Effect of Preveraison Water Deficits on the Yield Components of 15 Winegrape Cultivars

Yield and berry FW

Vine yield responded similarly to the irrigation treatments across cultivars and years from 2012 to 2015, with consistently large reductions in yield due to preveraison water deficits that reduced berry size together with clusters per vine. In contrast, postveraison water deficits during that time had an inconsistent effect on final yield across cultivars and years. There was a strong seasonal yield response as well, with yields generally declining from high to low from 2012 to 2013, to 2015.

When comparing preveraison and postveraison water deficit effects on yield, the reduction of yield due to preveraison water deficits is the most consistent observation in the literature despite relatively variable preveraison weather conditions (Matthews and Anderson 1989, Williams et al. 1994, Romero et al. 2010, Santesteban et al. 2011, Junquera et al. 2012, Intrigliolo et al. 2012, Uriarte et al. 2015). In contrast, previously reported effects of postveraison water deficits on yield are inconsistent, with yields either lower (Junquera et al. 2012), the same (Matthews and Anderson 1989), or even higher (Intrigliolo et al. 2012), compared to well-watered controls. Furthermore, there were substantial interyear inconsistencies of yield responses to postveraison water deficits within the aforementioned studies, despite relatively stable interyear weather conditions late in the season. Similar interyear inconsistencies among LD yields were found in this study.

Berry FW was more responsive to the irrigation treatments across all years and cultivars, and it largely drove the differences in vine yield. In addition, berry FW response to the irrigation treatments was more consistent and less variable than that of berries per vine (the product of clusters per vine and berries per cluster) across years and cultivars. Berry FW is highly sensitive to vine water deficits—particularly between anthesis and veraison—which limit berry growth through reduced cell enlargement in the berry mesocarp (Matthews and Anderson 1989, Ojeda et al. 2001), whereas postveraison water deficits have a more limited effect on berry growth, due to the shift in water source at veraison from xylem to phloem (Greenspan et al. 1994). Accordingly, berries grown on vines subjected to postveraison water deficits often have a similar FW to that of the controls (Matthews and Anderson 1989, Castellarin et al. 2007).

In this study, preveraison water deficits of Ψ_l < −1.3 MPa in ED vines consistently and significantly reduced berry FW relative to the SD and LD irrigation treatments for many cultivars, which ultimately reduced their final yield. By comparison, berry FW (and ultimately yield) was less affected by the LD treatment, with few significant differences among cultivars between SD and LD vines. It should be noted that without well-watered vines postveraison, interpretation is made more difficult considering that this may have somewhat limited the maximum potential yield and berry FW. However, as mentioned above, several other studies that included a well-watered control have shown limited to no effects of postveraison water deficits. The results of this study indicate that the potential to maximize yield across all cultivars is greatest before veraison by avoiding water deficits of Ψ_l < −1.3 MPa.

Sensitivity of clusters per vine versus berries per cluster

In addition to reducing berry FW, preveraison water deficits also reduced yield through a reduction in clusters per vine that drove the overall reduction in total berries per vine. In contrast, berries per cluster did not respond strongly to the irrigation treatments, despite the treatments’ strong impact on vine water status both pre- and postveraison. Finally, the lack of any significant differences among cultivars in response sensitivity to pre- or postveraison water deficits for either yield component provides evidence that these traits are under tight genetic control and that this control is highly conserved across V. vinifera L. cultivars.

Temporally, inflorescence (cluster) primordia must be induced in the bud prior to any differentiation of floral primordia, and this occurs mid-summer between anthesis and veraison (Williams 2000). Accordingly, clusters per vine response to water deficits was only observed preveraison, during primordia development, with reductions in clusters per vine due to preveraison water deficits but little effect due to postveraison water deficits. The significant reductions in clusters per vine were observed in ED vines (relative to SD vines) whose preveraison Ψ_l averaged -1.5 to -1.6 MPa. This suggests that preveraison water deficits of Ψ_l ≤ -1.5 MPa are significant enough to disrupt development of cluster primordia in the bud. On the other hand, the lack of differences in clusters per vine between LD and SD vines suggests that inhibition of cluster primordia development is likely unaffected at preveraison water deficits down to Ψ_l = -1.3 MPa and that severe postveraison water deficits (Ψ_l << -1.5 MPa) do not affect primordia development at all, presumably because this process is completed by veraison. Therefore, an average preveraison vine water status of Ψ_l = -1.3 MPa may be considered a preliminary critical value below which bud fruitfulness is negatively affected.

The inhibition of cluster primordia development under the water deficits in this study is more likely due to changes in shoot carbon partitioning as opposed to simple shoot growth inhibition. In general, plant cell expansive growth is more sensitive to water deficits compared to net photosynthesis (Hsiao 1973). However, while grapevine shoot elongation is typically inhibited at Ψ_l = −1.0 MPa (Schultz and Matthews 1988, Williams et al. 2010a), net photosynthesis is only reduced by 9% (Williams 2012b). However, at Ψ_l = −1.3 MPa, Williams (2012b) showed that net photosynthetic rate was reduced by 49%, and at Ψ_l = −1.5 MPa, it was reduced by 94%. Recently, a water deficit-induced inhibition of photosynthesis was reported to negatively affect floral differentiation (Dayer et al. 2013). Thus, the inhibition of cluster primordia development—as inferred from the reduction in clusters per vine—in ED vines relative to SD vines and the lack of differences in clusters per vine between LD and SD vines—can be at least partially explained by a change in the carbon economy of the developing shoot.

There is some disagreement in the literature about whether clusters per vine or berries per cluster is more sensitive to water deficit, although this may have been due to cultivar, rootstock, and/or site differences across studies. For example, Matthews and Anderson (1989) and Keller et al. (2016) showed that berries per cluster was more sensitive to water deficits compared to clusters per vine in Napa-grown Cabernet franc (grafted) and in eastern Washington-grown Cabernet Sauvignon (own-rooted), respectively. In contrast, Williams (2010) showed that clusters per vine is more sensitive to water deficits in Cabernet Sauvignon across multiple rootstocks in the Central Coast region of California (Paso Robles), suggesting that rootstock may have limited effect on sensitivity differences between the two yield components when that factor was controlled.

Other work has also shown that clusters per vine is more sensitive to water deficits compared to berries per cluster in Thompson Seedless (Williams et al. 2010b) and Merlot (Williams 2012a) grown in the southern San Joaquin Valley (SJV) of California. In addition, Keller et al. (2012) found fewer differences in berries per cluster compared to clusters per vine among three cultivars: Syrah, Chardonnay, and Merlot grown in eastern Washington. Those studies took place across varied regions of the western United States, potentially confounding direct comparisons, but their evidence supports the notion that clusters per vine is the yield component that is more sensitive to water deficits.

In this study, there were many more statistically significant differences among treatments in the response of clusters per vine compared to berries per cluster, and these responses were consistently observed across all 15 cultivars grown on the same rootstock over several years. Although rootstock has been shown to significantly influence yield formation in V. vinifera L. scion cultivars, the scion effect is typically much stronger (Keller et al. 2012); therefore, the cultivar comparisons are valid. There were also significant differences among cultivars for each yield component variable on an absolute basis, but when cultivars were compared on a relative basis, there were no differences in response sensitivity for either variable. This analysis may be somewhat confounded by the facts that (1) ED and LD treatments were not perfectly mirrored pre- and postveraison, and (2) the SD treatment designated as the control was not well watered. Future studies should aim to compare cultivars, taking these factors into account. Nevertheless, the results of this study coupled with evidence from previous research suggests it is unlikely that the sensitivity of yield components’ response to water deficits is cultivar dependent.

Interrelationships among flowers per cluster, percent berry set, and berries per cluster

To ensure that the lack of significant differences in berries per cluster among irrigation treatments was not due to differences in percent berry set, several representative cultivars were chosen in 2014 and 2015 for the determination of percent berry set. The selected cultivars were chosen based on a range of berries per cluster measured in 2013. The irrigation treatments did not have a significant effect on percent berry set either year. This was not surprising considering that all vines across treatments were irrigated similarly and had the same water status until after berry set (Levin et al. 2019). The irrigation treatments imposed the previous year may have affected starch reserves in the vines because of a water deficit-induced inhibition of photosynthesis that can inhibit floral differentiation (Dayer et al. 2013). However, it is unlikely that there would be a carryover effect of the irrigation treatments on set, considering that floral differentiation is highly controlled by environmental and physiological factors the current year (Skinner and Matthews 1989, Dokoozlian 2000). Based on the mean temperatures during the period of anthesis and berry set in both years (data not shown), it did not appear that set was negatively affected by environmental conditions in this study.

The interrelationships between flowers per cluster, percent berry set, and berries per cluster obtained in this study demonstrated that berry set was relatively high (~60%) at very low numbers of flowers per cluster (<150) and remained stable as flower number increased until ~250 flowers per cluster, at which point it began to decrease. Keller et al. (2010) reported a nearly linear relationship with berry set at ~55% at ~500 flowers per shoot across three years in Cabernet Sauvignon. In that study, there was an average of just under two clusters per shoot across all years and treatments, which would equate to comparable percent berry set values found in this study on a per cluster basis (i.e., ~250 flowers per cluster). Furthermore, Keller et al. (2012) also found significant differences among cultivars in percent berry set and flowers per cluster with no significant interactions between cultivar and rootstock. The authors reported higher percent berry set values with lower flower numbers and also an across cultivar percent berry set of ~50% at 250 flowers per cluster (Keller et al. 2012). Thus, the data in this study corroborate other findings regarding percent berry set in winegrape cultivars grown across different climates; namely, that flowers per cluster is strongly cultivar dependent, and that percent berry set levels much higher than 60% are unattainable in cultivated winegrapes.

Interestingly, it appears that across several cultivars with widely varying numbers of flowers per cluster, berries per cluster reached a maximum at ~150 and would not increase significantly with increasing flower number either year. Although grapevines have been previously reported to exhibit compensatory responses with respect to reproductive development (Keller et al. 2012), there were no significant relationships found in this study between berries per cluster and clusters per vine (data not shown). Data from multiple studies across different regions, years, cultivars, rootstocks, and experimental treatments show that berries per cluster is almost always ~150 or below and rarely above 200 (Skinner and Matthews 1989, Matthews and Anderson 1989, Keller et al. 2008, 2010, Santesteban et al. 2011, Intrigliolo et al. 2012, Junquera et al. 2012, Keller et al. 2012, Dayer et al. 2013, Romero et al. 2013, Guilpart et al. 2014, Mendez-Costabel et al. 2014, Uriarte et al. 2015, Nelson et al. 2016).

Consequently, ~150 berries per cluster may represent somewhat of a physiological limit to the number of berries per cluster in V. vinifera L. cultivars under nonlimiting environmental conditions. From a mechanistic perspective, this would likely be a function of photoassimilate partitioning among flower sinks, whereby at low numbers of flowers per cluster, percent berry set is sink limited, while at high numbers of flowers per cluster percent berry set is source limited. From a management perspective, cane pruning can be employed as a management tool to increase flowers per cluster in cultivars that consistently set fewer than 150 berries per cluster, as more fruitful mid-cane buds are retained with this technique (Williams 2000, Sanchez and Dokoozlian 2005).

Carryover effects and year-to-year variation

In 2016, the treatments in the ED and LD plots were altered such that vines received 100% of estimated ET_c to evaluate any potential carryover effects on reproductive development after several years of either early or late-season water deficits. Congruently, the SD plots were unchanged to account for any year effects. Yields were significantly increased across all treatments from 2015 to 2016 (including SD), mainly due to larger berries, and to a smaller degree, due to more berries per vine. Although berry sizes were similar across treatments in 2016, the lower yields in ED vines in 2015 were not made up for in 2016 when water was returned.

The general year-to-year reduction in berries per vine over the course of the study was primarily due to the reduction in clusters per vine, and to a much lesser extent, in berries per cluster. The largest year-to-year reduction in clusters per vine (and concomitant reduction in yield) across all cultivars was from 2013 to 2014, with little change from 2014 to 2015, and was increased slightly in 2016. While it is difficult to draw definitive conclusions about the cause of year-to-year variation in yield components (outside of the effects of the experimental treatments), it has been shown that dry conditions during the winter can reduce grapevine yield in the following season by primarily reducing the number of clusters per vine (Mendez-Costabel et al. 2014). That study was conducted in the SJV of California over two average rainfall winters in which the soil surface was covered to prevent absorption of water into the soil profile.

While most assume a full soil water profile at the beginning of the growing season, the measured values of SWC prior to the beginning of the 2015 growing season were below the soil type’s typical field capacity rating of 18.2% v/v (websoilsurvey.sc.egov.usda.gov). In addition, the values measured were also below those previously observed for well-watered grapevines grown on this soil type (Williams et al. 2010a). Thus, the low amount of dormant season rainfall in the winters of 2013 to 2014 and 2014 to 2015 did not fill the soil to field capacity, and vines were potentially under slight water deficit during dormancy. Reductions in clusters per vine and yield of Zinfandel grown on two rootstocks in the SJV were also reported by Nelson et al. (2016) from 2013 to 2014. The maintenance of cluster primordia during the dormant season has been shown to be sensitive to environmental conditions (Skinner and Matthews 1989). Therefore, it is possible that dry conditions during dormancy led to water deficits that caused an abortion of cluster primordia, reducing clusters per vine from 2013 to 2014 in this study as well.

Finally, it should be noted that SD vines also had significantly larger berries in 2016 compared to 2015 across all cultivars. Midday Ψ_l was also greater in SD vines throughout the 2016 growing season compared to the 2015 growing season. While it is commonly understood that berry size is primarily a function of the current season’s growing conditions (May 2000), berry size may be largely determined well before the season begins, or at least before fruit set. Because the berry is the result of a fertilized ovary that itself developed in the flower, environmental conditions during the final stages of floral differentiation—such as temperature (Keller et al. 2010) or precipitation (Mendez-Costabel et al. 2014)—may significantly affect ovary size and thus final berry size. When Mendez-Costabel et al. (2014) excluded winter precipitation and kept soil dry during dormancy, berries were significantly smaller than controls at two weeks postanthesis. Accordingly, the significant rainfall during dormancy from 2015 to 2016 may have increased the water status of the dormant vines, influenced respiration, and ultimately encouraged more growth and development of primordial floral structures that ultimately led to larger berries in 2016.