The Australian Wine Research Institute

Understanding and mitigating the impact of smoke exposure on grape and wine composition and wine sensory properties

Project summary

A major aim of this project is to determine the relationship between the concentration of smoke-derived compounds (phenols and phenolic glycosides) in smoke-exposed grapes and wine with the intensity of smoke characters in wine assessed through sensory evaluation. The consequences of early-season and later smoke exposure will be assessed, with multiple regions and both white and red grapes targeted with low, medium and high levels of smoke compounds. Wines will be made under standardised conditions and assessed using a screened, qualified smoke sensory panel. The sensory intensity of smoke characters will be related to consumer preference response. The role of specific key phenol compounds in smoke characters will be confirmed through reconstitution sensory studies, targeting low to moderate levels of smoke characters.

The project will also assess whether, following early-season exposure, the smoke phenolic glycoside concentration of unripe grapes and/or of grapevine leaves can be used to determine the extent of the smoke characters that would be expected in wine.

The effectiveness of treatments to remove or eliminate taint will also be assessed, notably activated carbon products, as well as enzyme preparations and specific yeast strains.

The project has attracted $100k co-investment from the South Australian State Government, in support of research towards an improved understanding of the consequences of early smoke exposure on grapes at harvest and wine composition and sensory attributes.

Latest Information

Early-season smoke exposure in the Adelaide Hills 2019/2020 season
It is now well known that smoke can contaminate wine-grapes via the accumulation of volatile phenols and glycosides, which can contribute to undesirable ‘smoky’ aromas and flavours in the resulting wines. Model smoke experiments conducted in the early 2000s concluded that at the early stages of grape ripening, when grapes were still hard and green, the risk of smoke taint was low to variable, based on measurement of guaiacol and 4-methylguaiacol, which were the smoke markers available at the time.

During late 2019, the Adelaide Hills region was affected by one major smoke event, which coincided with the early stages of grape ripening (average growth stage E-L 27), and the region was clear of smoke for the rest of the ripening period. This presented a unique opportunity to test the effect of early-season bushfire smoke exposure on wine-grapes and the wines made from them.

Chardonnay, Pinot Noir and Shiraz grapes were sampled from 24 sites across the Adelaide Hills, representing a range of suspected smoke exposure from non-smoke-exposed sites through to vineyards that had experienced direct fire damage but were still able to produce a crop. Replicated berry sampling commenced in mid-January 2020, four weeks after the fire started, and the grapes were harvested in March and made into wines at 50 kg scale. Standard winemaking protocols were applied: Chardonnay was pressed off skins and the clarified juice fermented; Shiraz and Pinot Noir wines were made in a full-bodied style with skin contact. PIRSA and Wine Australia contributed to the trial design, which they also supported with funding.
Smoke exposure early in the growing season clearly led to elevated phenolic glycosides in grapes at harvest for most of the smoke-exposed vineyards studied, corroborating the use of phenolic glycosides as smoke exposure markers during the early stages of grape ripening. The concentration of phenolic glycosides in early-season grapes was not diluted by the increase in berry size during ripening, contrary to expectations.

A range of concentrations of phenolic glycosides were quantified in the berries at harvest, and later compared with sensory ratings of smoke flavour intensity in the wines. One vineyard very close to the fire front in Woodside (labelled sample 12 in Figure 21), which might have been expected to be smoke affected, did not have elevated phenolic glycosides at harvest. This vineyard was reported by the vineyard manager to have escaped the full force of the smoke plume due to a wind change. Results from analysis of all three varieties at this vineyard were consistent with grapes that had not been exposed to smoke, and this was backed up by the chemical data and sensory assessment of the wines. On the other hand, results from a vineyard located 4 km away from the fire scar to the north demonstrated elevated concentrations of phenolic glycosides (sample 14 in Figure 21), and Pinot Noir wine made from that site had distinct ‘smoky’ flavours, confirming smoke exposure. These examples showed that smoke exposure may not correlate directly with proximity to the fire. Hence it is important to test for presence of smoke exposure markers in grapes from individual sites and not make generalisations about regions that have been exposed to smoke.

A relationship was seen between the concentration of taint markers in fruit early in the season and sensory assessment of smoke characters in the wines made from such fruit after harvest. The ranking of marker compounds early in the season matched the ranking of smoke characters in the wines (Figure 21). This relationship was particularly strong for the lighter Pinot Noir wines. While the ranking of marker compounds early in the season matched the ranking of smoke characters in the Shiraz wines in general, dominant ‘green’/‘eucalyptus’ characters seemed to mask the ‘smoky’ flavour in one wine, particularly when assessed six weeks after bottling.

For the Chardonnay wines the ranking of smoke marker compounds early in the season matched the ranking of smoke characters in the wines, but the wines were rated much lower in ‘smoky’ flavour than the red wines. This likely reflects the winemaking process for white wine, which involves early removal of the grape skins, resulting in minimal skin contact, whereas red wine production involves skin contact, aligning it more closely with the grape analysis which involves extraction from all grape components, including skins.

Lessons from the Australian 2019/2020 bushfire season beyond the Adelaide Hills
A number of sample sets have been created to understand the sensory impact of smoke exposure on wines made without any remediation treatments. These include 23 wines from Adelaide Hills wines, 42 wines from various regions across south-eastern Australia, 110 micro-ferments from the Adelaide Hills and 41 micro-ferments from various regions. Sensory and chemical analysis have been completed for all samples.

A dilution/blending study to assess the consumer rejection threshold for ‘smoky’ characters in Chardonnay was conducted with 124 consumers. The results showed that consumers responded negatively to blends with 50% or more of the smoke-affected Chardonnay in a clean control wine, with many consumers also giving low liking scores to the 25% blend. A similar study with a full-bodied red wine is planned for next year.

Smoke flavour intensity in wine was generally related to the concentration of guaiacol, cresols and glycosides. The smoke markers were highly correlated, with some varietal differences, in line with previous observations. No wines had ‘smoky’ flavour that could not be explained by the measured volatile phenols and their glycosides, although some attributes that can be confused with smoke were identified. Masking of smoke by ‘green capsicum’, ‘eucalyptus’ and ‘tropical’ notes was also observed.

Industry support activities
Through a partnership with Wine Victoria and funding from the Victorian Government, the AWRI developed a process for establishing sensory assessment panels in four key regions. Wine benchmarking test kits were prepared and distributed to producers and regional associations, which included wines with various levels of smoke compounds, provided with associated chemical and sensory data. A study was also undertaken to cross-validate results between the two principal laboratories in Australia that offer smoke taint analysis, but use somewhat different methods. A report on the results of this study is available from the smoke page on the AWRI website. Further trials are also underway testing a range of winemaking techniques and processing aids for remediating smoke-affected wine, the outcomes of which will be presented at a series of sensory workshops for winemakers. The smoke marker compounds background database, which started in 2011, has been further expanded to include an additional five varieties important in Victorian viticulture. A similar activity is also underway in NSW, with a focus on five varieties important in NSW regions. This project, funded by New South Wales Wine, is also creating a dataset from the grape chemical analytical data from AWRI helpdesk investigations.