Comparison of methods for the analysis of smoke related phenols and their conjugates in grapes and wine

Authors


Dr Kerry Wilkinson, fax +61 8 8303 7116, email kerry.wilkinson@adelaide.edu.au; Dr Davinder Singh, fax +61 3 5051 4523, email davinder.singh@dpi.vic.gov.au; Dr Yoji Hayasaka, fax +61 8 8303 6601, email yoji.hayasaka@awri.com.au

Abstract

Background and Aims:  Australian grape growers and winemakers have typically relied on guaiacol and 4-methylguaiacol measurements to determine smoke exposure of grapes following bushfires or prescribed burns. However, the guaiacol and 4-methylguaiacol content of grapes does not always correlate with the extent of taint in resultant wines. This study compared several methods for the analysis of smoke related phenols and their conjugates in grapes and wine, to determine their capacity as diagnostic assays for smoke exposure.

Methods and Results:  Smoke-affected grapes were sourced from commercial vineyards exposed to bushfire smoke and from experimental field trials involving the application of smoke to grapevines, and small-scale wines were made from a number of these samples. Several analytical methods were applied to grapes and wine to determine the concentration of smoke related phenols and their conjugates. Strong correlations were observed between the glycoconjugate content of smoke-affected grapes and the concentration of guaiacol and 4-methylguaiacol released following acid hydrolysis of juice.

Conclusions:  Where smoke-affected grapes contain low or non-detectable levels of guaiacol and 4-methylguaiacol, analytical methods that quantify their glycoconjugate forms (either directly or indirectly) provide a better indication of the extent of smoke exposure.

Significance of the Study:  This is the first study to compare different methods for assessing smoke exposure in grapes and wine, through analysis of free and bound guaiacol and 4-methylguaiacol. These methods will allow grape growers and winemakers to more reliably assess smoke exposure of grapes, enabling better informed decisions to be made with regards to harvesting and processing smoke-affected grapes.

Introduction

The impact of grapevine exposure to smoke on the composition and sensory attributes of grapes and wine has been the subject of several recent publications (e.g. Kennison et al. 2008, Sheppard et al. 2009, Hayasaka et al. 2010a, Dungey et al. 2011, Singh et al. 2011), in response to the recurrent incidence of bushfires in proximity to wine regions in Australia, as well as overseas. Smoke comprises numerous volatile compounds (Baltes et al. 1981, Maga 1988); however its characteristic aroma is thought to be largely caused by the presence of volatile phenols (Wittkowski et al. 1992). Kennison et al. (2007) reported guaiacol, 4-methylguaiacol, 4-ethylguaiacol, 4-ethylphenol and eugenol at elevated concentrations in wine made from grapes exposed to straw-derived smoke post-harvest. Of these, guaiacol and 4-methylguaiacol were most abundant. While guaiacol and 4-methylguaiacol are not considered solely responsible for the objectionable ‘smoky’, ‘burnt’ and ‘smoked meat’ characters associated with smoke-tainted wine (Kennison et al. 2009), they are useful marker compounds for assessing the exposure of grapes to smoke; particularly given they can be readily quantified by existing gas chromatography-mass spectrometry (GC-MS)-based analytical methods (Pollnitz et al. 2004).

However, the guaiacol and 4-methylguaiacol content of grapes is not always a reliable indicator of the extent of smoke exposure by grapes. Grapes harvested from grapevines exposed to smoke for 1 h at either pre-veraison, post-veraison or maturity were found to contain guaiacol and 4-methylguaiacol at levels that ranged from 2 to 26 µg/kg and from 0 to 4 µg/kg, respectively (Sheppard et al. 2009). In contrast, Kennison et al. (2008) detected only trace levels of guaiacol, 4-methylguaiacol, 4-ethylguaiacol and 4-ethylphenol (i.e. ≤1 µg/L) in free-run juice derived from fruit harvested from grapevines exposed to repeated smoke applications between veraison and maturity; but significant quantities of these phenols were evolved when the juice was subsequently fermented or hydrolysed under strong acid or β-glucosidase enzyme conditions. This disparity is attributed to the glycoconjugation of volatile phenols in grapes, following grapevine smoke exposure (Hayasaka et al. 2010a, Dungey et al. 2011). Hayasaka et al. (2010a) confirmed the presence of a β-D-glucopyranoside of guaiacol in juice of smoke-affected Chardonnay and Sangiovese grapes. Subsequent stable isotope tracer experiments tentatively identified six additional guaiacol disaccharides (Hayasaka et al. 2010b). Glycoconjugate forms of smoke related volatile phenols are not detected by current GC-MS analytical methods and so the extent of smoke exposure in grapes can be under-estimated by these methods. The assessment of smoke exposure is further complicated given that guaiacol has been identified in both free and conjugate (bound) forms, as a natural component of grapes of several Vitis vinifera varieties, including Merlot (Sefton 1998), Shiraz (Wirth et al. 2001), Tempranillo and Grenache (Lopez et al. 2004).

Grape growers and winemakers clearly need reliable diagnostic assays for grape exposure to smoke, in order to make informed decisions with regards to harvesting and processing smoke-affected fruit. This study was therefore undertaken to compare methods for the analysis of smoke related phenols and their conjugates in grapes and wine, in order to investigate their capacity as diagnostic assays for smoke exposure. Analytical methods included: (i) a GC-MS based stable isotope dilution assay (SIDA) method for the quantification of guaiacol and 4-methylguaiacol (Pollnitz et al. 2004); (ii) a high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS)-based SIDA method for the quantification of guaiacol glycoconjugates (Dungey et al. 2011); and (iii) a direct acid hydrolysis technique based on the glycosyl-glucose (GG) assay (Williams et al. 1995) and an adaption of this assay incorporating solid phase extraction (SPE), for the quantification of glycosylated guaiacol and 4-methylguaiacol (Singh et al. 2011).

Materials and methods

Smoke-affected grapes: grapes exposed to experimental smoke

Control and smoke-affected Grenache, Merlot and Viognier grapes were sourced from field trials involving the application of smoke to grapevines under experimental conditions described by Ristic et al. (2011) and Dungey et al. (2011). Grenache vines (approximately 100) growing in a vineyard located in the Barossa Valley region of South Australia (34°30′S, 138°59′E) and Merlot and Viognier vines (three vines per variety, in triplicate) growing in vineyards located at the University of Adelaide's Waite campus in Adelaide, South Australia (34°58′S, 138°38′E) were exposed to straw-derived smoke at approximately 7 days post-veraison, i.e. at juice total soluble solids (TSS) concentrations of approximately 15°Brix, determined using a digital handheld refractometer (PAL-1, Atago, Tokyo, Japan). The duration of smoke exposure was 20 min for Grenache vines and 30 min for Merlot and Viognier vines. Grapes were harvested from smoked and control vines when juice TSS reached 23 ± 1°Brix.

Smoke-affected grapes: grapes exposed to bushfire smoke

Smoke-affected grapes were sourced from several Victorian vineyards exposed to smoke from bushfires that occurred between February 7 and March 14, 2009. Shiraz grapes were sourced from Yarra Valley (37°42′S, 145°30′E) and Shiraz, Cabernet Sauvignon, Chardonnay and Pinot Noir grapes were sourced from Goulburn Valley (37°42′S, 145°30′E). The Yarra Valley and Goulburn Valley Shiraz are hereafter differentiated as Shiraz 1 and Shiraz 2, respectively. Fruit was harvested at juice TSS levels of 23 ± 1°Brix, except for Cabernet Sauvignon, which was harvested at TSS of 19 ± 1°Brix and Shiraz 1, which was harvested at TSS > 30°Brix. Fruit was stored at −20°C prior to analysis and winemaking.

Winemaking

Fruit was randomly divided into parcels (approximately 5 kg) and processed (in triplicate) according to small-lot winemaking procedures (Holt et al. 2006). Irrespective of variety, all fermentations were conducted with skin contact to eliminate any effect of winemaking practice, particularly given the duration of skin contact during fermentation has been shown to influence the intensity of smoke taint in wine (Ristic et al. 2011). Bunches were de-stemmed and crushed, and tartaric acid added to adjust the pH to 3.5, prior to inoculation with Maurivin PDM yeast (200 ppm). Musts were fermented on skins for 7 days at 15°C, with the cap plunged at least twice per day. The wine was pressed at 1 °Baume, transferred to stainless steel vessels and held at 20°C until the residual sugar approached 0 g/L. Wines were then racked from gross lees and cold stabilised (at 4°C for 3 weeks). Wine pH and free SO2 were adjusted to 3.4 and between 20 and 40 ppm, respectively, before filtration (Z6 grade pad for non-sterile, 0.45 µm for sterile) and bottling (under screw cap closures). Because of the extremely high TSS content of Shiraz 1 (i.e. >30°Brix), water (700 mL) was added to each fermentation replicate, but all other winemaking protocols were as described earlier. Acid adjustments were not made to Chardonnay juice or wine, because of high natural acid content (i.e. pH 3.0), but all other winemaking protocols were as described above. Malolactic fermentation was not carried out for any of these wines.

Determination of guaiacol and 4-methylguaiacol by GC-MS

Guaiacol and 4-methylguaiacol were quantified according to SIDA methods developed by Pollnitz et al. (2004). For fruit analysis, grapes (200 berries, in triplicate) were randomly collected from each fruit parcel and average berry weight determined. Samples were then homogenised (T18 Ultra Turrax, IKA, Staufen, Germany) and an aliquot (25 g) of the resulting whole berry homogenate accurately weighed into centrifuge vials. Following addition of d3-guaiacol and d3-4-methylguaiacol (10 µg/mL in ethanol, 100 µL) as internal standards, samples were centrifuged (4300× g for 5 min). Organic solvent (n-pentane c. 2 mL) was added to an aliquot (5 mL) of the resulting juice and the mixture shaken briefly. A portion of the organic layer (c. 1.5 mL) was then taken for instrumental analysis. For wine analysis, internal standards (10 µg/mL in ethanol, 100 µL) were added to the sample (10 mL) in a screw cap vial and extracted with organic solvent as mentioned earlier, prior to analysis. The preparation of internal standards, method development and validation, and instrument operating conditions have been published previously (Pollnitz et al. 2004). Analyses were performed by the Australian Wine Research Institute's Commercial Services Laboratory (Adelaide, Australia) using an Agilent 6890 gas chromatograph (Agilent, Palo Alto, CA, USA), coupled to a 5973 mass selective detector. For these methods: linearity was confirmed for a concentration range from 1 to 1000 µg/L, with correlation coefficients of 0.999 and 1.000 for guaiacol and 4-methylguaiacol, respectively; the limit of detection (3 × standard deviation) was 1 µg/L; and the precision was better than 5% relative standard deviation.

Determination of guaiacol glycoconjugates by HPLC-MS/MS

Guaiacol glycoconjugates were quantified according to the SIDA methods previously described (Dungey et al. 2011, Ristic et al. 2011), using an Agilent 1200 HPLC system, coupled to a 4000 Q TRAP hybrid tandem mass spectrometer. For fruit analysis, d4-guaiacol β-D-glucopyranoside (100 µg/mL in water, 50 µL) was added as an internal standard to a subsample (10 g) of whole berry homogenate (as mentioned earlier) and the mixture centrifuged (4300× g for 5 min). The glycoconjugate fraction of the resulting juice was purified using C18-HF SPE cartridges (Grace, Rowville, Australia) and extracts filtered through a 0.45-µm GHP membrane (Acrodisc®, PALL Life Sciences, Cheltenham, Australia) prior to instrumental analysis. For wine analysis, d4-guaiacol β-D-glucopyranoside (100 µg/mL in water, 10 µL) was added to the sample (1 mL), the mixture shaken briefly and filtered as mentioned earlier, prior to analysis. The preparation of internal standards, method development and validation, and instrument operating conditions have been previously published (Dungey et al. 2011, Ristic et al. 2011). For these methods: linearity was confirmed for a concentration range from 1 to 10 000 µg/kg (for homogenate) and 10 to 50 000 µg/L (for wine), with correlation coefficients of 0.993 and 0.995 for homogenate and wine, respectively; the limit of detection (3 × standard deviation) was 25 µg/kg (for homogenate) and 7 µg/L (for wine); and the precision was better than 5% relative standard deviation.

Determination of glycosylated guaiacol and 4-methylguaiacol by GC-MS after acid hydrolysis

Direct acid hydrolysis.  Acid hydrolysates were prepared from each whole berry homogenate (as above). Homogenate samples (20 g) were centrifuged (4300× g for 5 min) and an aliquot (10 mL) of the resulting juice acidified to pH 1.0 with concentrated sulphuric acid, then heated at 100°C for 1 h, i.e. conditions used in the glycosyl-glucose assay (Williams et al. 1995). Samples were subsequently analysed by GC-MS and HPLC-MS/MS as mentioned earlier to determine guaiacol, 4-methylguaiacol and guaiacol glycoconjugate concentrations.

SPE/acid hydrolysis.  Bound forms of guaiacol and 4-methylguaiacol were quantified according to the modified glycosyl-glucose assay (Williams et al. 1995) as described by Singh et al. (2011). Whole berry homogenate samples (20 g) were centrifuged (2500× g for 10 min) and to an aliquot (10 mL) of the resulting juice was added aqueous sodium hydroxide (1.5 mL, 10 M), followed by vortex mixing and filtration through a 0.45 µm syringe filter. Wine (20 mL) was freeze-dried and the residue reconstituted in distilled water (10 mL), followed by the addition of aqueous sodium hydroxide (1.5 mL, 10 M) and filtration, as mentioned earlier. Homogenate and wine-derived filtrates (1 mL) were extracted for bound forms of guaiacol and 4-methylguaiacol using a 2 mL 96-well plate system (Oasis® HLB Plate, Waters Corporation, Milford, MA, USA) and vacuum manifold (96 Well Plate manifold, Waters Corporation). Extracts were acidified to pH 1.0 with sulphuric acid (2.5 M), and then heated at 100°C for 1 h. Samples were subsequently analysed by GC-MS to determine guaiacol and 4-methylguaiacol concentrations, using similar methodology as mentioned earlier, except: acid hydrolysates (5 mL) were extracted with n-hexane (2 mL); d5-4-ethylguaiacol (10 µL) was added as the internal standard; and analyses were performed with an Agilent 7890A gas chromatograph coupled with a 5975 mass selective detector. For these methods: linearity was confirmed for a concentration range from 1 to 500 µg/L, with correlation coefficients of 0.999; the limit of detection (3 × standard deviation) was 0.4 µg/L; and the precision was better than 10% relative standard deviation (Singh et al. 2011).

Statistical analysis

Chemical data were analysed using JMP (Version 5.0.1, SAS Institute, Cary, NC, USA) to generate a correlation matrix.

Results and discussion

The timing and duration of grapevine smoke exposure have been shown to influence the extent of smoke taint in wine, with prolonged or repeated exposure resulting in wines with higher volatile phenol content and more intense smoke related sensory attributes (Kennison et al. 2009). Grapevine smoke exposure can range from several hours to several weeks, depending on the burn time of a bushfire or prescribed burn and the prevailing wind conditions. In the current study, Grenache, Merlot and Viognier grapevines were exposed to smoke (for 20 or 30 min) under experimental conditions, whereas Shiraz, Cabernet Sauvignon, Chardonnay and Pinot Noir grapevines were exposed to bushfire smoke over a 5-week period, but specific details regarding the timing, duration and density of smoke exposure were not known. While the impact of grapevine exposure to smoke can be readily determined in wine via sensory analysis, it would be beneficial for the wine industry to objectively ascertain the extent of smoke exposure prior to the expense and effort associated with harvesting and processing grapes. The diagnostic potential of a range of analytical methods for the quantification of free and bound guaiacol and 4-methylguaiacol were therefore evaluated. These methods were applied to control and smoke-affected grapes, and where sufficient fruit was available for winemaking, their corresponding small-lot wines. Chemical data from grape and wine analysis are reported in Table 1.

Table 1.  Concentrations of free guaiacol and 4-methylguaiacol (by gas chromatography-mass spectrometry (GC-MS)), guaiacol glycoconjugates (by HPLC-MS/MS), and bound guaiacol and 4-methylguaiacol (by GC-MS after direct acid hydrolysis or solid phase extraction (SPE)/acid hydrolysis) in smoke-affected grapes and wine derived from grapevines exposed to experimental or bushfire smoke. Control grapes and wine were not exposed to smoke.
 GrenacheMerlotViognierShiraz 1Shiraz 2Cabernet SauvignonChardonnayPinot Noir
ControlSmoked§ControlSmoked§ControlSmoked§SmokedSmokedSmokedSmokedSmoked
  • Values are means from three sample replicates (n = 3) and were in agreement to c. 10%.

  • Values are means from three sample replicates, each measured in triplicate (n = 9) and were in agreement to c. 10%.

  • §

    §Smoke-affected grapes sourced from grapevines exposed to experimental smoke.

  • Smoke-affected grapes sourced from grapevines exposed to bushfire smoke. nd, not detected; tr, trace (i.e. positive identification but <1 µg/L for guaiacol and 4-methylguaiacol or <25 µg/L for guaiacol glycoconjugates). HPLC-MS/MS, high-performance liquid chromatography-tandem mass spectrometry.

Grapes           
 Guaiacol (µg/kg)ndndndnd1510812434
 4-methylguaiacol (µg/kg)ndndndndndndndnd2nd5
 Guaiacol glycoconjugates (µg/kg)3029434358342538753474152610121079
 Acid hydrolysate guaiacol (µg/L)426337630792831246983
 Acid hydrolysate 4-methylguaiacol (µg/L)2721027855302520
 Acid hydrolysate guaiacol glycoconjugates (µg/L)ndtrndndndndnd31trtr54
 SPE/acid hydrolysate guaiacol (µg/kg)8248351030722831136379
 SPE/acid hydrolysate 4-methylguaiacol (µg/kg)nd7311261781363027
Wine           
 Guaiacol (µg/L)tr51928760
 4-methylguaiacol (µg/L)nd1tr7413
 Total guaiacol glycoconjugates (µg/L)382908251660980880
 SPE/acid hydrolysate guaiacol (µg/L)12411211829887
 SPE/acid hydrolysate 4-methylguaiacol (µg/L)nd1616564728

Analysis of free guaiacol and 4-methylguaiacol in grapes by GC-MS

Guaiacol was detected in all of the bushfire smoke-affected grapes, at concentrations ranging from 4 µg/L for Chardonnay to 34 µg/L for Pinot Noir. For fruit derived from experimental field trials, control and smoke-affected Viognier grapes contained 1 µg/L and 5 µg/L guaiacol, respectively, but guaiacol was not detected in control or smoked Grenache and Merlot grapes. 4-Methylguaiacol was detected in Cabernet Sauvignon and Pinot Noir only, i.e. the two varieties containing the highest concentrations of guaiacol. The higher abundance of guaiacol relative to 4-methylguaiacol is consistent with literature concerning the composition of both wood smoke (Kornreich and Issenberg 1972) and smoke-tainted wine (Kennison et al. 2007, 2008).

Analysis of guaiacol glycoconjugates in grapes by HPLC-MS/MS

Guaiacol glycoconjugates (measured by HPLC-MS/MS) were present in all grape samples: at low concentrations for control grapes (approximately 30 µg/kg), as natural grape components; at concentrations between 253 and 358 µg/kg for grapes exposed to smoke under experimental conditions; and at considerably higher levels for grapes exposed to bushfire smoke (875 to 3474 µg/kg). Differences in grape glycoconjugate concentrations are likely to reflect the duration of smoke exposure. No smoke exposure for control grapevines, short (20 or 30 min) smoke exposure for Grenache, Merlot and Viognier grapevines under experimental conditions and prolonged smoke exposure for Shiraz, Cabernet Sauvignon, Chardonnay and Pinot Noir grapevines located in bushfire-affected regions. Pinot Noir and Cabernet Sauvignon grapes contained substantial quantities of guaiacol and 4-methylguaiacol and relatively high glycoconjugate concentrations, 1079 and 1526 µg/kg, respectively. However, guaiacol glycoconjugates were most abundant in Shiraz 2 (3474 µg/kg). While a proportion of these glycoconjugates might represent natural grape constituents, as observed in control grapes in the current study, as well as in previous studies (Wirth et al. 2001, Singh et al. 2011), the majority of the glycoconjugate pool is likely because of the glycosylation of smoke-derived guaiacol as demonstrated by Hayasaka et al. (2010b) and Dungey et al. (2011).

Analysis of bound guaiacol and 4-methylguaiacol in grapes by GC-MS after acid hydrolysis

Good agreement was seen between the two hydrolysis methods for bound guaiacol (i.e. direct acid hydrolysis and SPE/acid hydrolysis) and across the different types of grapevine smoke exposure (control, experimental or bushfire). As expected, hydrolysates of control grapes contained the least guaiacol being between 3 and 6 µg/L following direct acid hydrolysis and between 8 and 10 µg/L following adapted acid hydrolysis. Guaiacol concentrations between 24 and 37 µg/L were measured in hydrolysates prepared from smoke-affected Grenache, Merlot and Viognier grapes, following either direct or adapted acid hydrolysis. Hydrolysis of bushfire smoke-affected grapes yielded the highest guaiacol concentrations, ranging from 63 and 69 µg/L for Chardonnay, 72 and 79 µg/L for Shiraz 1, 79 and 83 µg/L for Pinot Noir, 113 and 124 µg/L for Cabernet Sauvignon, to 283 µg/L for Shiraz 2, depending on which acid hydrolysis based analytical method was employed. A similar trend was observed for 4-methylguaiacol. The lowest 4-methylguaiacol concentrations (2 to 4 µg/L) were observed in hydrolysates of control grapes; whereas hydrolysates from smoke-affected Grenache, Merlot and Viognier grapes contained between 6 and 11 µg/L. For bushfire smoke-affected grapes, greater amounts of 4-methylguaiacol were measured following adapted acid hydrolysis (from 17 µg/L for Shiraz 1 to 81 µg/L for Shiraz 2) than following direct acid hydrolysis (from 8 µg/L for Shiraz 1 to 55 µg/L for Shiraz 2). However, the relative order of concentrations within varieties was consistent regardless of the analytical method applied, i.e. Shiraz 2 > Cabernet Sauvignon > Chardonnay > Pinot Noir > Shiraz 1. Interestingly, despite containing the least amount of guaiacol, Chardonnay hydrolysates contained more 4-methylguaiacol than Pinot Noir and Shiraz 1 hydrolysates. At this stage it is unclear if this reflects differences in grape metabolism, smoke composition or perhaps the physiological responses of different grapevine varieties to smoke.

HPLC-MS/MS analysis after acid hydrolysis indicated small amounts of guaiacol glycoconjugates remained intact in hydrolysates of Shiraz 2 and Pinot Noir, i.e. 31 and 54 µg/L, respectively. However, in most cases, glycoconjugates were either not detected after hydrolysis, or detected at trace levels only (i.e. <25 µg/L), indicating near complete degradation under the acid hydrolysis conditions employed. On the contrary, a quantitative recovery of guaiacol was not observed. Given the molecular mass of guaiacol and its conjugate forms, i.e. 124 (free guaiacol), 286 (glucoside) and 448 (glucose-glucose disaccharide) atomic mass units, the guaiacol concentrations observed represent a recovery rate from acid hydrolysis between 15 and 30%. While enzyme hydrolysis occurs via cleavage of the glycosidic C–O linkage to release the aglycones, i.e. guaiacol or 4-methylguaiacol in the current study, Sefton (1998) suggested acid-catalysed hydrolysis of phenolic glycosides to involve cleavage of the phenolic ether linkage, to generate a reactive anisol carbocation intermediate. The low recovery of guaiacol after hydrolysis may therefore be attributed to the instability or reactivity of this intermediate, which is expected to rapidly undergo either further degradation and/or chemical reaction.

Analysis of smoke related phenols and their conjugates in wine

Compared with the aggressive conditions employed during the direct and adapted acid hydrolysis methods, i.e. high temperature and low pH, fermentation conditions are quite mild and in the current study resulted in hydrolysis of a considerably lower proportion of conjugated guaiacol and 4-methylguaiacol. Evolution of free guaiacol and 4-methylguaiacol during fermentation is most likely attributable to yeast enzyme activity, because smoke-derived phenols were shown not to be liberated under weak acid hydrolysis conditions (Kennison et al. 2008). Trace levels of guaiacol were detected in the control Grenache wine; with low levels (5 µg/L guaiacol, 1 µg/L 4-methylguaiacol) observed in the corresponding smoked Grenache wine. For wines made from bushfire smoke-affected grapes, Pinot Noir contained the highest guaiacol concentration (60 µg/L), followed by Cabernet Sauvignon (28 µg/L), Shiraz 1 (19 µg/L) and Chardonnay (7 µg/L). The free guaiacol concentrations of these wines were approximately twice that determined for their corresponding grapes. As such, for these particular wines, the grape guaiacol content was, to some extent, indicative of the level of smoke-derived volatile phenols in the resulting wine. However, this was not evident for the smoked Grenache wine, because neither guaiacol nor 4-methylguaiacol could be detected in smoke-affected Grenache grapes. The Grenache is therefore an example of a potential ‘false-negative’ classification, i.e. smoke-affected fruit for which determination of free guaiacol and 4-methylguaiacol would not accurately reflect the extent of smoke exposure by grapes. Application of a diagnostic assay involving quantification of glycoconjugates, either directly by HPLC-MS/MS or indirectly by acid hydrolysis would address this limitation.

As in previous studies (Hayasaka et al. 2010c, Ristic et al. 2011), significant concentrations of guaiacol glycoconjugates remained in the finished wine, having endured the fermentation process. Control Grenache wine contained the lowest glycoconjugate concentration (38 µg/L), followed by the smoked Grenache wine (290 µg/L) and then the bushfire smoke-affected Shiraz 1, Pinot Noir, Chardonnay and Cabernet Sauvignon wines, which contained 825, 880, 980 and 1660 µg/L, respectively. Significant quantities of bound guaiacol and 4-methylguaiacol were similarly observed following analysis of wine using the adapted acid hydrolysis method. The guaiacol and 4-methylguaiacol content of smoke-tainted wines has been shown to increase with bottle age (Kennison et al. 2008, Singh et al. 2011), which the authors attributed to the hydrolysis of glycoconjugate precursors remaining in wine after fermentation. Given the glycoconjugate content of wines in the current study, a similar increase in guaiacol and 4-methylguaiacol concentrations would be expected with time. Measurement of wine glycoconjugate concentrations, again either directly by HPLC-MS/MS or indirectly by acid hydrolysis, might therefore indicate the potential for smoke taint to intensify during storage.

Comparison of methods for the analysis of smoke-related phenols and their conjugates

Statistical analysis of compositional data obtained from the different analytical methods was used to generate a correlation matrix (Table 2). Grape guaiacol and 4-methylguaiacol concentrations were found to be poorly correlated with measurements of conjugate forms, whether by HPLC-MS/MS or either of the acid hydrolysis methods. Correlation coefficients ranged from 0.316 to 0.354 for guaiacol and from 0.128 to 0.208 for 4-methylguaiacol. This is not surprising, given the disparity in free and bound guaiacol and 4-methylguaiacol content observed for the different grape varieties. Namely, guaiacol and 4-methylguaiacol were not detectable in smoke-affected Grenache or Merlot grapes. Low levels of free guaiacol in Shiraz 2 accompanied high levels of guaiacol glycoconjugates and bound guaiacol and 4-methylguaiacol. High levels of free guaiacol in Pinot Noir corresponded to only moderate levels of guaiacol glycoconjugates and bound guaiacol and 4-methylguaiacol. These findings again highlight the potential for grapevine smoke exposure to be considerably underestimated by GC-MS measurement of free grape guaiacol and 4-methylguaiacol alone, and the need for glycoconjugate forms to be taken into consideration also.

Table 2.  Correlation matrix for grape compositional data obtained from the different analytical methods.
 Guaiacol4-methylguaiacolGuaiacol glycoconjugatesAcid hydrolysate guaiacolAcid hydrolysate 4-MethylguaiacolSPE/acid hydrolysate guaiacolSPE/acid hydrolysate 4-Methylguaiacol
  1. Correlation coefficients above 0.900 are significant at P < 0.001.

Guaiacol1.0000.9370.3540.3390.3440.3160.350
4-methylguaiacol0.9371.0000.1720.1490.2080.1280.181
Guaiacol glycoconjugates0.3540.1721.0000.9970.9740.9950.995
Acid hydrolysate guaiacol0.3390.1490.9971.0000.9610.9980.986
Acid hydrolysate 4-methylguaiacol0.3440.2080.9740.9611.0000.9550.989
SPE/acid hydrolysate guaiacol0.3160.1280.9950.9980.9551.0000.983
SPE/acid hydrolysate 4-methylguaiacol0.3500.1810.9950.9860.9890.9831.000

In contrast, all direct and indirect measurements of guaiacol glycoconjugates and bound guaiacol and 4-methylguaiacol were strongly correlated, with correlation coefficients ranging from 0.955 to 0.998 (Table 2). In the current study, the guaiacol glycoconjugate content of grapes was indicative of the amount of guaiacol and 4-methylguaiacol released during either direct or adapted acid hydrolysis. Glycoconjugate and hydrolysate guaiacol concentrations gave correlation coefficients greater than 0.995, whereas glycoconjugate and hydrolysate 4-methylguaiacol gave correlation coefficients greater than 0.974. Strong correlations were also observed between the direct and adapted acid hydrolysis methods, being 0.998 for bound guaiacol and 0.989 for bound 4-methylguaiacol. The strong correlations established for these different analytical methods indicate consistency in selectivity and relative robustness of the quantitative data generated across a range of samples, such that any of the methods could be implemented as diagnostic assays for the quantification of conjugated guaiacols.

Because the different techniques are characterised by similar and sufficient analytical method performance data, method selection may be determined by the analytical and instrumental capabilities of individual wine, research or commercial laboratories. Relying exclusively on the quantification of free guaiacol and 4-methylguaiacol in grapes has obvious limitations as discussed above. Quantification of bound guaiacol (after acid hydrolysis of grapes or wine) or guaiacol released from glycoconjugates isolated by SPE (Singh et al. 2011) would provide a suitable first pass screening tool and an alternative to the quantification of free guaiacol, offering improved diagnostic value. However, the concentrations and profiles of smoke related phenolic glycoconjugates have been shown to vary significantly between varieties and smoke events (Hayasaka et al. 2010c, Dungey et al. 2011) and the recovery of guaiacol from acid hydrolysis is relatively low (Hayasaka et al. 2010c) and may be affected by matrix variation. Hence, even quantification of guaiacol and 4-methylguaiacol following acid hydrolysis of glycoconjugates does not entirely resolve the risk of ‘false negative’ results and should be augmented through HPLC-MS/MS analysis of the full glycoconjugate profile where possible.

Conclusion

The methods for determination of bound guaiacol and 4-methylguaiacol evaluated in the current study were found to be highly correlated and yielded more informative data than quantification of free guaiacol and 4-methylguaiacol alone. Guaiacol and 4-methylguaiacol were again shown to occur naturally in some grape samples. As such, their presence at trace concentrations in grapes does not always provide definitive evidence of grapevine exposure to smoke. Furthermore, glycoconjugation of smoke-derived volatile phenols can result in trace or undetectable levels of free guaiacol and 4-methylguaiacol in smoke-affected grapes. Hence, analytical methods that either directly or indirectly account for their glycoconjugate forms are better suited to detect grapevine smoke exposure and may allow development of predictive models that assess the risk for smoke taint developing in wine.

Acknowledgements

The authors would like to gratefully acknowledge: Con Simos of the AWRI for sourcing smoke-affected fruit from Victoria; Louisa Rose and staff of the Yalumba Wine Company for participation in experimental field trials; Simon Odell and Gemma West of the AWRI for technical support with the winemaking; Randell Taylor and Commercial Services Laboratory (AWRI) for GC-MS analysis; and staff and students from UA and AWRI for participating in sensory trials. This research was supported by the Grape and Wine Research and Development Corporation, the Australian Research Council (Linkage Project LP0989138) with financial support from four industry partners, and the Victorian Department of Primary Industries. Kerry Pinchbeck and Anthea Fudge thank the GWRDC for the provision of research scholarships.

Ancillary