Wine sleuths seek answers to the mystery of ‘smoke taint’

After the smoke had cleared from the October wildfires that ravaged more than 970 km² in Northern California, Anita Oberholster started getting phone calls. Winemakers had harvested most of the grapes in the Golden State by the time the fires began—90% of the tonnage in Napa Valley had been picked, for instance—but some were still on the vines. And winemakers had questions.

Many were familiar with “smoke taint,” the decidedly unpleasant, ashy flavor wines can take on after grapevines have been exposed to smoke from wildfires. But the approximately 1,200 wineries in the area of California where the October fires burned had never experienced blazes quite as intense as these, particularly during harvest season, when grapes are most sensitive. The winemakers calling Oberholster wanted advice on how they could determine whether their grapes had been tainted.

“The more calls I received regarding advice about what to do, the more I realized how much we still don’t know” about how the compounds in smoke affect grapes and the wine made from them, says Oberholster, a wine researcher at the University of California, Davis, who also advises winemakers. Wine chemists in Australia have been studying the issue for more than a decade. But smoke taint hadn’t been as high a priority for California because the wildfire problem hadn’t affected the wine industry as much there, Oberholster says.

What we still don’t know about smoke taint is “all that’s been on my mind” for the past seven months, says Christopher Carpenter, a winemaker for Jackson Family Wines. Some of the vineyard blocks Carpenter presides over in California ripen later in the season because they’re in the mountains rather than the valley and hadn’t been harvested before the October fires rolled in. Since the fires, the grapes from those regions have been fermented, and their resulting wine is now aging in barrels. “In about a year and a half I have to make blending and bottling decisions,” Carpenter says. One type of wine he makes sells for $375 per bottle. So if he’s not 100% confident that the 2017 vintage meets his standards, he won’t damage the label’s reputation by selling it. What’s weighing on Carpenter is that there aren’t yet any analytical tests that can absolutely guarantee him the wine isn’t tainted.

“We still don’t know exactly what combination of compounds causes the different characteristics that we call smoke taint,” so analytical tests can’t be conclusive, Oberholster says. “And if you have smoke taint, we don’t have effective methods that can treat the wine and remove smoke taint permanently.”

The October 2017 wildfires included four of the 20 most destructive in California history. And climate change models from researchers in the U.S. suggest that future wildfires in parts of Northern California could burn up to 40% more land if carbon dioxide levels in the atmosphere double (Clim. Change 2004, DOI: 10.1023/B:CLIM.0000024667. 89579.ed). With that predicted increase and similar estimates from countries like Canada and Australia, more and more researchers—Oberholster included—are joining the effort to find answers.

First encounters

Although Napa and Sonoma, the Northern California counties known for wine making, had never experienced wildfires as intense as the ones in October, popular wine-growing regions in Australia had. Australia has a mostly hot, dry climate, so bushfires are frequent, and often devastating, there. The 2009 Black Saturday bushfires in the southeastern state of Victoria, for instance, killed about 170 people and caused many more to lose their homes. They also cost the wine industry in the region about $300 million in lost revenue due to smoke taint and other damage.

After one particularly bad set of bushfires near Canberra in 2003, vintners in Australia began reaching out to wine scientists for help with the ashy flavors they were tasting in their wines. Researchers from local universities, the Australian Wine Research Institute, and the Department of Agriculture & Food Western Australia responded then and in the ensuing years as more fires cropped up. Researchers like Kerry Wilkinson.

Now a professor of enology at the University of Adelaide, Wilkinson recalls being invited around 2005 to taste wines made by a vineyard that had been in the path of smoke from a nearby “prescribed” burn. Government agencies like the Department for Environment & Water in South Australia periodically burn vegetation to reduce the fuel available and prevent bushfires from damaging residential areas.

The wine that Wilkinson tasted was definitely off. “I remember thinking, ‘It’s not that bad,’ the first one or two sips,” she says. “The winemaker said to keep tasting it. And the more that we tasted the wine and moved it around in our mouths, the more the ashy character built up. After about three or four mouthfuls, you realized you really couldn’t get through a whole glass without knowing there was something wrong.”

Winemakers often intentionally introduce smoky flavor into red wines by aging them in oak barrels that have been toasted with flames. This wine was different. Oaked wine, Wilkinson says, has “more of a toasty, smoky character that comes through with coconut and vanilla spice” flavors. With wine tainted by smoke, she says, “it’s more of a cold ash, acrid, and medicinal-type character. There’s a very clear distinction.”

So Wilkinson and others in Australia started studying how fires impart “off” flavors to wines (Aust. J. Grape Wine Res. 2015, DOI: 10.1111/ajgw.12183). By comparing the compounds found in tainted wines with those in untainted wines of the same grape variety, the scientists determined that the tainted vino contained higher concentrations of volatile phenols—nontoxic compounds like guaiacol and 4-methylguaiacol commonly found in wood smoke. Through their experiments, the researchers also learned that the smoky compounds slip into grapevines through the plants’ leaves and berries, concentrating mostly in the grape skins. A series of other experiments, in which researchers put tents over grapevines in the field and exposed them to intense smoke during certain stages of the vines’ growth cycle, taught the scientists that the plants are most vulnerable to taking up smoke when the grapes begin ripening, closer to harvest time.

All these results seemed straightforward: The grapes absorb smoke compounds at certain, sensitive times, and voilà! Smoke taint. But further investigations revealed that the situation was more complicated.

Hot on the trail

When researchers in Australia monitored the wine-making process using smoke-exposed grapes, they noticed that during the early stages of fermentation, the wine tasted okay. By the later stages, though, the flavor of smoke taint began to rear its ashy head.

Along with Kristen Kennison (now Brodison) of the Department of Agriculture & Food Western Australia, Wilkinson and others first reported this phenomenon in 2008 (J. Agric. Food Chem., DOI: 10.1021/jf800927e). “We analyzed the change in volatile phenol concentrations during fermentation, tracked them every couple of days, and saw they were increasing,” Wilkinson remembers.

At first, this wasn’t surprising because the team was making merlot, a red wine. During fermentation of a red wine, the skins of the grapes stay in contact with the juice to give the wine its color and body. Wilkinson and her team assumed that the grape skins were releasing their volatile phenols, leading to the observed increases. “But then we pressed the wine away from the skins, and we continued to get an increase of phenols in the juice,” Wilkinson explains.

The scientists concluded that guaiacol and its fellow smoke-derived volatile phenols had somehow been masked in the juice and were slowly revealing themselves during fermentation. The most likely scenario was that these phenols had been bound to some other, larger compounds in the wine that couldn’t be readily detected with the gas chromatography/mass spectrometry methods researchers typically use to analyze aroma and flavor molecules. These larger, precursor compounds also would have prevented people from tasting the volatile phenols in the original juice.

Enzymes in the fermentation yeast, the scientists thought, must have been cleaving the phenols from these larger, nonvolatile compounds, making them not only “visible” to GC/MS but also tasteable once again. So Wilkinson and other researchers in Australia did some sleuthing to uncover the identity of the precursor compounds. What they found were glycoconjugates: volatile phenols that had been linked to one or more sugars.

“Adding sugars to molecules is a common mechanism in plants,” says Katja Härtl, a postdoc who studies the biochemistry of fruit-bearing plants like grapevines with Wilfried Schwab at the Technical University of Munich. Plants tack carbohydrates onto molecules in their cells to make the compounds soluble and easier to transport and store. In the case of the volatile phenols from wildfires, the grapevines probably use this mechanism to cope with the foreign compounds, Härtl says. The added carbohydrates enable the plants to transport and sequester the phenols, as well as to block the small molecules’ functional groups so they can’t react and cause trouble inside the plant cells. Härtl, Schwab, and coworkers recently pinpointed a few of the grapevine enzymes, called glycosyltransferases, capable of tacking a single sugar onto volatile phenols like guaiacol (J. Agric. Food Chem. 2017, DOI: 10.1021/acs.jafc.7b01886).

Wine scientists have now confirmed the identity of a few of the glycoconjugates in smoke-tainted wine—monosaccharide versions like guaiacyl-β-D-glucopyranoside and disaccharide versions like guaiacyl-β-D-gentiobioside. Armed with that knowledge, researchers might one day design better techniques for removing the offending compounds—both the volatile phenols and glycoconjugates—from tainted wine.

Currently, methods that can reduce smoke taint are filtration with fining agents like activated carbon from charcoal or filtration with reverse osmosis. But in most cases, none of these nonspecific methods “remove 100% of the nasties” and leave the desirable compounds alone, says Markus Herderich, research group manager at the Australian Wine Research Institute. Activated carbon nonselectively binds all sorts of organic compounds in the wine, not just the ones from smoke taint, and some of them may be important to the wine’s color and desired flavor. Standard versions of reverse osmosis remove small molecules like guaiacol but can leave behind the larger glycoconjugates. If these sugar-linked compounds stick around in wine, studies suggest, enzymes or microbes in a person’s saliva can break them down to unleash the smoky volatile phenols.

To improve winemakers’ options, scientists could design advanced materials to recognize, trap, and remove both the glycoconjugates and volatile phenols from wine. Wilkinson, for instance, says she’s examining cyclodextrins and molecularly imprinted polymers for this purpose. Or maybe a fermentation yeast could be genetically engineered to stick extra sugars onto the glycoconjugates, making it more difficult for enzymes to unleash the phenols—and their flavor.

But what if there is more to smoke taint than glycoconjugates and volatile phenols? Chemists in British Columbia have evidence that the story doesn’t end there.

The plot thickens

Graduate student Matthew Noestheden and professor Wesley Zandberg at the University of British Columbia, Okanagan, began investigating smoke taint a few years ago. Supra Research & Development, a local analytical services firm, approached them to better understand the problem and to design improved smoke taint detection methods that could help winemakers. Supra now funds Noestheden’s research through an industrial fellowship.

In the valley where UBC Okanagan sits, summers are hot and dry, making the region vulnerable to wildfires. Smoke from fires in Washington state also blows into the area frequently and settles into the valley, blanketing the vineyards there.

Zandberg and Noestheden wanted to develop a quick-turnaround analytical test that could detect both the smoke taint phenols and the glycoconjugates, not just in wine but also in potentially tainted grapes while they’re still on the vine. Vineyards typically have crop insurance that protects them from things like weather, pests, wildlife damage, and smoke taint. But to claim that insurance money, winemakers need to have proof that their crop is spoiled—before plucking it from the vines.

Complicated liquid chromatography/mass spectrometry methods are available to detect glycoconjugates directly, but Adelaide’s Wilkinson says many analytical testing companies are not yet set up to use them. Some firms test for the glycoconjugates in grapes and wine indirectly by using either enzymes like β-glucosidases or strong acids to break the volatile phenols and the carbohydrates apart, making the phenols detectable by GC/MS. Still, other companies don’t bother with the glycoconjugates, testing for only guaiacol and 4-methylguaiacol not bound to sugars, which they argue are adequate indicators of smoke taint. All these services can cost a few hundred dollars to analyze samples of wine and grapes and can take from a couple days to more than a week to return results.

Through careful testing with GC/MS and LC/MS techniques, Noestheden and Zandberg designed and fine-tuned an acid digestion method they say breaks down all the glycoconjugates in a sample that contain simple sugars like glucose. They first measure the free volatile phenols in their grape samples and then, after the digestion, the ones that had been bound. Their technique analyzes not just guaiacol and 4-methylguaiacol but also other prominent volatile phenols in smoke, including syringol and m-, o-, and p-cresol (J. Agric. Food Chem. 2017, DOI: 10.1021/acs.jafc.7b03225).

But the two noticed a few peculiar things when later using their protocol to monitor smoke-exposed grapes during fermentation. One was that even though some of the wines Zandberg and Noestheden made from those grapes tasted tainted, the researchers didn’t observe high levels of glycoconjugates, as many studies in Australia had before them (Food Chem. 2018, DOI: 10.1016/j.foodchem.2018.03.097). It’s possible that different soils, climates, smoke types, and other factors could explain the differences in findings, Noestheden says. The other peculiar thing was that during fermentation, volatile phenol concentrations increased, as expected. But so did the concentration of glycoconjugates (J. Agric. Food Chem. 2017, DOI: 10.1021/acs.jafc.7b04946). “What we think is happening is that even larger compounds are breaking down as a consequence of fermentation” and leading to the increase in glycoconjugates, Zandberg says. “But we haven’t proven anything yet.”

For wine researchers, the news that an unknown class of compounds may also be involved in smoke taint is exciting, Wilkinson says. But all the twists and turns in what’s amounting to a wine detective story can be frustrating for industry, she acknowledges. “We don’t have all the answers for them just yet.”

As Carpenter sits in California fretting over his 2017 wines, he holds out hope that by the time he’s ready to bottle, someone will have nailed all the smoke taint culprits and come up with a conclusive analysis method. In particular, the method should be one that the wine industry all agrees on to evaluate the bound forms of the volatile phenols, whatever they may be, he says. “Somebody would make a lot of money with it.”.