Guaiacol and 4-methylguaiacol accumulate in wines made from smoke-affected fruit because of hydrolysis of their conjugates_

Transcription

1 Singh et al. Guaiacol and 4-methylguaiacol glycosides in smoke tainted wines S13 Guaiacol and 4-methylguaiacol accumulate in wines made from smoke-affected fruit because of hydrolysis of their conjugates_ D.P. SINGH 1, H.H. CHONG 2, K.M. PITT 1, M. CLEARY 2, N.K. DOKOOZLIAN 2 and M.O. DOWNEY 1 1 Department of Primary Industries Victoria, PO Box 95, Mildura,Vic. 352, Australia 2 E & J Gallo Winery, Modesto, CA , USA Corresponding author: Dr Mark Downey, fax , mark.downey@dpi.vic.gov.au Abstract Background and Aims: Taint in smoke-exposed grapes have been associated with elevated levels of guaiacol and 4-methylguaiacol. Previous research has reported guaiacol and 4-methylguaiacol in both fruits and wines. In some cases, these compounds were not detected, or were detected at low levels in the fruit while high levels were subsequently identified during or after winemaking. Later research indicated that this was due to the presence of glycosidic conjugates. Here we report a method for the routine analysis of guaiacol and 4-methylguaiacol released after acid hydrolysis of glycoside precursors. Methods and Results: Chardonnay, Merlot, Shiraz, Sangiovese and Cabernet Sauvignon fruits were collected following bushfire events in 6 7 in the King Valley wine region of NE Victoria, Australia. Gas chromatography-mass spectrometry (GC-MS) was used to detect free guaiacol and 4-methylguaiacol in both fruits and wines. Low levels of free and bound forms were present in fruit not exposed to smoke. Substantial levels of free guaiacol and 4-methylguaiacol were detected in the wines made from the smoke-affected fruits. These compounds increased during bottle storage. Acid hydrolysis of wines and berries resulted in a several-fold increase in free guaiacol and 4-methylguaiacol. Conclusions: The validated GC-MS method is suitable for monitoring free and glycosidically bound guaiacol and 4-methylguaiacol after acid hydrolysis in both fruits and wines. Acid hydrolysis of wines provided evidence that bound volatiles, most probably glycosidically, act as reserve for guaiacol and 4-methylguaiacol, which are released during ageing of wines. Significance of the Study: This is the first study published in a refereed journal to demonstrate that smoke taint-associated volatiles increase during ageing of wine and bound forms of guaiacol and 4-methylguaiacol represent an aroma reserve for smoke taint in ageing/bottled wines. Keywords: 4-methylguaiacol, bushfire, gas chromatography-mass spectrometry, glycoside, grapes, guaiacol, smoke, wine Introduction Bushfires are a natural feature of the Australian environment (Gill 1979). While fire events are a necessary element in the life cycle of some native plants (Bell et al. 1993, Dixon et al. 1995), they can have a devastating effect on communities and industries in fire-prone regions ( In the grape and wine industry, the negative impacts of bushfires can also occur in areas not directly threatened by fire because of the effects of bushfire smoke (Whiting and Krstic 7). Smoke has been shown to directly impact grape composition and subsequent wine quality (Kennison et al. 7). Wines made from smoke-affected grapes have been described as having smoky, dirty, earthy, burnt, smoked meat, damp fire and ashtray characteristics (Kennison et al. 7, Whiting and Krstic 7). In grapes and wine, a number of compounds have been identified that have sensory characteristics similar to those used to describe wines made from smoke-affected fruit (Maga 1988, Kennison et al. 7). Two of these, guaiacol and 4-methylguaiacol, have been widely used as indicator compounds in assessing the degree to which fruit and wines have been affected by smoke (Kennison et al. 7, Whiting and Krstic 7, Sheppard et al. 9). Guaiacol and 4-methylguaiacol are lignin degradation products (Maga 1984, Wittkowski et al. 1992). These are commonly found in wines that have been aged in oak barrels (Towey and Waterhouse 1996, Pollnitz et al. ). Guaiacol and 4-methylguaiacol may also be derived from corks (Simpson et al. 1986) and occur naturally in the fruit and leaves of some grape varieties, for example Shiraz, Merlot and Muscat of Alexandria (Sefton 1998, Wirth et al. 1). At low levels, these compounds add complexity to wine flavour and aroma (Francis and Newton 5), but at higher levels are undesirable and considered as a taint or fault in the wine (Boidron et al. 1988, Kennison et al. 7). Boidron et al. (1988) reported that the detection thresholds for guaiacol and 4-methylguaiacol in red wines were 75 mg/l and 65 mg/l, respectively, and in white wines were 95 mg/l and 65 mg/l. Whereas, Simpson et al. (1986) reported a much lower detection threshold of 2 mg/l for guaiacol. Smoke taint in wine occurs as a result of incorporation of smoke components into the fruit and their subsequent extraction into wine (Kennison et al. 7, Sheppard et al. 9). To avoid the loss of quality associated with smoke taint, fruits from smoke-affected vineyards tends not to be harvested; thus, these fruits represent a financial loss to growers and winemakers. For doi: /j x

2 S14 Guaiacol and 4-methylguaiacol glycosides in smoke tainted wines Australian Journal of Grape andwine Research 17,S13 S21,211 example, in 6 7, an estimated AUD$2 million worth of winegrapes, or approximately AUD$75 million worth of wine were lost in the north east of Victoria because of smoke taint (Whiting and Krstic 7). Following bushfire events, grape samples are routinely analysed for the presence of guaiacol and 4-methylguaiacol as indicators of smoke taint in the fruits as well as in the fermenting musts and finished wines (Jones 9). It has been proposed previously that the concentration of these compounds increases during vinification because of the hydrolytic release of guaiacol and 4-methylguaiacol from their glycosylated forms (Sefton 1998, Wirth et al. 1, Kennison et al. 8). The hypothesis is supported by the recent finding that juice from smoke-affected grapes contained higher levels of glucosides of guaiacol and 4-methylguaiacol compared with unsmoked grapes (Hayasaka et al. 21) and proposed that enzymatic hydrolysis is the major source of the release of smoke taint volatiles during fermentation. Analysis of the free guaiacol and 4-methylguaiacol in fruit and wines may substantially underestimate the level of potential smoke taint compounds in grapes and wine (Kennison et al. 8). While this initial work demonstrated an increase in guaiacol and 4-methylguaiacol during fermentation and anticipated increasing levels of these compounds in wines made from smoke-affected fruit during bottle ageing, no evidence of this has been published to date. Here, we report levels of guaiacol and 4-methylguaiacol in wines made from fruits exposed to smoke in the north east of Victoria during bushfires in 6 7 analysed immediately post-bottling and after 2 years of bottle ageing. In addition, we have adapted an existing analytical methodology (Zoecklein et al., Whiton and Zoecklein 2) to enable routine analysis of the smoke taint potential represented by bound forms of guaiacol and 4-methylguaiacol in both fruits and wines. Materials and methods Chemicals High-performance liquid chromatography grade acetonitrile, methanol, ethanol, sulphuric acid (H 2SO 4) and sodium hydroxide (NaOH) was purchased from Merck and Co. Inc. (Darmstadt, Germany). Guaiacol, 4-methylguaiacol standards and n-hexane (gas chromatography (GC) grade) were obtained from Sigma Aldrich (St. Louis MO, USA). 4-ethyl-d 5-2-methoxyphenol (d 5-4-EG) was purchased from CDN isotopes (Pointe-Claire, QB, Canada). Purity of all standards was verified by GC-mass spectrometry (MS) before preparation of stock solutions. Deionised water was obtained through a MilliQ system (Milli-RX Analytical-Grade Water Purification System, Millipore, Billerica MA, USA). Wine Samples 6 7 and 8 9 Wines were made from Vitis vinifera L. cv. Chardonnay, Merlot, Shiraz, Sangiovese and Cabernet Sauvignon fruits collected from the King Valley wine region of north eastern Victoria (36 42 South, East), Australia. Fruits were collected in March 7 following bushfire events in December 6 and January 7 (Table 1), and again in the March 9 following bushfires events in February 9. Because the region is incorporated within a phylloxera inclusion zone, fruits were frozen at 2 C for at least 7 days prior to shipping to a small-scale winery for winemaking or to the research laboratory for analysis of guaiacol and 4-methylguaiacol. Wines were made according to a standardised methodology (Walker et al. 1998). For the purposes of comparison, non-smoked fruits of Shiraz and Cabernet Sauvignon varieties (Table 1) from the 6 vintage and wines prepared from these fruits were analysed for both free and bound forms of guaiacol and 4-methylguaiacol. In 5 6, there was no bushfire activity in the major Australian viticultural production areas. Wines prepared from 6 7 growth season were analysed for free guaiacol and 4-methylguaiacol by the Australian Wine Research Institute (AWRI; Adelaide, South Australia) as described by Pollnitz et al. (4). This method has been reported to have a limit of detection (LOD) of 1 mg/l in both grapes and wines, with an uncertainty of 1. mg/l or 1% (whichever is greater). AWRI analysed wines on two occasions: 2 weeks post-bottling in June 7 and in May 9. These wines were also analysed in March 21 by the method described below. Fruits collected at harvest in March 7 were also analysed by the AWRI laboratory for free guaiacol and 4-methylguaiacol in April 7 (Pollnitz et al. 4). Grape and wine samples (8 9 growth season) were analysed for both free and bound forms of guaiacol and 4-methylguaiacol according to the methodology described below. Sample preparation Whole berries. Frozen whole berries were homogenised to a fine powder using a Retsch grinder (GM, Retsch, Haan, Germany) in the presence of liquid nitrogen. Ground berry material was transferred to a 5-mL polypropylene tube and stored at -8 C until analysed. For analysis, frozen berry material was thawed and centrifuged for 1 min at 2469 g. Ten millilitres of the supernatant were transferred to a clean 15-mL tube to which was added 1.5 ml of 1 M NaOH before vortex mixing and filtration through a.45-mm syringe filter. Wine. A 2-mL aliquot of the wine to be analysed was frozen in liquid nitrogen and dried using a freeze dryer (Freezone, Labconco Corporation, Kansas City MO, USA) at -75 C. The dried sample was redissolved in 1 ml of deionised water, 1.5 ml of 1 M NaOH was added and then the samples passed through a.45 mm syringe filter. Hydrolysis of guaiacol and 4-methylguaiacol bound precursors. For the hydrolysis of bound forms of guaiacol and 4-methylguaiacol, the whole berry supernatant, or the freeze dried wine samples, were purified utilising solid phase extraction (SPE) adapted to a 2-mL 96-well plate system (Oasis HLB Plate, Waters Corporation, Milford MA, USA) and vacuum manifold (96 Well Plate Manifold, Waters Corporation, Milford MA, USA). Solid phase columns were conditioned with.5 ml methanol followed by a rinse of.5 ml deionised water. For whole berries,.5 ml of sample, or 1 ml of wine sample was loaded in triplicate into wells and the liquid phase removed. Samples were rinsed three times with 1 ml aliquots of deionised water. Solid phase plate columns were eluted with.17 ml ethanol (99.9%) and were rinsed with.33 ml water into a clean 2-mL 96-well plate. One millilitre of each of the three replicates of each sample was then transferred to 2-mL GC-MS headspace autosampler vials. To these was added 4 ml of H 2SO 4 (ph1.). Autosampler vials were then sealed and incubated for 1 h at 1 C. Samples were cooled on ice and transferred to Kimble tubes (PYREX Corning, New York, USA) containing 1.5 g NaCl. These samples were spiked with 1 ml of internal standard (4 mg/l d 5-4-EG in ethanol) and vortexed for 1 min followed by incubation at room temperature for 3 min. Immediately after incubation 2 ml of n-hexane was added to

3 Singh et al. Guaiacol and 4-methylguaiacol glycosides in smoke tainted wines S15 Table 1. Sites (or vineyards) and harvest Brix of winegrape varieties, exposed to smoke as a result of bushfires in the 6 7 season in north eastern Victoria. Variety Vineyard and location Brix at Harvest Cabernet Sauvignon (non smoked, control) Mildura South East Shiraz (non smoked, control) Mildura South East Chardonnay King Valley 23.9 Chardonnay Whitfield South East Merlot Moyhu South East Merlot King Valley 25.3 Merlot Chestnut South East Shiraz Myrrhee South East Shiraz Chestnut South East Shiraz King Valley 26.7 Shiraz Chestnut South South East Shiraz Whitfield South East Sangiovese Myrrhee South East Sangiovese King Valley 23.7 Sangiovese Whitfield South East Cabernet Sauvignon Edi Upper South East Cabernet Sauvignon Chestnut South South East Cabernet Sauvignon King Valley 23. the tubes and vortexed for 1 min. Samples were stored at room temperature for 2 3 h before spinning at 2469 g for 5 min. A 1-mL portion of the organic phase was transferred to a 2-mL GC autosampler vial, capped and analysed for guaiacol and 4-methylguaiacol using GC-MS method described below. Free forms of guaiacol and 4-methylguaiacol were measured by extracting 5 ml of the wine to be analysed in triplicate with 2 ml of n-hexane after spiking with 1 ml of d 5-4-EG and adding 1.5 g NaCl similar to the sample procedure described above. d 5-4-EG was used as an internal standard because purchase of suitable stable isotope labelled analogues of guaiacol and 4-methylguaiacol were not commercially available. Gas chromatography mass spectral analysis of guaiacol and 4-methylguaiacol. Grape and wine samples were analysed for guaiacol and 4-methylguaiacol using an Agilent 789A gas chromatograph and 5975 mass spectrometer (Agilent Technologies, Palo Alto CA, USA). The column used was a fused silica capillary (DB-171P,.25 mm I.D. 3 m length.25 mm film thickness, J&W Scientific, Folsom CA, USA). High purity helium (BOC Gases, Adelaide SA, Australia) was used as a carrier gas with an average linear velocity of 37 cm/s and a flow rate of 1 ml/min. Liquid samples (1 ml) were injected using a CTC-PAL autosampler (CTC Analytics AG, Zwingen, Switzerland) into front GC inlet injector at 24 C by using a 4-mm i.d. liner (Agilent Technologies, Palo Alto CA, USA). GC injector (inlet 1) was operated in the split/splitless mode with a splitless time of 1.5 min followed by a split flow of 5 ml/min. Oven temperature started at 4 C and held for 3 min before increasing to 5 1 C/min. Temperature was then increased by 4 C/min until it reached 18 C and then increased to 22 C at 8 C/min and held for 2.33 min. Under

4 S16 Guaiacol and 4-methylguaiacol glycosides in smoke tainted wines Australian Journal of Grape andwine Research 17,S13 S21,211 Table 2. Concentration of G and 4-MG in grapes and wines exposed to smoke as result of bushfires during the 8 9 season in north eastern Victoria. Variety Site Grapes Wine Free G (mg/kg) Free 4-MG (mg/kg) Bound G Bound 4-MG Free G Free 4-MG Bound G Bound 4-MG Shiraz (control) Mildura ND ND ND ND ND Cabernet Sauvignon Mildura ND ND ND ND ND (control) Merlot King Valley ND ND ND ND ND Shiraz Moyhu ND ND ND Sangiovese King Valley.9.1 ND ND ND The samples were analysed for free and bound G and 4-MG by the method described in materials and methods. Values represent mean standard deviation (n = 3); the limits of detection determined for G and 4-MG were between.32 and.38 mg/l. Control grapes and wines were not exposed to smoke. ND, not detected; G, guaiacol; 4-MG, 4-methylguaiacol. this temperature programme the analyte retention times (RT) were RT = 9.66 min for guaiacol, RT = min for 4- methylguaiacol and RT = min for d 5-4-EG. The MS ion source temperature was 23 C and the GC-MS transfer line temperature was 22 C. A solvent delay of 5 min was set up and data acquisition mode was set to selective ion monitoring mode. The ions monitored were m/z 81, 19 and 124 for guaiacol; m/z 95, 123 and 138 for 4-methylguaiacol; and m/z 124, 139 and 157 for d 5-4EG (National Institute of Standards and Technology virtual library). The selected ions were monitored for 5 ms each. Calibration standards and method validation. Solutions containing 5, 25, 1, 5, 25, 1, 5, 2.5 and 1. mg/l guaiacol and 4-methylguaiacol were prepared in n-hexane. The calibration standard curves were prepared by transferring 1. ml of the mixed guaiacol/4-methylguaiacol solution to a 2-mL vial and adding 1 ml ofd 5-4-EG internal standard solution. The specificity, precision and validation of the analytical method were determined by spiking a series of standards to red wines (Shiraz and Cabernet Sauvignon). The wines were spiked in triplicate with, 1, 2, 4, 8, 16 and mg/l of mixed guaiacol and 4-methylguaiacol standards to determine recovery of these chemicals in n-hexane phase. The values of LOD and limit of quantification (LOQ) of guaiacol and 4-methylguaiacol were determined by analysing n-hexane spiked with 1 mg/l of mixed guaiacol/4-methylguaiacol standards. Standard deviation (SD) from ten independent analyses was determined to calculate LOD (3 SD) and LOQ (1 SD) (Clesceri et al. 1995). Results Method validation The parameters of the method were optimised using standards of guaiacol and 4-methylguaiacol in n-hexane. It was verified that these analytes presented rates of recovery and levels of detection compatible with their thresholds of perception and the concentrations expected in non-smoked fruit and wine (Table 2). The recovery of smoke taint compounds was >1% with relative SD (RSD) < 1% (Figure 1). The detection limits (LOD) determined for guaiacol and 4-methylguaiacol were between mg/l. The detection limits of the method allowed us to use it in routine analysis G (μg/l) 4-MG (μg/l) y = x R 2 =.9995 y = 2.683x R 2 = G or 4-MG spiked (μg/l) Figure 1. Recovery of guaiacol (G) and 4-methylguaiacol (4-MG) in n-hexane extracts after spiking of Shiraz wine with known amount of these analytes for each recovery level. The bars indicate standard deviation (n = 3). of guaiacol and 4-methylguaiacol. For guaiacol and 4-methylguaiacol, the LOQ was 1.27 mg/l and 1.6 mg/l, respectively. These values were close to the lowest concentration level of the working range. The reproducibility and reliability of the method was evaluated by measuring free guaiacol and 4-methylguaiacol after acid hydrolysis of SPE-extracted bound forms of these compounds over 2 days by two operators (Table 3). The RSD obtained by the method described was less than 1% for all the analytes and no significant differences were observed between the sets of data produced by the two operators. It was apparent that guaiacol and 4-methylguaiacol levels increased significantly after hydrolysis with H 2SO 4 at 1 C for 1 h (Figure 2). A time course experiment revealed that acid

5 Singh et al. Guaiacol and 4-methylguaiacol glycosides in smoke tainted wines S17 Table 3. Validation of analytical method showing the concentration of guaiacol released after acid hydrolysis of purified glycosides of guaiacol in wine produced from smoke-affected Merlot winegrapes in 8 9 season in King Valley (North Eastern Victoria). Variety Source Day Operator 1 Operator 2 RSD (%) Merlot King Valley The samples were analysed in triplicate by two operators on two different days. Values represent mean standard deviation (n = 3). RSD, relative standard deviation. Analyte (μg/l) detected Guaiacol 4-methylguaiacol Incubation time (min) at 1ºC Figure 2. Time course acid hydrolysis at 1 C to determine optimum time required to release bound forms of guaiacol and 4-methylguaiacol of Shiraz wine prepared from grapes exposed to bushfire smoke during the 6 7 growing season (n = 3). After 2 years of bottle ageing, wines were re-analysed and the concentrations of both guaiacol and 4-methylguaiacol in most of the wines had increased (Figure 3). The highest concentration of guaiacol was still observed in the Sangiovese wines and had increased by 1 15% during 2 years of bottle ageing. The highest concentration of 4-methylguaiacol at this time was observed in a wine made from Cabernet Sauvignon grapes which had an increase of around 45% in the concentration of 4-methylguaiacol. Indeed, two of the three highest concentrations of 4-methylguaiacol were in Cabernet Sauvignon wines; the third highest level was observed in Cabernet Sauvignon wine from the Edi Upper vineyard, which had increased by 67% during 2 years of bottle ageing. The biggest increase in guaiacol and 4-methylguaiacol concentrations during ageing was observed in the Chardonnay wines with an increase of over % in guaiacol and ~3% in 4-methylguaiacol. These wines were analysed in March 21 for free guaiacol and 4-methylguaiacol by the GC-MS method described here; for most wines guaiacol and 4-methylguiacol had continued to increase. The Chardonnay and two of the Merlot wines remained below the sensory threshold for perception, with one of these Merlot wines also now close to that threshold (73.2 mg/l). hydrolysis at 1 C for 3 min would be sufficient to release all of the bound form of guaiacol and 4-methylguaiacol into free forms for analysis by GC-MS. Analysis of free guaiacol and 4-methylguaiacol in new and aged wines Wines were made from smoke-affected fruits collected after bushfires in Victoria (Australia), which occurred during the 6 7 growing season. Following bottling of the wines, made in a small-scale experimental winery, the wines were analysed for the presence of guaiacol and 4-methylguaiacol and then wines were re-analysed 2 years later (Figure 3). Guaiacol concentration in the fruits ranged from 6.9 mg/kg to 183 mg/kg of fruits, while 4-methylguaiacol concentration in the fruits ranged from 1.3 mg/kg to 28.7 mg/kg. No guaiacol or 4-methylguaiacol were detected in unsmoked control Shiraz and Cabernet Sauvignon grapes. In the wines measured 2 weeks post-bottling, the concentration of guaiacol ranged from 13 mg/l to 377. mg/l, and 4-methylguaiacol ranged from 5 mg/l to 82 mg/l. As with the fruits, the highest concentrations of both guaiacol and 4-methylguaiacol were observed in the Sangiovese wines, while the lowest concentrations were observed in wines made from Chardonnay grapes. Analysis of bound precursors in grapes and wine The method utilised to determine the concentrations of bound guaiacol and 4-methylguaiacol in grapes and wines was adapted from an assay previously used to determine the levels of flavour and aroma precursors in grapes and wine known as the glycosyl-glucose assay (Zoecklein et al. ). Here, we have applied that methodology to the purification of the glycosylated precursors of guaiacol and 4-methylguaiacol, the acid-catalysed hydrolytic release of guaiacol and 4-methylguaiacol and identification of the aglycones by GC-MS. The methodology was applied to wines made from fruit exposed to smoke during bushfires in 6 7 and to wines made from fruits that had not been exposed to smoke (5 6). Wines made from fruits exposed to smoke during bushfire events in 6 7 had detectable levels of guaiacol and 4-methylguaiacol when measured immediately post-bottling. It was subsequently observed that the concentration of guaiacol and 4-methylguaiacol had increased when these wines were analysed again in 9 and 21, indicating release of these compounds from a pool of conjugated precursors. To examine the pool of potential smoke taint compounds these wines were hydrolysed with acid to release the remaining guaiacol and 4-methylguaiacol. Concentrations of free guaiacol and 4-methylguaiacol detected after acid hydrolysis of purified glycosidic forms of guaiacol and 4-methylguaiacol for

6 S18 Guaiacol and 4-methylguaiacol glycosides in smoke tainted wines Australian Journal of Grape andwine Research 17,S13 S21,211 (a) (b) (c) Guaiacol or 4-ethylguaiacol in fruit (µg/kg) Guaiacol in wine (µg/kg) 4-methylguaiacol in wine (µg/kg) Guaiacol 4-Methylguaiacol Bound 21 Total Bound 21 Total 21 KV W M KV C My C KV CS W My KV W EU CS KV Figure 3. Concentration of guaiacol and 4-methylguaiacol in grapes (a), and guaiacol (b) and 4-methylguaiacol (c) in wines prepared from these grapes exposed to smoke as a result of bushfires in the 6 7 season in north eastern Victoria (King Valley (KV), Whitfield (W), Moyhu (M), Chestnut (C), Myrrhee (My), Chestnut South (CS), Edi Upper (EU)). Grapes were analysed following harvest in 7 for free guaiacol and 4-methylguaiacol by Pollnitz et al. (4) method. (n = 1). Wines were analysed for free guaiacol and 4-methylguaiacol immediately post-bottling in 7 and 2 years later in 9 (n = 1) by Pollnitz et al. (4) method. Wines were subsequently analysed by the method described here for both free and bound forms of guaiacol and 4-methylguaiacol in 21, values represent mean of analytical replicates (n = 3). Totals represent the sum of free and bound guaiacol analysed in 21. Chardonnay Merlot Shiraz Sangiovese Cabernet Sauvignon unsmoked (control) and smoked grape varieties are shown (Figure 3). Following acid hydrolysis of the wines from 6 7, high concentrations of both guaiacol and 4-methylguaiacol were observed in all of the wines. In all cases, the concentrations were above the sensory thresholds for detection and in many cases the concentration was greater than ten times the threshold for sensory detection. Different levels of guaiacol and 4-methylguaiacol were observed in the wines made from grapes, collected from various sites (Table 1), exposed to bushfire smoke during the 6 7 growing season. For example, in Chardonnay wines, the guaiacol and 4-methylguaiacol concentrations ranged from 192 mg/l (King Valley) to 262 mg/l (Whitfield) and 86 mg/l (King Valley) to 152 mg/l (Whitfield), respectively (Figure 3). These values include both the free form and that released by acid hydrolysis at the time of analysis in 21 (Figure 3). Highest levels of guaiacol were found in Merlot ( mg/l) and Shiraz wines ( mg/l) made from fruits collected at Cheshunt site. By comparison, wines made from fruits that had not been exposed to smoke had relatively low levels of the bound precursors of guaiacol. In Shiraz, the detected level of guaiacol was mg/l and in Cabernet Sauvignon, mg/l (Table 2). Furthermore, bound precursors for 4-methylguaiacol were not detected through acid hydrolysis of the wines. To explore the levels of free and bound smoke taint precursors in fruits, the methodology was refined using grapes and the wines made from those grapes following bushfire events in 8 9, and also from fruits that had not been exposed to smoke (Table 2). No free or bound precursors of 4-methylguaiacol were detected in control fruits. But wine made from control fruit had low levels of free and bound forms of guaiacol. In contrast, smoke-exposed fruit and wines showed elevated levels of both free and bound precursors of smoke

7 Singh et al. Guaiacol and 4-methylguaiacol glycosides in smoke tainted wines S19 taint. Among the three varieties studied, the Shiraz fruits and wines had highest levels of free and bound guaiacol and 4-methylguaiacol (Table 2). Discussion A GC-MS based liquid injection method was developed to determine free analytical levels of guaiacol and 4-methylguaiacol after acid hydrolysis of purified precursors of taint compounds. The optimised conditions for releasing the free form of guaiacol and 4-methylguaiacol were hydrolysis with 2.5 M H 2SO 4 (ph 1.) at 1 C for 3 min (Figure 2). It is still remained to be seen whether acid hydrolysis completely hydrolysed the precursors under experimental conditions used in this study. Nevertheless, these findings are consistent with recently published data (Hayasaka et al. 21), where glycoside precursors were found to be the source of free taint compounds in smoke-exposed winegrapes. They reported that both acid and enzymatic hydrolyses released free guaiacol and 4-methylguaicol from glycosidic precursors. The red wine matrix (Shiraz and Cabernet Sauvignon) did not have any significant impact on recovery, precision and specificity of the method for the detection of guaiacol and 4-methylguaiacol (Figure 1 and data not shown). The recovery of smoke taint compounds was better than 1% with RSD < 1% (Figure 1). This could be due to hydrolysis of soluble guaiacol and 4-methylguaiacol precursors at higher injector block temperatures as has also been reported previously especially when polar solvents were used at high injector temperatures (Pollnitz et al. 4). Another reason for this discrepancy could be matrix enhancement of the GC response; usually from the active sites in the liner and column being shielded by compounds in the matrix resulting in a larger response for the target compounds. The RSD for the lowest concentration (i.e. 1. mg/l) of guaiacol and 4-methylguaiacol was 9 1% indicating high repeatability of the method. The LOD and LOQ values enabled us to identify and quantify guaiacol and 4-methylguaiacol routinely with 99% confidence. Furthermore, under analyses conditions (described in Materials and Methods), the expected values of controls and unknown samples determined by two operators were within <1% RSD. Using this method, we have examined the levels of free and bound guaiacol and 4-methylguaiacol in a range of both smokeaffected and non-smoked grape and wine samples. Previous research has shown that smoke exposure alters the chemical composition and aroma of winegrapes and the wine products. Guaiacol and 4-methylguaiacol were the main compounds implicated in taint in smoke-affected fruit and wines. In this study we have demonstrated that the smoked grapes and resultant wines had enhanced levels of guaiacol and 4-methylguaiacol compared with wines and fruit that had not been exposed to smoke. In addition, we have shown that during 2 years of storage, the concentration of guaiacol and 4-methylguaiacol increased in bottled wines, providing evidence that the wines contained high levels of guaiacol and 4-methylguaiacol precursors, most likely as glycosides, acting as a pool of taint precursors. To our knowledge this is first demonstration of an increase in smoke taint during bottle ageing. Grapes smoked, as a result of bushfires during 6 7 in the King Valley wine region of north eastern Victoria, had elevated levels of guaiacol and 4-methylguaiacol (Figure 3). These compounds were not detected in control non-smoked fruits. The highest concentrations of guaiacol were observed in Sangiovese and the lowest in Merlot. Although duration and intensity of smoke for all these sites needs to be determined, it is tempting to speculate that there is differential accumulation of smoke taint compounds based on variety. This hypothesis is further supported by high levels of guaiacol observed in Sangiovese (1 mg/l) and Shiraz (19 mg/l) wines, prepared from fruit collected at King Valley site, compared with Cabernet Sauvignon (522 mg/l), whereas, Merlot wines had intermediary levels of guaiacol i.e. 823 mg/l (Figure 3) at the same site. It would be interesting to compare smoke intensity and duration for all these sites and correlate it to the responsive phenological stage of each variety. Recently, 7days post-veraison was identified as a peak period of Merlot vine sensitivity to smoke (Kennison et al. 9). The 4-methylguaiacol concentrations were also highest in the Sangiovese fruits, while both Shiraz and Merlot had similarly low levels. The ratio of guaiacol and 4-methylguaiacol concentrations in fruits varied between , suggesting a differential absorption of guaiacol and 4-methylguaiacol into grape skin. The ratio resulting from combustion of different wood sources such as Ponderosa pine and Quercus sp. (oak) have been reported to range between 1.6 to 1.85 (Edye and Richards 1991, Guillén and Manzanos 2). Of the wines prepared from smoke-affected fruit, only the Chardonnay, Merlot and one of the Shiraz wines had guaiacol and 4-methylguaiacol concentrations below the sensory detection threshold (Figure 3). Rapp and Versini (1996) reported guaiacol to have a negative impact on wine aroma at concentrations exceeding 8 mg/l. Simpson et al. (1986) reported guaiacol to be responsible for an off-flavour in wine; the taint, originating from contaminated corks, was attributed to guaiacol levels ranging from 7 mg/l to 263 mg/l with a detection threshold of 2 mg/l being reported in the study. Wines produced from Shiraz and Cabernet Sauvignon winegrapes not exposed to smoke had low levels of guaiacol and 4-methylguaiacol (Table 2). It is possible that these levels could originate from other environmental factors or that guaiacol and 4-methylguaiacol are synthesised or accumulate to detectable levels in these cultivars. These compounds may exist in all varieties below the current level of detection. Previous studies have reported guaiacol as a glycoside in the berries of Tempranillo, Grenache (Lopez et al. 4), Shiraz (Wirth et al. 1), Merlot (Sefton 1998) and in Chardonnay juice (Hayasaka et al. 21). Analysis of 6 7 season wines in 9 and 21, showed significant increases in free guaiacol and 4- methylguaiacol (Figure 3) during ageing in bottles. Because these wines were never in contact with wood at any stage of winemaking or storage, the most likely source of additional free smoke taint analytes is the hydrolysis of bound forms of guaiacol and 4-methylguaiacol present in the wines at bottling. If the taint- and wood-derived precursors are the same, then data presented in this paper are consistent with previously published results where concentrations of free guaiacol and 4- methylguaiacol had been shown to increase in wines during cooperage in oak barrels (Sefton 1998, Wirth et al. 1, Moreno and Azpilicueta 7, Ortega-Heras et al. 7). This increase in free guaiacol and 4-methylguaiacol had been suggested to be due to hydrolysis of glycosides of compounds present in the wood (Wittkowski et al. 1992, Kennison et al. 8). Release of free guaiacol and 4-methylguaiacol from their precursors after acid hydrolysis in this study, supports this hypothesis. Concentrations of free 4-methylguaiacol increased in all the bottled wines during 3 years of bottle storage but remained lower than the sensory detection threshold and were always significantly lower than concentrations of free guaiacol (Figure 3). Because guaiacol levels increased substantially

8 S2 Guaiacol and 4-methylguaiacol glycosides in smoke tainted wines Australian Journal of Grape andwine Research 17,S13 S21,211 during storage and were above the sensory detection threshold, we suggest that for routine analysis, it would probably be sufficient for the wine industry to use only free and bound guaiacol concentrations as indicators of smoke taint and smoke taint potential in grapes and wine. Concentrations of hydrolytically released guaiacol from its bound forms ranged from mg/l to mg/l and 4-methylguaiacol ranged from 62.9 mg/l to 76.8 mg/l in the wines measured 2 years post-bottling (Figure 3). The highest concentrations of hydrolytically released guaiacol and 4-methylguaiacol were observed in Merlot and Shiraz, while the lowest concentrations were observed in wines from Chardonnay grapes. This observation was consistent with previous reports that wines made from white grapes tended to have lower levels of guaiacol and 4-methylguaiacol and that this was because of the absence of skin contact during winemaking with the wines being made from free-run juice (Kennison et al. 8). This suggests that guaiacol and 4-methylguaiacol are concentrated in the skin (or seeds) of the berry. The thickness of grape berry skin can vary between 3 8 mm and depending on vine variety; skins constitute 1 12% by weight of a mature grape berry. Recently, Sheppard et al. (9) observed that thicker-skinned berries absorbed less guaiacol when exposed to smoke. Earlier observations of low levels of guaiacol and 4-methylguaiacol in white wines made without skin contact suggested that smoke taint compounds were confined to the skins of the fruit, possibly adhering to the cuticular wax of the berry (Høj et al. 3, Whiting and Krstic 7, Kennison et al. 8, Sheppard et al. 9). While this may be the case for the free analytes, it is possible that the skin is not the sole source of guaiacol and 4-methylguaiacol extracted from the berries and that high levels of the glycosylated precursors may be present especially in the mesocarp (flesh) of berries exposed to high smoke levels. This probably is the case in the samples analysed here as indicated by the significant increases in guaiacol and 4-methylguaiacol in Chardonnay wines upon acid hydrolysis (Figure 3). This is consistent with previous work where higher levels of these compounds were detected in the flesh of the berry because of the very high smoke exposure of grapes (Kennison et al. 7). It would be interesting to analyse these compounds in each of the berry tissues (skin, seeds and flesh) separately to localise their distribution in the berry. As discussed above, hydrolysis of the glycosylated precursors resulted in an increase in free smoke taint compounds in wines during bottle ageing (Figure 3). The presence of smoke taint precursors in wine has negative implications for the shelf life of bottled wines, although the full implications of this are uncertain because the rate of release is unknown. It would be necessary to study the kinetics of hydrolysis of these precursors in different wine matrices and under different storage conditions to develop a predictive model for release of guaiacol and 4-methylguaiacol and to determine whether an equilibrium between free and bound forms is reached. This information could then be used to determine shelf life of wines containing levels of free smoke taint compounds below the sensory threshold but with a significant pool of bound compounds. A further gap in our current knowledge is the extraction of both free and bound forms into the wine during fermentation or maceration. Together, these data could be used to evaluate the potential risk posed by a range of guaiacol levels in fruits at harvest. As a corollary to this work, it is important to understand the smoke intensity and its duration in the field, resulting in levels of guaiacol that represent a significant risk. This risk will vary with variety and with wine style and with the expected shelf (or cellar) life of the wine. While there has been some research into the dose-response of grapes to smoke, this needs to be extended to include both bound and free forms of guaiacol in all of the varieties currently in production in smoke prone regions. Different winemaking styles will also need to be considered. This investigation must also include a network of smoke detectors in these regions that will both generate valuable research data and a risk management tool for the industry. Conclusions In this study, we demonstrated that guaiacol and 4-methylguaiacol accumulate in bottled wines, produced from smoke-exposed grapes, during ageing. GC-MS data strongly suggest that hydrolysis of precursors of guaiacol and 4-methylguaiacol is the most likely source of increase in smoke taint content during storage of bottled wines. Guaiacol was always present at concentrations considerably in excess of 4-methylguaiacol. The optimised and validated GC-MS method was demonstrated to be useful for quantitative analysis of smoke taint compounds in grape berries and wines. In an event of bushfire, this technique could be used to verify the results of sensory analysis, and to certify the criteria for acceptance or rejection of fruit by wine makers. Acknowledgements The authors gratefully acknowledge the assistance of Mr Stephen Lowe of the King Valley Vignerons in the collection and transport of samples from bushfire-affected regions of NE Victoria. The authors also thank Mr Peter NanKervis (Agilent Technologies Australia) for his tremendous help during method development and Mrs Marica Mazza (DPI Vic.) for providing non-smoked grapes and wines for comparison. Analysis of free guaiacol and 4-methylguaiacol, in the 6 7 season grapes and wines, was conducted by the Australian Wine Research Institute (AWRI). The authors also recognise the invaluable support of Mrs Joanne Butterworth-Gray of the Victorian Wine Industry Association and Mr Liam Fogarty of the Victorian Department of Sustainability and Environment. This research was supported by E& J Gallo Winery, California (USA), the Victorian Department of Primary Industries and the Victorian Department of Sustainability and Environment. References Bell, D., Plummer, J. and Taylor, S. (1993) Seed germination ecology in southwestern Western Australia. The Botanical Review 59, Boidron, J.N., Chatonnet, P. and Pons, M. (1988) Influence du bois sur certaines substances odorantes des vins. Connaissance de la Vigne et du Vin 22, Clesceri, L.S., Greenberg, A.E. and Eaton, A.D. (Eds.) (1995) Standard methods for the examination of water and wastewater, 2 th edition (American Public Health Association: Washington, DC). Dixon, K.W., Roche, S. and Pate, J.S. (1995) The promotive effect of smoke derived from burnt native vegetation on seed germination of Western Australian plants. Oecologia 11, Edye, L.A. and Richards, G.N. (1991) Analysis of condensates from wood smoke components derived from polysaccharides and lignins. Environmental Science and Technology 25, Francis, I.L. and Newton, J.L. (5) Determining wine aroma from compositional data. Australian Journal of Grape and Wine Research 11, Gill, A.M. (1979) Fire in the Australian landscape. Landscape and Planning 6, Guillén, M.D. and Manzanos, M.J. (2) Study of volatile composition of an aqueous oak smoke preparation. Food Chemistry 79, Hayasaka, Y., Dungey, K.A., Baldock, G.A., Kennison, K.R. and Wilkinson, K.L. (21) Identification of a b-d-glucopyranoside precursor to guaiacol in grape juice following grapevine exposure to smoke. Analytica Chimica Acta 66, Høj, P., Pretorius, I. and Blair, R. (3) Investigations conducted into the nature and amelioration of taints in grapes and wine, caused by smoke

9 Singh et al. Guaiacol and 4-methylguaiacol glycosides in smoke tainted wines S21 resulting from bushfires. In: The Australian Wine Research Institute annual report 3. (AWRI: Adelaide) pp Jones, L. (9) Wine industry compassion rises from the ashes of Victorian vineyards. Australian Viticulture 13, Kennison, K.R., Wilkinson, K.L., Williams, H.G., Smith, J.H. and Gibberd, M.R. (7) Smoke-derived taint in wine: effect of postharvest smoke exposure of grapes on the chemical composition and sensory characteristics of wine. Journal of Agricultural and Food Chemistry 55, Kennison, K.R., Gibberd, M.R., Pollnitz, A.P. and Wilkinson, K.L. (8) Smoke-derived taint in wine: the release of smoke-derived volatile phenols during fermentation of Merlot juice following grapevine exposure to smoke. Journal of Agricultural and Food Chemistry 56, Kennison, K.R., Wilkinson, K.L., Pollnitz, A.P., Williams, H.G. and Gibberd, M.R. (9) Effect of timing and duration of grapevine exposure to smoke on the composition and sensory properties of wine. Australian Journal of Grape and Wine Research 15, Lopez, R., Ezpeleta, E., Sanchez, I., Cacho, J. and Ferreira, V. (4) Analysis of the aroma intensities of volatile compounds released from mild acid hydrolysates of odourless precursors extracted from Tempranillo and Grenache grapes using gas chromatography-olfactometry. Food Chemistry 88, Maga, J.A. (1984) Flavor contribution of wood in alcoholic beverages. In: Progress in flavour research. Ed. J. Adda (Elsevier: Dourdan) pp Maga, J.A. (1988) Smoke in food processing (CRC Press Inc.: Boca Raton). Moreno, N.J. and Azpilicueta, C.A. (7) Binding of oak volatile compounds by wine lees during simulation of wine ageing. Lebensmittel- Wissenschaft und-technologie 4, Ortega-Heras, M., González-Sanjosé, M.L. and González-Huerta, C. (7) Consideration of the influence of aging process, type of wine and oenological classic parameters on the levels of wood volatile compounds present in red wines. Food Chemistry 13, Pollnitz, A.P., Pardon, K.H. and Sefton, M.A. () Quantitative analysis of 4-ethylphenol and 4-ethylguaiacol in red wine. Journal of Chromatography A 874, Pollnitz, A.P., Pardon, K.H., Sykes, M. and Sefton, M.A. (4) The effects of sample preparation and gas chromatograph injection techniques on the accuracy of measuring guaiacol, 4-methylguaiacol and other volatile oak compounds in oak extracts by stable isotope dilution analyses. Journal of Agricultural and Food Chemistry 52, Rapp, A. and Versini, G. (1996) Flüchtige phenolische verbindungen in wein. Deutsche Lebensmittel Rundschau 92, Sefton, M.A. (1998) Hydrolytically-released volatile secondary metabolites from juice sample of Vitis vinifera grape cvs Merlot and Cabernet Sauvignon. Australian Journal of Grape and Wine Research 4, Sheppard, S.I., Dhesi, M.K. and Eggers, N.J. (9) Effect of pre- and post-veraison smoke exposure on guaiacol and 4-methylguaiacol concentration in mature grapes. American Journal of Enology and Viticulture 6, Simpson, R.F., Amon, J.M. and Daw, A.J. (1986) Off-flavour in wine caused by guaiacol. Food Technology in Australia 38, Towey, J.P. and Waterhouse, A.L. (1996) The extraction of volatile compounds from French and American oak barrels in Chardonnay during three successive vintages. American Journal of Enology and Viticulture 47, Walker, R.R., Clingeleffer, P.R., Kerridge, G.H., Rühl, E.H., Nicholas, P.R. and Blackmore, D.H. (1998) Effects of the rootstock Ramsey (Vitis champini) on ion and organic acid composition of grapes and wines, and on wine spectral characteristics. Australian Journal of Grape and Wine Research 4, Whiting, J. and Krstic, M.P. (7) Understanding the sensitivity to timing and management options to mitigate the negative impacts of bush fire smoke on grape and wine quality scoping study. Report (Victorian Department of Primary Industries: Melbourne). Whiton, R.S. and Zoecklein, B.W. (2) Evaluation of glycosyl-glucose analytical methods for various glycosides. American Journal of Enology and Viticulture. 53, Wirth, I., Guo, W., Baumes, R. and Günata, Z. (1) Volatile compounds released by enzymatic hydrolysis of glycoconjugates of leaves and grape berries from Vitis vinifera Muscat of Alexandria and Shiraz cultivars. Journal of Agricultural and Food Chemistry 49, Wittkowski, R., Ruther, J., Heike, D. and Rafiei-Taghanaki, F. (1992) Formation of smoke flavour compounds by thermal lignin degradation. In: Flavour precursors. Eds. R. Teranashi, G.R. Takeoka and M. Güntert (American Chemical Society: Washington, DC) pp Zoecklein, B.W., Douglas, L.S. and Jasinki, Y.W. () Evaluation of the phenol free glycosyl-glucose determination. American Journal of Enology and Viticulture 51, Manuscript received: 12 July 21 Revised manuscript received: 16 November 21 Accepted: 2 December 21