Growing & Winemaking


Terroir and Other Myths of Winegrowing

April 2016
by Mark A. Matthews

Editor’s note: Terroir and Other Myths of Winegrowing explores popular winegrowing concepts including yield, berry size, vine balance, a critical ripening period, and terroir. The book is currently available from the University of California Press ( The following excerpt from Terroir and Other Myths of Winegrowing explores popular concepts related to wine quality and crop yield.

What is the Expected Relationship of Grapes and Wines to Yield?

By Mark A. Matthews

Two of the most widely accepted articles of faith in winegrowing are that: 1) low crop yields and 2) small berries are key factors in producing the best wines. Both winegrowers and the popular wine press frequently invoke the high yield–low quality (HYLQ) and big bad berry (BBB) concepts when discussing wine quality in general, or with respect to specific wines.

A visual starting point for examining the HYLQ paradigm is a linear decrease in quality with increasing yield (see Curve 1 in “Hypothetical Relationships of Quality Vs. Yield or Berry Size”). In this straightforward hypothetical (suggested by many proponents of the HYLQ concept), any increase in yield will result in a sacrifice in fruit and wine quality—but this purported relationship is merely a starting point. The relationship between yield and quality could take many forms, including a very steep slope or an almost flat line, the latter indicating little or no dependence of fruit or wine quality on yield.

Some claim to have more specific knowledge of the shape of this supposed relationship or curve. According to Karen MacNeil, author of The Wine Bible, “The relationship of yield of grapes to wine quality is extremely complex and nonlinear.” No further explanation of the complexity is offered with that statement, but potentially important insight is given in her introductory comments about yield and quality: “We do know this: For every vineyard, there is a breaking point—a point where too many grapes will cause the vineyard to be out of balance, and where the subsequent quality of the wine will plummet.”

Wine journalists James Halliday and Hugh Johnson include a short chapter dedicated to yield and quality in The Art and Science of Wine, in which they claim, “However yield is measured, there is no question of relating it directly to quality.” Apparently by “directly,” they mean something similar to MacNeil, as they cite the “the French rule of thumb” of 3.7 tons per acre as a yield not to be crossed when attempting to produce fine wine. Whereas MacNeil sees this threshold as vineyard-specific, Halliday and Johnson report this as a broader overall principle indicating, for example, that high-quality German Riesling arises from yields similar to those required for Chardonnay in Burgundian vineyards. It is interesting to note that the Napa average yield since the 1970s is curiously close to the 3.7 tons per acre standard suggested by Halliday and Johnson. The question is whether trial and error has worked out a fundamental truth, or if a herd mentality is in effect.

Curve 1 represents a direct (linear) dependence of fruit and wine qualities on yield or berry size. Curve 2 represents quality as relatively insensitive to yield or size increases up to a point, beyond which quality falls. Curve 3 represents quality as relatively insensitive to high yield or big berries, but below some yield or berry size quality becomes increasingly positive with further decreases in yield or size.

The hypothetical breaking point—whether for wine grapes in general or for a vineyard—is represented by Curve 2 in “Hypothetical Relationships of Quality Vs. Yield or Berry Size,” in which the real action in loss of quality occurs above some specific relatively high yield point. If the relationship of quality to yield indeed follows Curve 2, reducing crop below “60” would have no effect on the wine, and hence sacrificing yield would hardly be the virtue that it is generally held out to be (in contrast to Curve 1, where every reduction in yield brings an increase in quality). Additionally, if that hypothetical threshold of Curve 2 were known, producers could exploit this knowledge to produce a higher volume of similarly fine wines rather than needlessly restricting production—a practice that itself often costs money.

If, alternatively, the quality response is that shown in Curve 3, yields above some intermediate value will all produce similar and relatively low-quality wine, and only at very low yields would each incremental decrease lead to an increasingly better wine. The extreme importance of low yields for obtaining high quality that is proclaimed by some wine writers and producers (and printed on the back labels of some California wines) is reflected in Curve 3. Because the yield must be so low, very little of the high-quality wine could be produced and sold. Curve 3 expresses a very different relationship between yield and quality than that described by MacNeil for vineyards or suggested in a more general sense by Halliday and Johnson.

MacNeil gives a measured evaluation of the role of yield in her glossary, noting several caveats with respect to the HYLQ model. Yet, when discussing the designation of Grand Cru vineyards in Alsace, France, MacNeil examines “what, if any, limits should be set on a Grand cru’s yield. Clearly, the stricter the requirements, the more impact and validity the designation Alsace Grand cru would have.” She continues: “A good property (producing Merlot in St. Emillion/Pomerol) may well do an additional summer prune to further restrict yield to make only, say, 3,000 quarts of wine for every 2.5 acres. A basic Merlot from California’s Central Valley, France’s Languedoc or Italy’s Veneto may have been made to about three or four times this yield, and the taste is obviously stretched accordingly.”

Halliday and Johnson also note caveats with respect to the HYLQ relationship, including hail and rot. Yet, in general, these phrases “the stricter the requirements, the more impact and validity” and “stretched accordingly” imply a direct dependence of quality that arises from increased concentrations of good flavors in fruit from lower-yielding grapevines—like that of Curve 1. Claims that low yield is key to wine quality imply a relationship unlike Curve 2. Some high-end wines in California carry back labels touting their extremely low yield, which implies a relationship consistent with Curve 3 or possibly Curve 1.

The complexity of yield
Although the nature of the yield-quality relationship is fundamentally important to those in the business, there has been surprisingly little direct effort to resolve it. Crop yield in grapevine production depends on many factors over two seasons: varieties, weather and cultural practices. The propensity to initiate flower clusters and the number of flowers in a cluster varies among wine grape varieties, but also depends on environmental conditions including but not limited to temperature, light on the developing bud and vine water status. Fruit set (the fraction of retained flowers) of any variety is sensitive to environmental factors such as untimely rain and low temperatures. Vine row spacing (established at planting) affects yield rather directly. Once a vineyard is established, pruning takes place before the season begins, and shoot and cluster thinning during the season is the main means of regulating yield. Water and nutrient supply are usually regulated with the objectives of desirable vine growth and fruit quality, and therefore they impact yield as well. Because the sources of yield are so varied, the task of determining the facts about HYLQ is enormous.

It is important to note that the HYLQ concept tells us that whether yield is reduced by winter pruning to lower shoots per vine, summer pruning (or cluster thinning) or poor fruit set, each action is said to have the same positive impact on flavor intensity. Thus, in the popular press, it is the yield per se that determines the fruit quality. If great winegrowing is as simple as getting as far to the left along Curve 1 or 3 as possible, then it is hardly a challenge worth appreciating. In the following sections, we will turn to the empirical evidence related to various means of altering yield, keeping in mind that with so many paths to yield, it could turn out that none of the hypothetical scenarios for quality response to yield are a reliable generalization.

Applying the HYLQ concept interchangeably among varieties ignores the biology of what makes one variety distinct from another. There is no biological reason or precedent in other crops that should lead us to assume that more yield in a variety is linked inversely and inescapably to lower concentrations of flavors in the fruit when compared to a lower yielding variety. For example, Petit Verdot is more fruitful (produces more clusters) than Malbec and many other red varieties, yet it has much more red color in the fruit and wines. It is likely that varieties differ in their responses to crop load, but the limited empirical evidence available thus far does not support a high sensitivity in any particular variety. We like wines so much that we try to grow wine grapes everywhere, including locations to which some popular varieties may not be best suited. It is possible that part of what has been attributed to a variety’s sensitivity to yield has as much or more to do with the variety interaction with the soil and weather aspects of the environment, about which there is still much to be discovered.

Pruning and cluster thinning
Pruning and cluster thinning are the most direct and common means for manipulating yield in winegrowing, and they can be adjusted each season. A third means is shoot thinning, although this technique is also used to manage the canopy microclimate. When the HYLQ concept is raised in discussions of vineyard management, pruning and cluster thinning are usually the implied subjects.

The HYLQ myth did not come into existence by mistake; there are studies dating back as far as 1904 containing results and interpretations consistent with HYLQ predictions, but the number and history of studies with contrary results are quite surprising.

Investigations of putative yield effects on wine sensory attributes evidently began in the 1950s at the University of California, Davis (UC Davis), long after the HYLQ myth had assumed dogma status. Professors Maynard Amerine, Robert Weaver and others conducted field trials with varied pruning, cluster thinning, or both. They then measured vine yields, followed a consistent protocol to make the fruit into wine and used the recently developed sensory methods to evaluate wine quality. Combining several vintages (1949-51), their data show little or no response of wine scores to wide changes in yield (up to a fivefold range of yields) of Alicante, Carignane, Grenache and Zinfandel; however, wine scores for Zinfandel may have decreased at highest yields per vine. In 1961, Amerine and Weaver reported on another set of experiments with Grenache and Carignane that appear to show more sensitivity of wine scores to yield but at much higher yields. The authors concluded, “The final wine rating was just as high in over-cropped Carignane vines as in normal-crop fruit, and the difference was only slightly in favor of the normal crop in Grenache.”

Brian Freeman and Mark Kliewer conducted studies in the early 1980s that manipulated yield via irrigation and pruning in both Shiraz and Carignane. The large study on Shiraz in Australia, in which severe pruning (i.e., dramatic reduction in yield) in fact decreased or had no effect on wine color in three of four years, has already been mentioned. From their work together in California, Freeman and Kliewer concluded, “Contrary to popular belief, increasing the pruning level, hence yield, generally increased these quality parameters.” In the late 1980s, a study of Cabernet Sauvignon in Napa Valley by Kliewer at UC Davis found that taste panels could successfully distinguish between 6.6 and 11.0 tons per acre wines in only one of two years, and could not distinguish between wines made from vines with smaller differences in yield.

In addition to several studies in the 1980s already mentioned, many studies over the next two decades reported similar observations. In Pinot Noir, large differences in yield (twofold) caused no differences in fruit color or pH, and cluster thinning to reduce yield about 35% had no effect on skin tannin or color in a study of Merlot in three consecutive years. Other Australian studies in hot regions reported a nil to weak relationship between yield and berry color in Shiraz and Cabernet Sauvignon vineyards. In yet another comprehensive study of Shiraz wines produced in South Australia, John Gray and coauthors were unable to find a relationship between yield and their “wine value index” across a large sample of growers in several regions.

Leading grapevine scholar Markus Keller and colleagues conducted a series of studies in Washington state that involved changing yield in various ways. In one five-year study, after reducing clusters by about 25% in Cabernet Sauvignon, Riesling and Chenin Blanc, Keller and colleagues found that “cluster thinning and its timing had little or no influence on shoot growth, leaf area, pruning weight, berry number, berry weight and fruit composition (soluble solids, titratable acidity, pH, color) in both the current and subsequent seasons.” In another five-year study, reducing clusters by about 40% had no effect on fruit composition (including color) in Cabernet Sauvignon. In 2010, Keller and colleagues conducted novel experiments in which the temperature of buds was increased in order to enhance their fruitfulness and found that “although yield per shoot varied threefold among treatments, differences in fruit composition were minor.” Yet the world of winegrowing has not assimilated these many experimental contradictions of HYLQ.

There are a few authors who have reported positive changes in fruit or wines (such as increased fruity aroma) with increasing yield, regardless of harvest criteria. However, for fruit harvested at similar Brix, the consequences of yield reduction by pruning or cluster thinning for grapes and wine, when present, are often similar to the slightly increased astringency in Cabernet Sauvignon wines reported by Chapman, Matthews, and Guinard (2004). That is, changes in sensory attributes that are sometimes statistically significant but also small. In that 2004 study, wine astringency ratings were closely correlated with the tannin concentrations, but varied only from about five to not quite seven on a scale of one to 10, while the corresponding differences in yield varied more than threefold.

WINE QUALITY SCORES -- Sauvignon blanc
Expert wine panels assigned scores to Sauvignon Blanc grown in the Golan Heights, Israel, and harvested in 1992, 1993 and 1994. (Naor et al., 2002.)

Yield and ripening
To be sure, there are studies showing negative relationships of color or other quality attributes to yield, as predicted by HYLQ. For example, Sauvignon Blanc wine quality scores were highest at the lowest yield (and lowest at the highest yield) in a study that included some (but not all) fruit harvested at the same Brix (see “Wine Quality Scores – Sauvugnon Blanc”). The maturity (Brix) of the fruit at harvest is important for understanding the role of yield in wine grape quality, because most viticulture experiments harvest all fruit on the same day, but the most common fruit response to yield reduction by pruning and cluster thinning is advanced fruit ripening. Thus, fruit reach ripeness sufficient for winemaking earlier when lower crop loads are carried, and they are riper at a given harvest date. This creates a problem for interpreting the results of those studies, unless the only quality issue is the advancement of ripening.

In experiments with Shiraz in Australia’s Barossa Valley, researchers Tony Wolfe, Peter Dry and colleagues found a consistently negative relationship between fruit anthocyanins and crop yield in three different seasons (color decreased as yield increased). In this study, however, sugar accumulation lagged in the higher yielding treatments, making it difficult to know how much of the differences in color were due to differences in fruit maturity. When the fruit of vines with significantly different yields are harvested on the same date, wines made from unripe (higher yielding) grapes are often found lacking. For example, in one early study when crop was thinned, the intensity of “good wine aroma” was dramatically increased, but the higher cropped vines had less mature fruit at harvest. This is consistent with HYLQ, but the aromas probably reflected riper fruit at lower crop loads rather than an effect of yield per se.

In most cases, when yields varied but fruit were harvested at the same Brix (rather than on the same date), studies have found little or no significant differences or loss of quality attributes in fruit or wine (as shown in the table “Studies Report Little or No Effect in Yield, Fruit or Wine Attributes”). However, my analysis is based on what could be gleaned from studies conducted for purposes other than testing HYLQ, and it is notoriously difficult to harvest fruit at precisely the desired Brix. This issue should be addressed experimentally, and that work should include harvesting fruit based on an experienced winemaker’s sense of ripeness, in addition to harvesting on the basis of Brix.

Furthermore, for growers who do not have the luxury of waiting until later in the season, getting to a higher sugar concentration is tantamount to quality. For these growers, a “vintage year” was a year when the sugar concentration reached an adequate value for winemaking at all, and chaptalization was not required. In the milder climates of southern Europe, California and other warm regions, every year is a vintage year in this regard, and growers in these accommodating climates can simply wait a little longer and harvest a higher crop load at the same ripeness. Indeed, longer “hang times” have become associated with desirable wine flavors for some (but not all) wine experts.

To many both inside and outside the realm of viticulture research, it may come as a surprise that for more than half a century, pruning and fruit-thinning studies have reported small or undetectable effects on fruit and wine when fruit is harvested at similar Brix. Despite this contrary research, many continue to consider higher yield, however it is attained, to be a problem that dilutes the fruit and reduces wine quality (a topic to be taken up later in this book). While it is widely recognized that increased yield has one major effect in delaying ripening, the p ossibility that higher yield could be a means of achieving better flavors derived from longer hang times (a longer ripening period) seems to go unrecognized.

Mark A. Matthews is a professor and plant physiologist at the Robert Mondavi Institute for Wine and Food Science at the University of California, Davis. In his recently released book, Terroir and Other Myths of Winegrowing, Matthews applies a scientist’s skepticism, as well as facts from the historical record, plant physiology, scientific research and a bit of economics to debunk myths that may be preventing advancements in viticulture and enology.

Currently no comments posted for this article.