Effect of smoke application to field-grown Merlot grapevines at key phenological growth stages on wine sensory and chemical properties

Authors

  • K.R. KENNISON,

    Corresponding author
    1. Department of Agriculture and Food WA, PO Box 1231, Bunbury, WA 6230, Australia
    2. Curtin University of Technology, School of Science, Department of Environment and Agriculture, PMB 1, Margaret River, WA 6285, Australia
    Search for more papers by this author
  • K.L. WILKINSON,

    1. Curtin University of Technology, School of Science, Department of Environment and Agriculture, PMB 1, Margaret River, WA 6285, Australia
    2. The University of Adelaide, School of Agriculture, Food and Wine, PMB 1, Glen Osmond, SA 5064, Australia
    Search for more papers by this author
  • A.P. POLLNITZ,

    1. The Australian Wine Research Institute, PO Box 197, Glen Osmond, SA 5064, Australia
    Search for more papers by this author
    • Present address: Forensic Science South Australia, 21 Divett Place, Adelaide, SA 5000, Australia.

  • H.G. WILLIAMS,

    1. Curtin University of Technology, School of Public Health, GPO Box U1987, Perth, WA 6845, Australia
    Search for more papers by this author
  • M.R. GIBBERD

    1. Curtin University of Technology, School of Science, Department of Environment and Agriculture, PMB 1, Margaret River, WA 6285, Australia
    Search for more papers by this author

Ms Kristen Kennison, fax +61 8 9780 6136, email kristen.kennison@agric.wa.gov.au

Abstract

Background and Aims:  Smoke exposure of grapevines and development of smoke taint in wine are issues of increasing incidence and severity. There is limited understanding of the effect of phenological stage at the time of smoke exposure on taint development. The aim of this study was to demonstrate the variation in smoke uptake and taint development between and within seasons.

Methods and Results:  Smoke was applied to field-grown Merlot grapevines at 12 stages of vine development over three growing seasons. Key periods of vine sensitivity to smoke taint in wine were (i) from shoots at 10 cm to full bloom (low levels of smoke taint); (ii) from berries at pea size to the onset of veraison (variable levels of smoke taint); and (iii) between 7 days post-veraison and harvest (high levels of smoke taint).

Conclusions:  The severity of taint in wine varied depending on the phenological timing of grapevine smoke exposure. Taint was elevated when exposure occurred between 7 days post-veraison and harvest. The carry-over of smoke constituents the following season was not detectable in wine but yields were reduced.

Significance of the Study:  This is the first study to demonstrate the timing of smoke exposure to critically affect wine chemical and sensory characters. These effects were consistent and reproducible over three seasons.

Introduction

The exposure of grapevines to smoke and the subsequent development of smoke taint in wine is an important issue for wine producers globally. Increases in the incidence of wildfires and fire-risk weather (actual and predicted) in Australia, Canada and the USA have been attributed to climate change (Gillett et al. 2004, Hennessy et al. 2005, Westerling et al. 2006). As such, the presence of smoke in viticultural areas is likely to occur more frequently, resulting in the increased occurrence of smoke taint for wine producers.

Smoke produced from the combustion of vegetative biomass contains numerous substances including inorganic gases (carbon monoxide, ozone, nitrogen dioxide), polycyclic aromatic hydrocarbons, volatile and semi-volatile organic compounds, particulate matter (PM2.5 and PM10) and oxygenated organics (Schauer et al. 2001, Lee et al. 2005, Naeher et al. 2007). The production of these substances varies with the combustion conditions such as moisture and oxygen availability, temperature and fuel composition (Maga 1988b, Hays et al. 2002, Simon et al. 2005). Fuel composition can also vary depending on the fuel type and source and it is generally comprised of lignin (18–35%), cellulose (40–45%) and hemicellulose (20–35%; Maga 1989). To date, there are no published reports on the effect of different fuels and fuel pyrolysis conditions on the development of smoke taint in wine.

The presence of smoke taint in wine can result in the wine being unacceptable for consumption with tainted wines exhibiting ‘burnt rubber’, ‘smoked meat’, ‘leather’, ‘disinfectant’, ‘ash’, ‘smoked salmon’ and ‘salami’ characters (Høj et al. 2003, Kennison et al. 2009). However, smoking of foods is one of the oldest methods of food processing and has traditionally been used to impart flavours, aromas, colours and for food preservation by reducing antimicrobial spoilage (Wittkowski et al. 1992, Fellows 2000). Smoke and liquid smoke flavourings utilised in food processing have been shown to contain compounds including carbonyls, aldehydes, lactones, ketones, furans, pyrans and phenols (Maga 1988a, Guillén et al. 1995, Guillén and Manzanos 1996). The volatile phenols guaiacol and 4-methylguaiacol, which derive from the thermal degradation of lignin, are present in smoke and have been reported to exhibit ‘smoky’, ‘phenolic’, ‘sharp’, ‘smoked meat’ and ‘burning’ aromas and flavours (Baltes et al. 1981, Boidron et al. 1988, Maga 1988a, Rocha et al. 2004). Guaiacol and 4-methylguaiacol are routinely detected in wines aged in toasted oak barrels at concentrations up to 100 and 20 µg/L, respectively (Pollnitz et al. 2004).

Research on the effects of smoke exposure at a range of grapevine growth stages and the subsequent development of smoke taint in wine is limited. A direct link between smoke exposure to grapes and the development of smoke taint in wine has been established for both grapes exposed to smoke postharvest (Kennison et al. 2007) and field-grown grapevines exposed to smoke between veraison and harvest (Kennison et al. 2009). The later study identified a peak period of grapevine sensitivity to smoke uptake at ‘veraison + 7d’. Furthermore, Kennison et al. (2009) also utilised repeated smoke exposures to grapevines to demonstrate an accumulation effect of volatile phenols in the final wine product (388 µg/L guaiacol, 93 µg/L 4-methylguaiacol) with these wines demonstrating elevated smoke-like aromas of ‘burnt rubber’, ‘smoked meat’ and ‘leather’. In another study conducted in British Columbia, application of smoke to field-grown grapevines pre-veraison, post-veraison and at maturity resulted in the detection of guaiacol (2 to 26 µg/L) in grapes; however, the study did not investigate the effects of smoke exposure on resultant wines (Sheppard et al. 2009). Therefore, previous studies, to date, have not considered the implication of smoke exposure across the key phenological stages of an entire growing season nor between subsequent growing seasons.

The current study builds on previous research to investigate the effect of smoke exposure of grapevines over three growing seasons. Smoke applications were made over a comprehensive range of phenological stages and the development of volatile phenols and sensory smoke aromas in resultant wines were investigated. This enabled the assessment of treatment effects both within and between seasons.

Materials and methods

Trial establishment

The trial was established on a commercial vineyard (Vitis vinifera cv. Merlot) located at Capel (33.575°S, 115.577°E) in the Geographe region of Western Australia. The trial site was selected because of its remote location from forested areas and infrequent history of smoke exposure. During the 3-year period of the trial, no incidences of external smoke generation and exposure were observed.

Smoke generation and application

Proven methodology previously employed for the application of smoke to field-grown grapevines (Kennison et al. 2009) was utilised in this study. In brief, smoke was produced from the combustion of dry barley straw in a 50-L lidded drum and pumped through outlet piping into tents (6 m long × 2.5 m high × 2 m wide) which enclosed field-grown grapevines (three vines per replicate). Tents were constructed from galvanised steel and greenhouse-grade plastic (Solaweave, Gale Pacific, Braeside, Vic., Australia) that enabled light transmission. The density (30% obscuration/m) and duration (min) of smoke applications were measured using portable nephelometer equipment (VESDA Laser FOCUS™ VLF-250, Mount Waverly, Vic., Australia).

In order to assess smoke application to a range of grapevine growth stages and the reproducibility of this application between seasons, smoke treatments were applied to field-grown grapevines over three growing seasons (2006/2007, 2007/2008 and 2008/2009). In each season, different grapevines were selected for experimental treatments in order to minimise any potential carry-over effects. Experimental treatments comprised the application of smoke to grapevines (in triplicate, three vines per replicate) for 30 min. Smoke was applied to different grapevines at various phenological stages of grapevine growth as designated by the modified Eichhorn-Lorenz (E-L) system (Coombe 1995). The E-L system was further modified for this study to incorporate additional smoke application timings (Table 1). Smoke was applied to field-grown grapevines at least once (n = 1 to 3) at the growth stages corresponding to E-L: 12 (10-cm shoots); 23 (full bloom); 31 (pea-sized berries); 32 (bunch closure); 35(a) (onset of veraison); 35(b) (veraison + 3d); 35(c) (veraison + 7d); 35(d) (veraison + 10d); 36(a) (intermediate total soluble solids (TSS)); 36(b) (intermediate TSS + 3d); 37 (berries not quite ripe); and 38 (harvest).

Table 1.  Key grapevine growth stages from the modified E-L system and additional growth stages used for reference in this study.
Modified E-L grapevine growth stageAlternative interpretation of modified E-L growth stages for reference in this study
Stage numberDescriptionStage numberDesignation
  • Adapted from Coombe 1995. TSS, total soluble solids.

125 leaves separated; shoots approximately 10 cm in length; inflorescence clear1210-cm shoots
2317–20 leaves separated; 50% caps off; full bloom23Full bloom
31Pea-sized berries (7 mm diameter)31Pea-sized berries
32Beginning of bunch closure; berries touching (if bunches are tight)32Bunch closure
35Berry colouring and softening begins; berries begin to enlarge35(a)Onset of veraison
35(b)Veraison + 3d
35(c)Veraison + 7d
35(d)Veraison + 10d
36Berries with intermediate TSS values36(a)Intermediate TSS
36(b)Intermediate TSS + 3d
37Berries not quite ripe37Berries not quite ripe
38Berries ripe for harvest38Harvest

Separate smoke treatments were performed to investigate: the potential sequestration of smoke constituents within the grapevine; the phenological carry-over potential of smoke constituents' compounds; and the recovery of grapevines from one season to another. In order to ensure a heavy smoke application, repeated smoke treatments (n = 8) were applied to the same vines from the onset of veraison to harvest with wine made from fruit of these vines for sensory and chemical analysis. Wine was also made from fruit produced by the same vines in the following grapegrowing season to investigate grapevine recovery and the potential carry-over of smoke aromas and flavours between years.

Winemaking

Wine was produced from all smoke and control treatments in each year of the trial when grapes reached TSS of 21.4 ± 2.9°Brix as measured by refractometry. Fruit (approximately 15 kg) from each replicate (n = 3) of smoke and control (unsmoked) treatments was processed separately to avoid contamination. Fruit was crushed, destemmed and inoculated with Saccharomyces cerevisiae EC1118 yeast (Lallemand Inc., Montreal, Canada) at a rate of 200 mg/L and fermented in 15-L stainless steel fermentation vessels. On average, fermenting musts were plunged twice daily until the wine approached a TSS of 0°Brix before being pressed off skins. Wines were then inoculated for malolactic fermentation with Oenococcus oeni (Viniflora Oenos, Chr. Hansen, Hørsholm, Denmark) and were stored in 4.6-L glass fermenters at 15°C until completion of malolactic fermentation as determined by quantitative malic acid analysis (Iland et al. 2004). Post-malolactic fermentation, wine sulphur dioxide (SO2) concentrations were measured by aspiration (Iland et al. 2004) and adjusted to 30 ppm. Wines were subsequently cold stabilised for 28 days at 2°C. Wines were subsequently filtered (5 µm) and bottled.

Quantitative analysis of guaiacol and 4-methylguaiacol

Gas chromatography-mass spectrometry (GC-MS) was utilised to determine guaiacol and 4-methylguaiacol concentrations in both grape and wine samples using previously reported methodology (Spillman et al. 1997, Pollnitz et al. 2004, Kennison et al. 2008). Guaiacol and 4-methylguaicol were selected as analytes of interest as they have previously been used as indicators of smoke taint (Kennison et al. 2009).

Sensory analysis of wine aroma

Sensory analysis of all wines in this study was conducted by quantitative descriptive analysis (Meilgaard et al. 2007) using a trained panel of eight people (four males and four females). Panellists were pre-screened and selected for experience with wine sensory education (i.e. >100 h), availability, interest in the study, being non-smokers, regular wine drinkers and of good health. Panellists' ages ranged between 21 and 30 years and all were able to detect the aroma of smoke taint in red wines at a predetermined threshold ascertained by Kennison et al. (2007).

Prior to formal wine evaluation, all panellists participated in eight training sessions where they identified and agreed on six descriptive aroma terms and learnt to measure the intensity of smoke aroma on an unstructured 100-point line scale. Formal wine aroma evaluation was conducted on 30 wines (i.e. two replicates of 15 wine treatments) over six sessions each held at the same time of day on different week days. Samples (20 mL) were presented in three-digit coded ISO standard wine tasting glasses that were lidded with glass covers to avoid contamination of other samples and the testing area. Sample order was completely randomised with each panellist receiving a different sample at any one time in order to avoid bias. In order to avoid sensory fatigue, each panellist left the testing area for approximately 10 min to an outdoor environment after evaluating each sample.

Statistical methods

All data was analysed using Genstat 11th Edition (VSN International Limited, Hemel Hempstead, UK). Analysis of variance (ANOVA) was utilised to analyse wine sensory and chemical data at the 5% level of significance (P < 0.05). Further wine sensory data was analysed by Principal Component Analysis (PCA). Chemical data from the phenological smoke application experiments was analysed by the residual maximum likelihood (REML) procedure that was utilised to fit a linear mixed model over 3 years (fixed effect = smoke treatment).

Results

Effects of smoke exposure on chemical properties of wine

The degree of smoke taint, according to guaiacol and 4-methylguaiacol content of wine, varied considerably depending on the phenological timing of grapevine exposure to smoke. Over the three growing seasons, the concentration of guaiacol and 4-methylguaiacol in wines ranged from a low of 0.6 and 0.3 µg/L, respectively, for wines corresponding to grapevine smoke exposure at ‘10-cm shoots', to a high of 60.7 and 14.1 µg/L, respectively, for wines made from vines exposed to smoke at ‘veraison + 7d’ (Figure 1). Control wines, i.e. wines produced from fruit harvested from unsmoked vines, contained either no detectable or trace concentrations of guaiacol (i.e. <2 µg/L) and 4-methylguaiacol (i.e. <0.5 µg/L). As in previous studies (Kennison et al. 2007, 2008), 4-methylguaiacol occurred at lower concentrations than guaiacol in all wines, however, followed similar trends across the three growing seasons.

Figure 1.

Concentration of guaiacol and 4-methylguaiacol in wine made from fruit of vines exposed to a single smoke application at either E-L 12 (10-cm shoots), 23 (full bloom), 31 (pea-sized berries), 32 (bunch closure), 35(a) (onset of veraison), 35(b) (veraison + 3d), 35(c) (veraison + 7d), 35(d) (veraison + 10d), 36(a) (intermediate TSS), 36(b) (intermediate TSS + 3d), 37 (berries not quite ripe) or 38 (harvest). Three separate periods of vine sensitivity to smoke taint uptake are represented by P1 (low uptake), P2 (variable uptake) and P3 (high uptake). Data is from 2006 to 2008, analysed by REML to produce predicted means and standard errors. C, control; n = 3 to 9; error bars show two standard errors of the mean.

On the basis of guaiacol and 4-methylguaiacol concentration in wine, three key periods of susceptibility to smoke exposure were identified within the annual cycle of active grapevine growth. These were period 1 (P1), defined as the period from ‘10-cm shoots’ to ‘full bloom’ when exposure to smoke resulted in relatively low concentrations of volatile phenols observed in wine; period 2 (P2) defined from the stage of ‘berries of pea size’ to the ‘onset of veraison’ during which moderate but variable concentrations of volatile phenols were observed in wine; and period 3 (P3) defined from the stage of ‘veraison + 7d’ to ‘harvest’ during which the highest concentrations of volatile phenols were observed in wine (Figure 1). Average guaiacol and 4-methylguaiacol concentrations in wine were 1.0 and 0.5 µg/L, respectively, for P1; 21.4 and 5.0 µg/L, respectively, for P2; and 48.9 and 8.9 µg/L, respectively, for P3.

Effect of smoke exposure on sensory properties of wine

Trained panellists rated the intensity of smoke-like aromas such as ‘burnt rubber’, ‘smoked meat’, ‘leather’ and ‘disinfectant/hospital’ to quantify the degree of taint present in wine, together with desired wine attributes such as ‘red berry fruits’ and ‘confection’ (Table 2). An ANOVA of sensory data showed the wine treatments to be a source of variation (P < 0.001 to P < 0.05) for all aroma attributes except for the wine aroma attribute of ‘confection’. Panellists were also a source of variation (P < 0.001) in wine sensory data, however, were consistent in their evaluation between sessions. A wine by panellist interaction in the aroma of ‘burnt rubber’ (P < 0.01), ‘leather’ (P < 0.05) and ‘disinfectant/hospital’ (P < 0.05) was also present. Wine replicates showed reproducibility as not being a source of variation except for the wine by replicate aroma attribute of ‘confection’.

Table 2.  Analysis of variance for wine sensory attribute ratings of ‘burnt rubber’, ‘smoked meat’, ‘leather’, ‘disinfectant/hospital’, ‘red berry fruits’ and ‘confection’.
Aroma descriptorWine (W)Panellist (P)Rep (R)W × PP × RW × R
  1. F ratios are shown as sources of variation. Significance indicated by *P < 0.05,**P < 0.01 and***P < 0.001.

  2. Variance is represented for wine (W), panellist (P), replicate (R), wine by panellist (W × P), panellist by replicate (P × R) and wine by replicate (W × R).

Burnt rubber4.922***4.000***2.1332.21**0.811.15
Smoked meat5.844***2.994***0.081.360.651.84
Leather2.342*3.747***1.0521.78*0.331.31
Disinfectant/hospital2.455*9.265***0.8811.89*0.660.53
Red berry fruits2.399*7.236***1.8130.971.351.45
Confection1.8128.161***3.3580.731.202.12*

Panellists determined all aroma characters to be present in the 3-year wine set with wine-like aromas of ‘red berry fruits’ and ‘confection’ in higher mean intensities (44.6 to 44.9) than smoke-like aromas (15 to 18.1; Figure 2). However, while the intensity of smoke-like aromas was more subtle the wines that exhibited higher expression of smoke-like aromas generally also exhibited low wine-like aromas and vice versa. PCA of mean aroma results accounted for 94.4% of overall variation as being comprised of principal component 1 (88.5%) and 2 (5.9%) (Figure 3). Principal component 1 consists of positive loadings on smoke-like aromas of ‘burnt rubber’ (0.42), ‘smoked meat’ (0.41), ‘leather’ (0.41), ‘disinfectant/hospital’ contrasted with negative loadings on wine-like aromas of ‘red berry fruits’ (−0.41) and ‘confection’ (−0.4). Principal component 2 is dominated by negative loadings on all aroma attributes (−0.19 to −0.55) except for the attribute of ‘disinfectant/hospital’ that has a positive loading (0.15).

Figure 2.

Mean aroma intensity descriptor scores of ‘burnt rubber’, ‘smoked meat’, ‘leather’, ‘disinfectant/hospital’, ‘red berry fruits’ and ‘confection’ detected in wines made from grapes of vines exposed to smoke at either E-L stage 12 (10-cm shoots), 23 (full bloom), 31 (pea-sized berries), 32 (bunch closure), 35(a) (onset of veraison), 35(b) (veraison + 3d), 35(c) (veraison + 7d), 35(d) (veraison + 10d), 36(a) (intermediate TSS), 36(b) (intermediate TSS + 3d), 37 (berries not quite ripe) or 38 (harvest). Error bars indicate two standard errors of the mean.

Figure 3.

PCA biplot of mean wine sensory scores (inline image) made from fruit of vines exposed to smoke application at E-L stage 12 (10-cm shoots), 23 (full bloom), 31 (pea-sized berries), 32 (bunch closure), 35(a) (onset of veraison), 35(b) (veraison + 3d), 35(c) (veraison + 7d), 35(d) (veraison + 10d), 36(a) (intermediate TSS), 36(b) (intermediate TSS + 3d), 37 (berries not quite ripe) and 38 (harvest). Aroma descriptors are indicated by arrows labelled BR (burnt rubber), SM (smoked meat), L (leather), DH (disinfectant/hospital), RBF (red berry fruits) and CF (confection). C, control (unsmoked) wine.

The intensity of specific wine aromas was found to vary depending on the phenology at the time of smoke application to grapevines (Figure 3). Wines produced from fruit of unsmoked (control) grapevines and fruit produced from vines smoked in P1 contained dominant wine-like aromas of ‘red berry fruits’ and ‘confection’. Conversely, wines produced from grapevines exposed to smoke at E-L stage 35(b) (veraison + 3d), 35(c) (veraison + 7d), 35(d) (veraison + 10d) and 36(a) (onset of veraison) in P3 were dominated by smoke-like aromas of ‘disinfectant/hospital’, ‘burnt rubber’, ‘smoked meat’ and ‘leather’. Interestingly, wine made from fruit of grapevines exposed to smoke at E-L 31 (pea-sized berries) from P2 had elevated smoke-like aromas, levels similar to those detected in wines made from fruit of vines exposed to smoke at E-L 38 (harvest; Figure 2).

A relationship was also evident between the concentration of guaiacol and 4-methylguaiacol determined by GC-MS and that of the smoke and wine-like aromas in wine determined by sensory analysis. Regression analysis (not shown) revealed a strong positive linear correlation between guaiacol and the smoke-like aroma characters of ‘burnt rubber’ (r = 0.8), ‘smoked meat’ (r = 0.78), ‘leather’ (r = 0.79) and ‘disinfectant/hospital’ (r = 0.87) and a strong negative linear correlation between guaiacol and the wine-like aroma descriptors of ‘confection’ (r = −0.84) and ‘red berry fruits’ (r = −0.88). Guaiacol and 4-methylguaiacol were therefore adequate indicators of the intensity of smoke-like aromas in wine.

Seasonal effect of smoke exposure on grapes and wine

In order to investigate grapevine recovery and the potential for carry-over of smoke constituents from one season to the next, repeated smoke applications (n = 8) were applied to the same vines in a single growing season. The most notable carry-over effect was decreased fruit yield of grapevines in the subsequent season. In the season of repeated smoke application (i.e. year 1), smoked grapevines yielded an average of 11 kg per vine of fruit at harvest, in comparison to control vines, which yielded an average of 15.8 kg per vine (Table 3). In the subsequent season (i.e. year 2), the same vines were not exposed to smoke but continued to produce reduced yields (6.4 kg) and bunch numbers (73) in comparison to the unsmoked (control) vines, which yielded 12.9 kg from 115 bunches.

Table 3.  Concentration of guaiacol and 4-methylguaiacol in wines made from fruit of vines exposed to eight repeated smoke applications from the period of veraison to harvest (2006), in comparison to wines made from fruit of the same vines 1 year post-repeated smoke exposure (2007) and wines from fruit of control (unsmoked) vines.
 Compound concentration (µg/L) in wineYield components (average per vine)
Guaiacol4-methylguaiacolFruit yield (kg)Bunch no.
  1. Means followed by the same letter within columns are not significantly different at P ≤ 0.05, results are per treatment replicate (three vines), mean values are from three replicates. Fruit yield and average bunch number produced from the same grapevines (i.e. for the 2006 and 2007 seasons) are also represented. †Data from Kennison et al. 2009. n.d. = not detected.

Year 1: smoke applied to vines
 Smoke388.3a9311.0a129.3ab
 Control4.3bn.d.15.8b148.3a
Year 2: 1 year post-smoke application
 Smoke2bn.d.6.4c73.7c
 Control0bn.d.12.9d115.3b

However, from a chemical and sensorial perspective, no carry-over effect from repeated smoke exposure was observed in the subsequent season. Sensory analysis of wines made from grapes harvested one season after smoke exposure (i.e. year 2) indicated a low level (not significant at P < 0.05) of ‘smoked meat’ aroma, but these wines also exhibited intense ‘red berry fruits’ and ‘confection’ aromas, i.e. at similar levels to those of the control wine (Figure 4). In the same wines, the concentration of guaiacol (2 µg/L) and 4-methylguaiacol (0 µg/L) also showed no chemical carry-over effect (Table 3).

Figure 4.

Mean intensity ratings of smoke-like aromas of ‘burnt rubber’, ‘smoked meat’, ‘leather’, ‘disinfectant/hospital’ and wine-like aromas of ‘confection’ and ‘red berry fruits’ in smoke tainted wines, recovery effects (1 year post-smoke exposure) and wines made from fruit of control (unsmoked) grapevines. Scale represents 0 = non-detectable aroma to 8 = highly detectable aroma; †data from Kennison et al. 2009.

Discussion

This study builds on past research (Kennison et al. 2009) and demonstrates that the development of smoke taint depends greatly on the phenological timing of grapevine smoke exposure. In particular, the timing of peak periods of smoke uptake was consistent over the three growing seasons. Furthermore, this study has demonstrated that the carry-over effect of smoke exposure between seasons is limited to physiological responses (i.e. yield and bunch number) and there was no evidence of sequestration of smoke constituents by grapevines. To our knowledge, this is the first study concerning the development of smoke taint as a function of phenology for any crop.

The demonstrable link between phenology and the intensity of smoke taint in wine enabled the identification of three periods of susceptibility to smoke taint. The first period (P1) corresponded to a low level of smoke taint susceptibility and wine produced from fruit of grapevines exposed to smoke during this period contained trace levels of guaiacol (<1 µg/L) and 4-methylguaiacol (<0.5 µg/L) only (Figure 1). The dominant sensory attributes of these wines were ‘red berry fruits’ and ‘confection’. The sensory panel gave low scores for ‘burnt rubber’, ‘smoked meat’, ‘leather’ and ‘disinfectant/hospital’ aromas (Figure 2). The possible reasons for the low levels of taint following exposure during P1 are most likely related to the lack of fruit because of the early stage of fruit development rather than the lack of uptake by the vine per se. As P1 occurred prior to fruitset, there was minimal surface area for direct uptake by the fruit. Furthermore, translocation of smoke compounds taken up by the vine would have been limited by the lack of a strong source-sink relationship between the leaves and fruit (Ollat and Gaudillère 2000, Ollat et al. 2002). Likewise, compounds taken up during this time would be expected to be diluted through subsequent vine and fruit growth and may also potentially be lost through volatilisation or degradation in a similar manner to that reported for pesticides (Cabras and Angioni 2000).

During P2, when fruit was present on the vine, increased concentrations of taint were observed in wines, i.e. 21.4 µg/L for guaiacol and 5 µg/L for 4-methylguaiacol. Compared with P1, phenol levels were higher, but absolute levels were variable between seasons and exposure timings. P2 wines exhibited enhanced ‘red berry fruits’ and ‘confection’ characters, but often also showed enhanced ‘burnt rubber’, ‘leather’, ‘smoked meat’ and ‘disinfectant/hospital’ characters. This period comprises the initial period of rapid berry growth via cell division, followed by a lag phase of slowed growth and seed maturation (Coombe 1992, Mullins et al. 2000). During this phase, bunches are a comparatively weak sink for photosynthates until the onset of veraison (Hale and Weaver 1962, Ollat et al. 2002). P2 concludes with a heightened level of smoke taint for wines made from fruit of vines exposed to smoke 3 days after veraison. This may represent a transition from P1 to P3 or even a distinct period in itself.

The highest risk period (P3) for the development of smoke taint corresponded to smoke exposure between ‘veraison + 7d’ and harvest. Elevated concentrations of volatile phenols were measured in wines, being on average, 60.7 µg/L for guaiacol and 14.1 µg/L for 4-methylguaiacol. Wines also exhibited intense ‘leather’, ‘smoked meat’ and ‘burnt rubber’ aromas and were generally disagreeable to panellists. The timing of P3 smoke applications corresponded with the grape berry ripening, i.e. when bunches represent major carbohydrate sinks (Coombe 1992, Ollat et al. 2002). The development of smoke taint during P3 is therefore likely to indicate both direct berry uptake and translocation from leaves to the berry.

Hayasaka et al. (2010a) demonstrated that the marker compound, guaiacol, was assimilated by leaves, conjugated and translocated between leaves and berries. In their study, which incorporated glasshouse-grown vines, the rates of translocation of the extrogenous source of labelled guaiacol were considered to be slow. If translocation is an important contributor to taint accumulation and if the rate is limited, then it is logical that it will be time dependent. Smoke exposure in the late stages of P3 resulted in markedly lower levels of taint (40.3 µg/L guaiacol and 7.3 µg/L 4-methylguaiacol) than smoke exposure in the earlier stages of P3 (60.6 µg/L guaiacol and 14.1 µg/L 4-methylguaiacol). In the same period (P3), there was a strong negative correlation between the timing of smoke application to vines and concentration of both guaiacol (r = −0.866) and 4-methylguaiacol (r = −0.882) in the wine. This indicates that there is likely to be a rate limitation in the translocation of taint compounds to the fruit and this has implications for the timing of harvest relative to the timing of smoke exposure.

Smoke taint was not found to carry-over in wine in the season that followed high levels of smoke exposure although grapevine yield was reduced. The reduced yield from smoke-exposed vines (6.4 kg) was found to be substantially lower in relation to those vines that were not exposed to smoke (12.9 kg/vine). The reduction in vine yield may be related to the negative impact that smoke could have on the photosynthetic capacity of the vine. A short duration (1 min) of smoke has been documented to reduce the stomatal conductance, CO2 assimilation rate and intercellular CO2 levels of Chrysanthemoides monilifera with full plant recovery not achieved for 24 h (Gilbert and Ripley 2002). In grapevines, smoke exposure has been shown to decrease the grapevines' ability to accumulate sugar in grape berries and has produced damage on leaf surfaces as evidenced by necrotic lesions (Kennison et al. 2009). The reduction in grapevine yield may therefore likely be a consequence of the effect that smoke may have on the physiological functioning of the vine.

The current study raises several important implications for the wine industry. Firstly, if grapevine smoke exposure occurs early in the grapevine growth cycle, i.e. prior to flowering, then the intensity of smoke taint in resulting wines is likely to be relatively low. However, if smoke exposure occurs between fruit set and harvest, then there is far greater potential for the development of smoke taint in wine, so the presence of marker compounds should be further investigated in smoke-exposed grapes prior to vinification. Since it has now been established that smoke-derived volatile phenols are conjugated following smoke exposure (Hayasaka et al. 2010b), detection of marker compounds should involve acid hydrolysis of juice samples (as described by Kennison et al. 2008), small-scale fermentations or direct measurement of guaiacol glycoconjugates (Dungey et al. 2011). However, if smoke exposure occurs immediately prior to harvest then smoke taint could be minimised if fruit is harvested as soon as possible after the exposure. To avoid the carry-over yield reduction effect caused by high levels of smoke exposure, consideration may need to be given to retaining higher vine bud numbers in the season following smoke exposure.

Acknowledgements

The authors wish to acknowledge the invaluable assistance of Eric Wootton, Glynn Ward and Richard Fennessy from the Department of Agriculture and Food Western Australia and the Capel Vale vineyard. This research was supported by Australia's grape growers and winemakers through their investment body the Grape and Wine Research and Development Corporation, with matching funds from the Australian Federal Government (Project RD 05/02-3).

Ancillary