Arguing “microbial terroir” from microbe to metabolite

Short: New microbial terroir research provides even more evidence that local differences in yeast and bacteria associated with a vineyard make a difference to wine quality.


Bokulich and Mills of UC Davis* have published a series of papers about the communities of yeast and bacteria that live in various winemaking-associated places and arguing for why differences in those communities – usually differences in place; sometimes differences in time – matter to wine as a finished product. Some of their past work outlines distinctive microbial communities around California and connects those communities to climate at a regional level. Their newest publication, out this week, tries to show microbial distinctiveness at the level of near-neighbor wineries and to connect those microbial profiles to wine composition.**

The article is open-access on MBio but, to be honest, it’s a challenging read if you’re interested in following the methods. The study at first seems to involve a two-way comparison between Far Niente and Nickel and Nickel, both in Oakville, California in the Napa Valley. But both of those wineries pulls grapes from multiple vineyards distributed around Napa and Sonoma, and so the comparison is actually among 13 chardonnay and 27 cabernet sauvignon wines, from different vineyard locations, fermented individually in one of those two wineries.

The first part of the study is about showing that grape musts from each of these vineyard sites have unique microbiomes (including bacteria and yeast populations), though their sample size is too small in this case to convincingly argue for a microbial basis to Napa-Sonoma distinctiveness at the AVA level. Unsurprisingly, the diversity of the microbial population and its distinctiveness decreased as fermentation progressed.

The second part of the study is about connecting initial microbial distinctiveness observed in grape musts to wine composition. The authors – and a bunch of statistics – drew some specific connections between the presence of specific metabolites (i.e. chemical compounds created during fermentation by microbial metabolism) with important sensory implications and the presence of specific microbes. On that basis, they argue (they = the authors and the statistics) that they “demonstrate that the microbial composition of grapes accurately predicts the chemical composition of wines made from these grapes and are therefore biomarkers for predicting wine metabolite composition.” That’s true in the sense that they have – in these wines, in this place – created data that mean that identifying X microbe in a must predicts finding Y chemical in a wine, for a limited number of Xs and Ys. It remains a pretty strong way of making the statement.

To be fair, the authors initial frame this as a proof of concept study. Do geographically neighboring vineyards have different microbiologies that matter to wine chemistry? Yes, in these cases, they do. More can be done to substantiate that point, and to follow up on any number of the other questions this paper raises about what some of the microbes linked to specific wine metabolites but with unknown roles in fermentation are actually doing (if they are, in fact, doing anything at all rather than serving as a marker for something else) to make that link happen.

An important note: the framing of this paper, and some others dealing with microbial terroir, can suggest the idea that terroir is quantifiable, reducible to measurable differences in straight-forward wine chemistry. That’s balderdash. Terroir is about regional character. Quantifiable chemical differences are absolutely part of that character, but so is the human history of a place, the character of the people who live there and who make the wine, and the stories that come along with it. Some of those more nebulous influences surely do translate into chemical differences, but not all of them, or at least not all of them in ways contemporary science has an easy time handling. I have no trouble believing that the stories we tell about the wine we’re drinking produce neurochemical changes that affect the sensorineural mechanics of taste perception and that effectually alter the flavor of the wine. Someday soon our sciences may be sophisticated enough to measure those changes. Someday further away, maybe our sciences will be sophisticated enough not to imagine that those measurements explain away why stories are important, too.


*Bokulich has newly moved to Northern Arizona University per the UC Davis press release, though that move is so new that he doesn’t yet seem to have a web presence at his new institutional home.

**Bokulich and Mills’ work is an interesting complement, along different lines, to the microbial terroir work Dr. Matthew Goddard’s group is doing to understand connections amongst regional populations of Saccharomyces cerevisiae, something he presented at the recent International Cool Climate Wine Symposium in Brighton and about which I’ll be writing elsewhere.

On Palate Press: Old wine research we’re still trying to finish

My piece for Palate Press this month asks what California (proto-Davis) wine researchers were doing in the era before mass spectrophotometers and DNA sequencers and even automated pH meters and all the other fancy stuff wine scientists consider essential today. The short story is that they were trying to figure out what grows best where, and how, which is fundamentally what we’re still trying to do. The long story is on Palate Press.

The long story didn’t have space for me to really geek out over the fun of reading old research articles. I think it’s fair to say that science writing — of the by scientists, for scientists variety — wasn’t as dry then as it is now, not just because antiquated language is quaint but because the distance between normal-talk and science-talk was shorter then than it is now. It’s pretty accessible and often entertaining. There’s the simple, voyeuristic pleasure of being astonished at just how backward they sometimes were, and sometimes at realizing that they weren’t as backward as we tend to assume. And then there’s the higher-order pleasure of making stories by connecting what they were doing to what we’re doing and finding new meaning in both the historical and the modern.

But reading about someone else geeking out over light archival wine reading isn’t near as fun as doing it yourself, and the archives of Hilgardia: a Journal of Agricultural Science from the University of California, including much about wine, are freely available via the University of California Agriculture and Natural Resources Respository. When so much is pay-walled and protected, free access to land grant university resources — not just for subscribers, not just for local winemakers, and not just for the taxpayers of California or even the United States — seems increasingly meaningful, and a good reminder of this massive, excellent, egalitarian knowledge-sharing project we practice through land-grant universities and agricultural extensions. I won’t ask you to excuse my unfashionable patriotism.

Does cross-flow filtration affect wine flavor? No easy answers.

Cloudy wine (mostly) doesn’t sell. Neither does (most) wine spoiled by spoilage microbes that produce off-aromas. And so, while it is entirely possible to remove floating particles that make wine cloudy and (most) microorganisms via careful winemaking without sterile filtration, most winemakers appreciate the extra insurance it gives. Filters are essentially membranes punctuated by lots of little pores: all good stuff should flow through the pores; bad stuff you want to keep out of the bottle should be too big to pass. To reliably remove yeast and bacteria, filters need to have really tiny pores, and the question always is: are those filters excluding stuff other than the microbes, good-tasting stuff that I want the wine to keep? We’re talking .45 μm pores here, an order of magnitude smaller than the diameter of a red blood cell; anything bigger risks allowing bitsy bacteria through.

Pro-filtration folks say that all molecules important to wine quality are much to small to get caught in even these super-stringent filters. Filtration-shy folks say that wine tastes different post-filtration; no matter what molecular measurements say should happen, something is happening. For an excellent discussion of the arguments on both sides, see Tim Patterson’s excellent review in Wines and Vines (updated here, but this one is still behind a pay wall). Bottom line: scientific evidence suggests that while some important molecules could get stuck to the surface of the filter, all the important stuff can pass through safely; dissenters can still taste a difference, and maybe that has something to do with big conglomerates of molecules.

So it stands for conventional filtration. But what about cross-flow filtration, they shiny new-ish solution to some of the conventional process’s major hassles? Conventional filtration points a stream of flowing wine directly at the filter membrane and waits for it to percolate. Enough push needs to be behind that wine stream to keep things moving, but too much push and the force of the flowing wine will rip right through the filter. And the wine must be almost entirely free of particles in the first place, else biggish stuff will cover the surface of the filter, block the holes, and slow down flow. Either way, the filter membrane will need to be replaced, and they’re expensive. Cross-flow filtration instead points the wine stream across the surface of the membrane. Liquid still flows through but with less direct pressure on the membrane, and the constant stream sweeps pores clean of junk, too.

So cross-flow is better than traditional filtration for a few technical reasons. Is it also better for wine quality?

A study addressing that question has just been published (as a provisional draft; it’s not yet appeared in the print journal) with work done by a team from UC Davis. A few published articles have shown chemical analyses of cross flow-filtered wines, but this study is unique and helpful in two ways: 1. The emphasis was on whether a trained tasting panel could detect flavor differences in filtered wine; and 2. Wines were tested not just immediately after being filtered, but from bottle samples taken at intervals out to eight months post-filtration. Kitchen-sink white and red blends were included.

Though the group’s experiments aimed at looking for sensory differences following filtration, their results uncovered something else. The flavor of the unfiltered red wine changed more over time — more earth, less fruit beginning at two months and continuing to the eight-month end point — while the flavor of the filtered red remained more or less constant; in other words, the filtered wine was more stable. The obvious explanation is that the unfiltered wine suffered from some kind of microbial growth after bottling, even though the idea of a UC Davis-crafted experimental wine having microbial spoilage issues does seem strange.

More to the original point: even though chemical analyses showed that filtration decreased phenolics in the wine — filtered reds had lower pigmentation and up to 26% lower tannin levels — the tasting panel didn’t pick up corresponding differences in astringency. That’s surprising. The only explanation offered in the paper is that the magnitude of the change mustn’t have been big enough to be detectable.

In the end, then, this study probably does more to fuel the filtration debate than to help resolve it. Pro-filtration folk can point to filtration’s apparent lack of sensory impact, and to the likely spoilage of the unfiltered wine. Filtration-caution folk can point to the color and tannin changes and say that, even if those changes didn’t affect flavor in this wine, similar changes might indeed be important in other wines. So instead, we have one more example of what may indeed be Rule No. 1 in winemaking: there are no easy answers.