The Pseudo-Science of Investigating Terroir (and Why Some Reports May Prove the Opposite of What They Think)

I am fed up with reading reports in the press that scientists have shown a role for microorganisms in terroir. I don’t know which is worse: the lack of logical analysis in the original scientific papers, or the uncritical acceptance of the conclusions by the press.

The running defect in all these papers is a misunderstanding of the meaning of terroir. The concept of terroir is simple enough: fruit grown in one place will have consistently different characteristics from fruit grown in another place. In the case of grapes, this translates to differences in wines according to their origins. But a crucial feature is consistency: it does not prove a basis for terroir to show a difference between grapes from different sources, but that difference has to be persist over multiple vintages.

Whether local microbial populations – in particular yeast, which are responsible for creating most of the flavors of wine during fermentation – contribute to terroir is the subject of paper just published by a group led by Sarah Knight at the University of Auckland.[i] It goes so far as to put “evidence for a microbial aspect to terroir” in the title. But the work is fatally flawed.

The approach superficially seems controlled: yeast were collected from different vineyards and then used to ferment examples of the same batch of sterilized juice from Sauvignon Blanc grapes. Differences in the resulting wine must be due to the yeasts. The analysis rests upon previous work from the same group that showed genetic differences in S. cerevisiae populations in different locations in New Zealand.[ii] Ignoring the fact that only 4 yeast genotypes were found on grapes (potentially able to affect the properties of wine) while the remaining 21 genotypes were limited to soil, this makes the unsurprising point that there are variations in natural yeast populations.

Each isolated yeast population gave wine with a different profile of volatile compounds. Talking about reinventing the wheel! There have been too many demonstrations to count that different cultivars of yeast can have profound effects on the character of wine. Indeed, this is the basis for a sizeable industry in which yeasts selected to emphasize (or de-emphasize) particular aspects of wine character are available commercially. It’s almost trivial to choose yeast to bring out herbaceous character in Sauvignon Blanc, for example.

The logical error in the paper is to conclude that the observations demonstrate any role for microbes in terroir. Taking a single snapshot entirely misses the point, because the population of yeast may be quite different the next year. According to Ribéreau-Gayon’s authoritative book on viticulture and vinification: “In a given vineyard, spontaneous fermentation is not systematically carried out by the same strains each year; strain specificity does not exist and therefore does not participate in vineyard characteristics. Ecological observations do not confirm the notion of a vineyard-specific yeast.”[iii] Surely it behooves any new work to explain why this wrong?

A couple of years ago, a group led by Dr. David Mills at the University of California showed directly by DNA sequencing that different microbes were present on the skins of grapes in different vineyards.[iv] That work was fatally flawed because most of the microorganisms were what are known as spoilage organisms, and are probably not part of useful fermentation (see The Answer to Terroir Does Not Lie on the Skin).

After that, a group led by Régis D. Gougeon at the University of Dijon sampled two vineyards in Burgundy and claimed that they could find differences in both grapes and wine, although differences between vintages were more significant than differences between vineyards. I reckon their samples were too small to be significant because they analyzed only 100 berries from each vineyard, but anyway it’s interesting that vintage was a bigger effect than origin (see Will people please stop trying to prove terroir exists. It’s more useful to look for gold at the end of the rainbow).

These studies are naively touted in the press as showing the involvement of microbes in terroir, but to date there is really no evidence at all: in fact, if the microbes vary significantly from year to year, they may dilute the effect of terroir rather than contributing to it. One could make a career of debunking these studies, which could provide a really good exercise for students of science in how to misinterpret uncontrolled studies.


[i] Sarah Knight, Steffen Klaere, Bruno Fedrizzi & Matthew R. Goddard.Regional microbial signatures positively correlate with differential wine phenotypes: evidence for a microbial aspect to terroir. Nature Scientific Reports 5, 2015, doi:10.1038/srep14233.

[ii] Sarah Knight and Matthew R Goddard. Quantifying separation and similarity in a Saccharomyces cerevisiae metapopulation. (The ISME Journal (2015) 9, 361–370; doi:10.1038/ismej.2014.132; published online 25 July 2014).

[iii] Pascal Ribéreau-Gayon et al., The Handbook of Enology, Volume 1, The Microbiology of Wine and Vinifications, 2nd edition (John Wiley & Sons, New York, 2000), p. 46.

[iv] Nicholas A. Bokulich, John H. Thorngate, Paul M. Richardson, and David A. Mills. Microbial biogeography of wine grapes is conditioned by cultivar, vintage, and climate (Proc. Nat. Acad. Sci. USA doi/10.1073/pnas.1317377110).


Even the Bad Times are Good: Mastering Sauvignon Blanc in Sancerre and Bordeaux

Visiting France in the Spring of 2013, it seemed likely it would be a difficult year: it was cold everywhere and bud break was substantially delayed. Things never really caught up during the growing season. Harvest was small and just about achieved ripeness.

The previous year had been difficult in many places: conditions in the Loire and in Bordeaux were somewhat similar overall, with rain in July, dry conditions in August and September, and then rain again. Harvested earlier, the whites may have come off better than the reds in Bordeaux. At the eastern edge of the Loire, Sancerre and Pouilly Fume produced quite rich 2012s, better than 2011 where there had been problems with rot.

At tastings of the 2012 and 2013 vintages in Sancerre and Pessac-Léognan, I was struck by the comparison between the regions and the vintages. The whites from Pessac offer a fascinating contrast with Sancerre. They range from 100% Sauvignon Blanc, which should be more or less directly comparable, to wines with up to 50% Sémillon. The more common use of oak in Bordeaux tends to soften the wines; and where new oak is used the flavor profile is quite distinct at this young age. Where Sémillon is high there is more of a nutty texture.

But Sancerre is no longer as distinct from Bordeaux as it used to be: the consequence of greater ripeness is that there are Sancerres where the fruits point more to peaches and apricots than citrus. Today, effectively Bordeaux and Sancerre each show a range of styles, with quite a bit of overlap. The old image of grassy herbaceousness has definitely gone.

Bordeaux tends to be citrus driven when there is 100% (or close to it) Sauvignon, with only occasional notes of grassiness. With Semillon the fruits become rounder and more stone-driven. But Sancerre is achieving a similar effect through greater ripeness. The richest Sancerre might be confused with Bordeaux; and some of the 100% Sauvignons in Bordeaux might be confused with Sancerre.

There seems, at this stage anyway, to be more of a distinct difference between 2012 and 2013 in Pessac than in Sancerre. In Pessac, acidity is noticeably more pressing in 2013, with fruits tending towards lemon and grapefruit, whereas 2012 gives more of a stone fruit impression.

In Sancerre, I noticed a great difference as to whether the current wine on offer was the 2013 or 2012 vintage. The difference was not so much in the intrinsic quality of the vintages (barrel samples show that 2013 was actually quite successful in Sancerre) or even the fact that the 2012 had had a year’s extra aging. The real point was that producers whose current vintage was 2013 had bottled it after no more than about four months on the lees; whereas producers who had not yet bottled 2013 and whose current vintage was 2012 had usually given the wine around eight months on the lees. The extra complexity from longer exposure to the lees was really evident. Is this the major difference between the artisanal and commercial approach?

The High Life: Wine at 37,000 feet

Flying transatlantic isn’t a great opportunity for fine wine and dining, but there really isn’t much else to do on the flight besides eat, drink, and sleep, so I thought I’d make an assessment of the present state of the high life. Flying American Airlines from New York to London, en route to start research in Bordeaux for my book on Cabernet Sauvignon, my assessment got off to a poor start in the lounge before takeoff, when the sparkling wine was Gloria Ferrer Brut from Sonoma. Judged by the taste, Brut is clearly a misnomer. Sweet to the point of being sickly, the monotonic palate had a strong taste of green apples (although without the matching acidity you might expect), and if you had told me it was a sparkling apple cider, I would have been hard put to argue. There are some fine sparkling wines made in northern California, but this is not one of them, and proved to be a sad harbinger for what was to follow.

Things improved briefly after takeoff when the Champagne was Pommery. Now this has never been one of my favorites – it always used to strike me as too thin and lacking in fruit – but it has definitely improved since Vranken acquired the brand name in 1990. I don’t know whether taste is affected by the low pressure at altitude, but this now seems to be a respectable, if rather ordinary sparkling wine. There’s not much character to it, and the dosage is just a touch too high for my taste – I wonder whether my impression that dosage has been increasing is right or whether my palate has changed – giving an impression that sugar is being used to compensate for lack of flavor interest. The wine seems essentially uninteresting and its flat flavor profile gave me some trouble in trying to find descriptors for a tasting note. You don’t expect originality from Grand Marque Champagnes, but I still think Pommery could do a better job to disguise its mass produced origins.

The white wines offered a choice between L’Ecole No. 41 Chardonnay from Washington State and Thilion Torbato Sauvignon Blanc from Sardinia. I had some trouble distinguishing them. If I were to be unkind I would say that the Chardonnay was a forlorn attempt to achieve the New World style. The palate has been loaded with oak to disguise the lack of ripeness in the fruit. The oak flavors stand aside from the fruits, and if I didn’t know that the wine had been aged in barriques I would have wondered about the use of oak chips The oak gives a hard, disjointed, phenolic note to the finish. This is one of those rare wines that would have been improved by a shorter finish, as what mostly lingers on are those disjointed oak phenolics.

Despairing of the Chardonnay, I turned to my wife’s Sauvignon Blanc (actually a blend of Sauvignon Blanc with the indigenous grape Torbata), although its relatively deep golden color made me feel suspicious even before I tasted it. A sniff made things worse. Instead of the expected grassiness or herbaceousness came a sort of slightly astringent citrus note. Maybe this is due to the Torbata, which is supposed to have a smoky aroma. Again the palate was loaded with harsh phenolics. I would have placed this as an aromatic variety in a blind tasting, but I think I would have had some trouble recognizing Sauvignon Blanc in it. I wonder whether I would have been able to identify the wines, if I’d been given them blind and told that one was Chardonnay and one was a Sauvignon blend. It wouldn’t be easy to find varietal typicity in this pair, but perhaps the greater acidity and aromaticity would identify the Sauvignon.

On to the reds, where we tried the Villa Mount Eden Cabernet Sauvignon from Napa and the Tres Picos Garnacha. Now I remember a period in the late 1970s when Villa Mt. Eden had a great reputation. In fact, based on Robert Parker’s recommendation, I bought a case of the 1978 vintage and drank it for several years. It never achieved greatness, but was still holding up quite well in 1993 when I had the last bottle. Sic transit gloria mundi. I do not think the old style is remotely recognizable in the current wine. When I swirled and smelled the wine, I wondered if there had been a mistake as the aromas of black cherry fruits with some piquancy came up at me. No, I decided on smelling my wife’s wine, which was even more aromatic and piquant. Following to the palate, the Cabernet showed no more typicity of Cabernet than the Chardonnay had of Chardonnay. The only note of relief was an oaky vanillin that seemed artificial. I’ve never really thought of Cabernet Sauvignon as an intensely aromatic variety, but after this wine I might have to change my opinion. (To be fair, you can find some high-end Napa or Barossa Cabernets with fairly distinct aromatics, but although I don’t usually like the wines, at least I can recognize a matching concentration and fruit intensity that hangs together.)

After this, I approached the Grenache with trepidation. The only information American Airlines provided about its origins is that it is produced in Spain. It turns out to come from the Campo de Borja DO, just to the south of Navarra, where Grenache is the principal grape. I have to disagree with Robert Parker’s high ratings for this wine. Aromatic and piquant on the nose, it followed through to the palate with bright red cherry fruits and a piquancy that made me wonder about acidification, with a slightly sickly nutty end to the finish.  But I have an idea. Add a little sugar and the profile would be perfect for a dessert wine. (After this, I decided not to sample the dessert wine, which however seemed to be a perfectly respectable vintage port.)

I can’t completely exclude the possibility that my palate was out of whack at 35,000 feet, but at all events the common feature of these wines seemed to be excessive striving for intensity. Subtle they ain’t. Even at the crunched price point – I calculate that if every passenger had a glass of wine American would be spending about $2.00 per passenger – there could at least be more variety of choice.

The food was better than the wine but not by a large margin. In all the years I have been on American Airlines, the food has never been up to much. Ranging from barely edible to inedible, sometimes it strikes me as unfit for human consumption (well, consumption by this human, anyway). In the past year or two there’s been some slight improvement. There was always a tendency on American to make the food highly spiced – just what you want at 35,000 feet where you tend to get dehydrated anyway – I assume to disguise the poor quality of the ingredients, but fortunately that phase seems to have passed. Of course, the days of  Krug and caviar on transatlantic flights are long since gone, but surely they could do better than to serve dried out hot meals. I’d settle for cold salads made from better ingredients any time (but I guess the bean counters won’t wear it). In any case, I’ll leave the last word to an American flight attendant, who some years ago said to me, “We’re not fine dining, we’re transportation.”