al. 2014) 4703(Lee and Noble 2003) 2 Methylbutanoic acid 2 MeBuAcid na Gambetta et al. (2014) Gambetta et al. (2014) 3 Methylbutanoic acid 3 MeBuAcid (Gambetta et al. 2014) (Gambetta et al. 2014) Hexanoic acid HexAcid (Gambetta et al. 2014) (Louw et al. 2010) Octanoic acid OctAcid (Gambetta et al. 2014) 1150 (Louw et al. 2010) (Gambetta et al. 2014) Decanoic acid DecAcid (Gambetta et al. 2014) (Gambetta et al. 2014) 2 Phenylethyl acetate 2 PEA (Gambetta et al. 2014) (Gambetta et al. 2014) 2 Phenylethanol 2 PhEtOH (Gambetta et al. 2014) (Gambetta et al. 2014) Hexan 1 ol HexOH (Guth 1997) 50 (Louw et al. 2010) 5022 (Lee and Noble 2003) (Z) 3 Hexen 1 ol 3 HexOH (Guth 1997) 62 (Lee and Noble 2003) (Welke et al. 2014) Linalool Lin (Gambetta et al. 2014) (Gambetta et al. 2014) α Terpineol Terp (Gambetta et al. 2014) (Gambetta et al. 2014) Nerol Ner (Ribéreau Gayon et al. 1975) (Welke et al. 2014) Geraniol Ger (Black et al. 2015) nd (Buettner 2004) β Damascenone DAM (Gambetta et al. 2014) 2 (Li et al. 2008) 170 (Gambetta et al. 2014) trans Ethyl cinnamate EthCinn (Gambetta et al. 2014) 1 33 (Gambetta et al. 2014) γ Octalactone γ OL (Cooke et al. 2009a,b) < (Cooke et al. 2009b) γ Nonalactone γ NL (Nakamura et al. 1988) <0.1 (Cooke et al. 2009b) 655 (Lee and Noble 2003) γ Decalactone γ DL (Gambetta et al. 2014) 3 53 (Gambetta et al. 2014) 3 S Cysteinylhexan 1 ol Cys 3 MH Odourless 15 (Capone et al. 2010) 3 S Glutathionyhexan 1 ol Glut 3 MH Odourless 89 (Capone et al. 2010) 3 S Cysteinylglycinehexan 1 ol CysGly 3 MH Odourless na In 10% aqueous ethanol unless indicated. Threshold determined in water. Threshold determined in wine. Threshold determined in beer. Three samples fell above the calibration limit of 1559 μg/l. na, no data available; nd, not detected.

9 Capone 46 et Potent al. thiols in Chardonnay wine Australian Journal of Potent Grape thiols and Wine Chardonnay Research 24, wine 38 50, compounds in these young wines was even greater than that of the commercial Chardonnay wines surveyed in this work, and in fact, the concentration was comparable to that found in Sauvignon Blanc wines (Benkwitz et al. 2012). The median concentration of 3 MH was 28 times higher than that of the aroma detection threshold (Tominaga et al. 1998b). Similarly, 3 MHA was found at a concentration approximately 12 times above its aroma detection threshold (Tominaga et al. 1996). Benzyl mercaptan was present at a much lower concentration than previously reported (Gambetta et al. 2014), but nevertheless, it was present well above the aroma detection threshold of this compound. As seen previously, thiol precursors were present in the finished wine at a concentration slightly higher than that previously reported in Chardonnay wine (Capone et al. 2010). Most of the other aroma compounds measured fell within the range of what had been previously reported in the literature; however, there were a few compounds that were found at somewhat higher concentration than had been previously reported. These included some of the oxidation related compounds (homofuraneol, sotolon and 2 phenylacetaldehyde) and some fermentation product compounds (ethyl 2 methylpropanoate, 2 methylbutanol and 3 methylbutanol, ethyl decanoate, 2 phenylethyl acetate and 2 phenylethanol) (Table 4). Two of the fermentation Figure 2. Biplot of principal components 1 and 2 for mean sensory scores for 16 Chardonnay wines made under standardised conditions from grapes from the following regions: Hunter Valley (HV), Margaret River (MR1 and MR2), Riverland (RL), McLaren Vale (MV), Padthaway (P), Rutherglen (R), Great Southern (GS1 and GS2), Adelaide Hills (AH), Yarra Valley (YV), Orange (O), Mornington Peninsula (MP), Great Western (GW), Tumbarumba (T) and Coal River Valley (TAS). A, aroma; F, flavour; AT, aftertaste. products (ethyl decanoate and 2 phenylethyl acetate) were at a concentration similar to that previously measured in commercial Riesling wines (Smyth 2005). The relatively high concentration of the oxidation related compounds found in some of the wines analysed in the present study was unexpected, as these wines were young and were all made in the same way. Sensory descriptive analysis Twenty four sensory attributes differed significantly (P < 0.05) among the 16 wines. Attributes that were not significant were melon, stonefruit, lemon, sweaty/cheesy and chemical aroma; tropical fruit flavour; sweet and bitter taste; and fruit aftertaste. Melon and sweaty/cheesy aroma and fruit aftertaste were found to be close to significant (P < 0.10) and were therefore included in subsequent analyses. The significant sensory attributes and those with P < 0.10 were subjected to PCA. The first three principal components (PCs) were important in showing the variation in the dataset; PC1 and PC2 accounted for 48.9% of the variation in the dataset, and PC3 accounted for a further 13.8% of variance (data not shown). Wines made from grapes sourced from Great Western (GW), Tumbarumba (T) and Tasmania (TAS) were rated highly in pineapple, floral, melon and confection. Wines made from grapes sourced from the Riverland (RL), Margaret River (MR), Great Southern (GS) and Yarra Valley (YV) were rated lower in those attributes and high in pungent, box hedge, flint and vegetal aromas and astringency, as seen along PC1 (Figure 2). The RL wine also had the highest concentration of 3 MH and 3 MHA. Separation along PC2 (Figure 2) was based on stonefruit flavour being highly rated for the Rutherglen (R) and Hunter Valley (HV) wines and the second Great Southern wine (GS2) being highly rated for lime and passionfruit aromas and green flavour. Wine GS2 also had the second highest concentration of both 3 MH and 3 MHA. Assessing the contribution of thiols to the sensory properties of the wines with PLS regression The multivariate relationships were determined between the significant sensory attributes and the chemical compositional data of the wines using PLS regression. Figure 3 shows the loadings of the volatile compounds with the sensory attributes. The model generated involved three optimum factors, with 73% of the variance of the sensory data explained by the model. Those volatiles located close to a sensory attribute are most associated positively with that attribute. The thiols 3 MH Figure 3. Plot of X and Y loadings from partial least squares regression for significant aroma and flavour terms using volatile compositional data for the 16 Chardonnay wines made under standardised conditions. X loadings (chemical variables) are shown in blue, and Y loadings (sensory attributes) are shown in red. Abbreviations of volatiles can be found in Table 4. A, aroma attributes; F, flavour attributes.

10 10 Capone et Potent al. thiols in Chardonnay wine AustralianPotent Journal thiols of Grape in Chardonnay and Wine Research wine and 3 MHA were associated with the passionfruit, citrus and green aromas, and with 4 MMP and BM, they were also related to the box hedge, pungent and flint attributes. As expected, the fermentation volatiles were associated with the terms overall fruit, confection and floral, with higher scores from the sensory panel for these attributes when the wines were lower in thiol concentration. The importance of the thiol compounds to tropical related attributes, particularly passionfruit and box hedge, can be seen in more detail by the regression coefficients for each compound (Figure 4). Compounds with high regression coefficients, either positively or negatively, are of greatest importance. 3 Mercaptohexan 1 ol and 3 MHA and to a lesser extent ethyl hexanoate were the main compounds positively associated with the passionfruit aroma (R 2 calibration predicted vs measured = 0.52), with the monoterpenes nerol and geraniol, as well as acetic acid and butanol, being negatively associated, indicating a potential masking or suppressing effect. In a similar pattern, 3 MH, 3 MHA and ethyl hexanoate were also highly related to box hedge aroma (R 2 calibration predicted vs measured = 0.68), but BM and 4 MMP were also associated with this attribute. The aroma attribute flint (defined by the sensory panel as struck match, smoky; R 2 calibration predicted vs measured = 0.52) was most strongly associated with the concentration of BM, which has previously been implicated as contributing to this character (Tominaga et al. 2003a,b). It should be noted that the reference standard for the flint sensory attribute was BM added to a base wine. The other fruity attributes, notably pineapple and confection, were related to several compounds, with confection aroma (R 2 predicted vs measured = 0.66) associated with linalool, nerol and geraniol, together with 2 methylbutyl and 3 methylbutyl acetate and two aldehydes, 2 methylpropanal and furfuryl aldehyde. Pineapple aroma was linked to octyl and hexyl acetate. The compounds related to fruit aftertaste (Figure 4) were of interest, as some of the highest positive regression coefficients for this attribute, and for fruit flavour, were from the thiols and the residual thiol precursors remaining in the wine, notably Glut 3 MH. This may indicate that breakdown of the Figure 4. Regression coefficients from a partial least squares model generated to relate volatile composition for the 16 Chardonnay wines made under standardised conditions with aroma and flavour attributes: (a) passionfruit aroma, (b) box hedge aroma, (c) green aroma, (d) flint aroma, (e) pineapple ( ) and confection ( ) aroma and (f) overall fruit flavour (F) ( ) and fruit aftertaste (AT) ( ). Abbreviations of volatiles can be found in Table 4.

11 Capone 48 et Potent al. thiols in Chardonnay wine Australian Journal Potent of Grape thiols and Wine Chardonnay Research 24, wine 38 50, precursor compounds in mouth, as demonstrated by Starkenmann et al. (2008), could be a contributor to these important sensory attributes. Consumer preferences A consumer study was conducted to determine the importance of thiol related characters to consumers. Six of the experimentally produced wines with varying sensory profiles were selected, and a group of regular white wine drinkers assessed the wines under blind conditions. Mean liking scores (n = 156) were relatively low, with only one wine scoring a mean above six on the nine point scale. Liking was generally strongly related to the intensity of passionfruit aroma and fruit flavour. Wines GS2, AH and GS1 were generally preferred by the consumers. Wine attributes, citrus, green apple, green and fruit aftertaste, were also positively related to liking scores, while stonefruit and flint were weak negative drivers for overall liking by the total sample of consumers. Cluster analysis indicated differences in consumer preferences for the thiol related attributes tropical, passionfruit and, in high concentration, box hedge. No wine was well liked by all four clusters of consumers. Average liking scores for each cluster are presented in Figure 5. The first cluster (cluster 1, 22% of consumers) preferred the wines AH, R and GS2, with higher pineapple, confection and floral attribute scores (Figures 2, 5), but did not appreciate the wine RL with stronger box hedge aroma and also gave a lower liking score to wines with higher vegetal and flint attributes. There were significantly more young consumers (18 25 years old) in this cluster compared to the other clusters and more people who did not drink Riesling (P < 0.05). Two clusters (clusters 2 and 3, 60% of consumers) preferred the wines with a high level of passionfruit aroma and disliked stonefruit, while a smaller cluster (cluster 4, 18%) liked less the wines with the tropical fruit attributes and those with higher astringency and green flavours. The stonefruit attribute had a small positive influence for this cluster, which contained more individuals in the age bracket and fewer experienced drinkers (<10 years) than the other clusters (P < 0.05). Significantly more consumers in cluster 4 agreed with the statements would pay more to expect a good bottle of wine, the best wines were more expensive and wine is too expensive to enjoy on a regular basis (P < 0.05). Figure 5. Consumer cluster analysis for the liking scores of the six Chardonnay wine samples made from grapes sourced from: Adelaide Hills ( ), Riverland ( ), McLaren Vale ( ), Rutherglen ( ), Great Southern 1 ( ) and Great Southern 2 ( ). Conclusions Overall, this study has shown that the highly potent thiols can be major contributors to Chardonnay wine flavour, with the concentration in Chardonnay being much higher than what has been reported in the earlier studies with a small number of wines. The compound 3 MH was at a surprisingly high concentration in many of the wines analysed, comparable to that found in Sauvignon Blanc. The thiols are likely to contribute tropical fruit aroma and flavour to Chardonnay, as well as enhancing the overall fruit flavour and aftertaste. Unwooded Chardonnay wines with high tropical fruit aroma and flavour were well accepted by consumers. High thiol concentration, however, notably high 4 MMP, in Chardonnay wines can give pungent box hedge/cat pee like aroma, which may cause a negative influence on liking for a sizeable proportion of consumers. The link of BM to smoky, gun flint type aromas confirms earlier reports relating this compound to this aroma in Sauvignon Blanc and Champagne wines. Acknowledgements The authors thank Mrs Amanda Aguis, Miss Joanna Vervey and Miss Sheridan Barter (AWRI) for technical assistance, Mrs Tracey Siebert for the γ lactone data, the sensory panellists and the Australian wine producers for donating grape juices. We also thank Dr Markus Herderich for manuscript evaluation and feedback. The work was conducted at the AWRI a member of the Wine Innovation Cluster at the Waite Precinct in Adelaide, and is supported by Australian grapegrowers and winemakers through their investment body, Wine Australia, with matching funds from the Australian Government. References Allen, T., Herbst Johnstone, M., Girault, M., Butler, P., Logan, G., Jouanneau, S., Nicolau, L. and Kilmartin, P.A. (2011) Influence of grape harvesting steps on varietal thiol aromas in Sauvignon Blanc wines. Journal of Agricultural and Food Chemistry 59, Benkwitz, F., Tominaga, T., Kilmartin, P.A., Lund, C., Wohlers, M. and Nicolau, L. (2012) Identifying the chemical composition related to the distinct aroma characteristics of New Zealand Sauvignon Blanc wines. American Journal of Enology and Viticulture 63, Bindon, K., Holt, H., Williamson, P.O., Varela, C., Herderich, M. and Francis, I.L. (2014) Relationships between harvest time and wine composition in Vitis vinifera L. cv. Cabernet Sauvignon 2. Wine sensory properties and consumer preference. Food Chemistry 154, Black, C.A., Parker, M., Siebert, T.E., Capone, D.L. and Francis, I.L. (2015) Terpenoids and their role in wine flavour: recent advances. Australian Journal of Grape and Wine Research 21, Buettner, A. (2004) Investigation of potent odorants and afterodor development in two Chardonnay wines using the buccal odor screening system (BOSS). Journal of Agricultural and Food Chemistry 52, Capone, D.L. and Jeffery, D.W. (2011) Effects of transporting and processing Sauvignon Blanc grapes on 3 mercaptohexan 1 ol precursor concentrations. Journal of Agricultural and Food Chemistry 59, Capone, D.L., Black, C.A. and Jeffery, D.W. (2012a) Effects on 3 mercaptohexan 1 ol precursor concentrations from prolonged storage of Sauvignon Blanc grapes prior to crushing and pressing. Journal of Agricultural and Food Chemistry 60, Capone, D.L., Sefton, M.A. and Jeffery, D.W. (2011b) Application of a modified method for 3 mercaptohexan 1 ol determination to investigate the relationship between free thiol and related conjugates in grape juice and wine. Journal of Agricultural and Food Chemistry 59, Capone, D.L., Sefton, M.A. and Jeffery, D.W. (2012b) Analytical investigations of wine odorant 3 mercaptohexan 1 ol and its precursors. Qian, M.C. and Shellhammer, T., eds. Flavor chemistry of wine and other alcoholic beverages (American Chemical Society: Washington, DC, USA) pp