Optimizing Irrigation for Mechanized Concord Juice Grape Production

Background and goals Economic considerations and water shortages associated with climate change are driving the conversion of many hand-pruned and furrow-or sprinkler-irrigated juice grape vineyards in arid eastern Washington to mechanical pruning and drip irrigation. However, juice grape producers have traditionally avoided plant water deficits, fearing a decline in vineyard productivity.


Introduction
Depending on the year and market conditions, between one-third and one-half of the juice grape production in the United States is concentrated in eastern Washington, much of it in the Yakima Valley.This region is characterized by an arid desert climate with warm but relatively short (<180 days) growing seasons and cold winters (Badr et al. 2018, Beck et al. 2018).Because the annual precipitation is only ~200 mm, adequate irrigation water supply is essential for vineyard establishment and sustained vine productivity.Consequently, the changes in precipitation patterns associated with global climate change are of some concern for the region, where irrigation relies on snow melt in the Cascade mountains.Like elsewhere in the western U.S., the water stored in those mountains' snowpack in early spring correlates strongly with water supply during the summer (Elsner et al. 2010).Because of a decreasing trend in snowpack and earlier snowmelt, peak river flows are shifting from May/June to February/March, and irrigation water availability in summer is declining (Elsner et al. 2010, Wagner et al. 2021).The decreased water supply, coupled with the ongoing rise in temperature and evapotranspiration, is a challenge for juice grape production because most growers in the region strive to apply ample irrigation water to avoid plant water stress.Although many vineyards are being converted from furrow or overhead sprinkler irrigation to more waterconserving drip irrigation, it is uncertain whether productivity can be maintained under drier conditions.The large canopy of well-watered juice grapes uses substantially more water than does the smaller canopy of deficit-irrigated wine grapes (Dragoni et al. 2006, Tarara andFerguson 2006).However, contrary to the abundance of research into deficit irrigation of wine and table grapes (reviewed by Costa et al. 2016, Permanhani et al. 2016, Scholasch and Rienth 2019, Mirás-Avalos and Araujo 2021), little research has addressed the question of how limited water supply impacts fruit yield and composition of juice grapes (Morris et al. 1983, Reynolds et al. 2005, Stout et al. 2017).A recent comprehensive review emphasized both the difficulty of generalizing optimal irrigation strategies across cultivars and regions, and the need for long-term studies (Mirás-Avalos and Araujo 2021).
The importance of the irrigation question is accentuated by the shift toward light mechanical pruning of much of the juice grape acreage in the region, driven by labor scarcity and the desire to cut production costs and maximize vineyard productivity.Because lower pruning severity increases canopy size, and hence, transpiration (Gómez del Campo et al. 1999, Lakso andDunst 1999), lightly pruned vines may be more vulnerable to water stress (Keller et al. 2004(Keller et al. , 2015)).Heavy crop loads typical of such vines can further exacerbate vine vulnerability to water stress (Poni et al. 1994).In addition to its impact on shoot growth and leaf photosynthesis, water stress during the preveraison berry development phase restricts berry size, and postveraison water stress can lead to undesirable berry weight loss or delayed fruit maturity (Hardie and Considine 1976, Keller et al. 2006, Zhang and Keller 2015).Unlike wine grapegrowers, juice grapegrowers therefore have been reluctant to adopt deficit irrigation strategies.
Because mechanical pruning could exacerbate problems associated with water shortages, this labor-saving practice calls for careful irrigation management.The present study was conducted to test a range of irrigation treatments in a drip-irrigated, mechanically pruned juice grape vineyard in the Yakima Valley.Our hypothesis was that the large canopy of machine-pruned vines requires ample water supply to maximize and stabilize crop yield in the long term.The field trial described here evaluated both the amount and timing of water supply and their influence on canopy development, plant water status, yield formation, and juice composition.Our goal was to identify one or several optimal irrigation strategies that may conserve valuable irrigation water while maximizing the production of high-quality juice grapes over multiple years.

Vineyard site and management
Own-rooted Concord grapevines (an interspecific hybrid cultivar with Vitis labrusca L. and Vitis vinifera L. ancestry) were planted in 2003 in a 3.2 ha vineyard block at Washington State University's Irrigated Agriculture Research and Extension Center (46°29´N; 119°74´W, 360 m asl).The soil type is Warden silt loam at pH 7.2 with <1% organic matter, volumetric water content (θ v ) at field capacity (FC) of 22.7% (v/v), and θ v at permanent wilting point (PWP) of 7.1% (https://websoilsurvey.sc.egov.usda.gov).A caliche layer over basalt at depths varying from 50 to 120 cm limits rooting depth.The vines are planted at 2.74 m between rows and 1.83 m within rows oriented north-south down a <2% south-facing slope.They are trained to a single cordon wire at 1.83 m with no foliage wires; the shoots hang under the weight of the developing fruit.The vineyard has been mechanically pruned by a cooperating grower since 2007 as described (Keller and Mills 2021) and the canopy is not managed during the growing season.A permanent midrow cover of resident vegetation is maintained between rows; it is mowed as needed and generally goes dormant in the summer.A 1.2 m herbicide strip is maintained in the rows during the growing season, but other pesticides are not applied in this vineyard.Nitrogen fertilizer, in the form of UAN-32, is applied by fertigation through the drip irrigation system at a rate of 66 kg N/ha split equally between the six-leaf stage, bloom, and fruit set.Drip irrigation is applied weekly as needed using 18 mm driplines with 3.4 L/hr emitters spaced at 0.91 m.To compensate for the low winter precipitation in this region (<80 mm on average; Table 1), the vine rows are usually irrigated to near FC around budbreak.Limited soil

Treatments and experimental design
An irrigation trial was conducted from 2011 through 2016.Irrigation treatments were designed to replace various fractions of water lost to crop evapotranspiration (ET c ) between fruit set and harvest.Instead of applying a crop coefficient (K c ), the amount of water to be applied under full irrigation (100% ET c ), which served as the control, was estimated based on maintaining θ v in the soil's top 60 cm constant near 18% (θ e = 0.7).Differences in water supply for the remaining treatments were achieved by retrofitting the irrigation system to accommodate pressure-compensating drip emitters with different flow rates; emitters were changed at key phenological stages as required by the different treatments.Those treatments were: mild early deficit (75% ET c from fruit set to veraison and 100% ET c thereafter); moderate early deficit (50% ET c from fruit set to veraison and 100% ET c thereafter); mild late deficit (100% ET c from fruit set to veraison and 75% ET c thereafter); moderate late deficit (100% ET c from fruit set to veraison and 50% ET c thereafter); mild full-season deficit (75% ET c from fruit set to harvest); and moderate full-season deficit (50% ET c from fruit set to harvest).In response to industry requests, the mild early deficit treatment was replaced in 2014 with abundant early irrigation at 150% ET c through veraison and 100% ET c thereafter.The other two mild (late and full-season) deficit treatments were not applied in 2016 because vine responses did not differ from the control in the first five years (see Results section).Each treatment was applied to 12 consecutive vines in three adjacent rows and replicated in four randomized blocks that were 48 vines long and six rows wide.Flow meters were installed in each block in 2013 to estimate irrigation water supply by treatment.

Weather data and plant measurements
Daily weather data were obtained from the WSU-Roza AgWeatherNet station (http://weather.wsu.edu),located at the same elevation and ~500 m to the east of the vineyard.Growing degree days (GDD) from 1 April through 31 Oct were calculated from daily maximum and minimum temperatures, using a base temperature of 10°C.Vine phenology was monitored regularly, and dates of 50% budbreak, bloom, and veraison, as well as the harvest date, were recorded.
Four consecutive data vines were designated in the middle row of each treatment replicate.Beginning at fruit set and on days that θ v was measured, we also measured midday stem water potential (Ψ s ) to determine irrigation effects on vine water status.Recently mature, sun-exposed leaves were enclosed in aluminum-coated plastic bags for ≥2 hr and measured in a pressure chamber (model 615, PMS Instrument Company) between 1300 and 1500 hr local time.Canopy dimensions (average height from top to bottom of pendant shoots and width across the fruit zone) were measured at veraison and before harvest to estimate the external canopy surface area and canopy volume, assuming a rectangular canopy cross section (Keller and Mills 2021).Within 30 min of solar noon on the same days, light penetration into the fruit zone was estimated using a ceptometer (AccuPAR LP-80, Decagon Devices) as described (Keller and Mills 2021).In 2013, 2014, and 2016, the trunk diameter was measured at the height of the dripline 45 cm above the vineyard floor as a proxy for vine size, since pruning weight is not a suitable indicator of vine vigor or size for mechanically-pruned grapevines.Trunk diameter (and hence, crosssectional area) correlates strongly with both aboveground and whole-vine structural biomass (Miranda et al. 2017).
All data vines were harvested manually on the same day once the vineyard exceeded an overall total soluble solids (TSS) target of 16 Brix, except in the cool 2011 season (Table 1), when this target was not reached and grapes were harvested at 15 Brix.The vineyard was then machine-harvested by a cooperating grower.Yield and its components (clusters per vine, cluster weight, berries per cluster, berry weight) were determined at harvest.A random 100-berry sample was collected at harvest from each treatment replicate to measure fruit composition.Juice TSS, titratable acidity (TA), pH, and red color intensity were determined as described (Keller et al. 2004).
Because of an unusually heavy crop in 2014, we measured bud cold hardiness and cane phloem and xylem hardiness during the subsequent winter.Measurements were conducted by differential thermal analysis as described (Mills et al. 2006) in mid-October, early and late November, early January, early and late February, and late March on four replicates of five buds or two cane internodes in the control, mild full-season deficit, and moderate full-season deficit treatments.

Data analysis
Data were analyzed using Statistica version 14 (TIBCO Software).The pH values were converted to H + concentrations for data analysis and means were converted back to pH for presentation.Because some irrigation treatments changed during the experiment, data were first analyzed by analysis of variance (ANOVA) for the control and the three moderate deficit treatments over all six years.Then, all pertinent treatments were analyzed across the years in which they were applied.In each case, the year effect was tested using a repeated measures design.Though year × treatment interactions were rarely significant (exceptions: θ v and Ψ s ), treatments were also analyzed by one-way ANOVA for each year to account for the changing treatments, resulting in an unbalanced design over time.Duncan's test was used for post-hoc means comparisons when treatment effects were significant within years.Associations between key response variables were tested using Pearson product moment correlation analysis.

Phenology and weather
The timing of phenological stages of our Concord vines varied widely among years, depending on seasonal weather conditions (Table 1).The date of budbreak ranged from 1 April (2015) to 30 April ( 2011), bloom occurred between 15 May (2016) and 20 June (2011), with fruit set typically one to two weeks later.Veraison occurred between 13 Aug (2015) and 14 Sept (2011), and the date of harvest varied from 16 Sept (2015) to 28 Oct (2011).The Yakima Valley, where the study vineyard is located, is considered one of eastern Washington's cooler grapegrowing regions (https://wine.wsu.edu/extension/weather).Compared with the long-term average of 1409 GDD and mean growing season temperature of 16.5°C, the 2011 growing season was unusually cool and 2015 was unusually warm (Table 1).The 2012 and 2013 seasons were close to the average, while 2014 and 2016 were warmer than average.The seasonal GDD accumulation is shown in Figure 1, and daily maximum and minimum temperatures and rainfall are shown in Supplemental Figure 1.Spring frosts were fairly common through mid-April and the latest frost (-1.7°C) occurred on 1 May 2013, 16 days after budbreak.Annual precipitation varied from 125 mm in 2011 to 241 mm in 2012.Rainfall during the growing season ranged from 50 mm in 2015 to 128 mm in 2012, which was far below the seasonal grass reference evapotranspiration (ET o ), which varied from 924 mm in 2011 to 1057 mm in 2015 (Table 1).The daily ET o (range: 0 to 10.5 mm) correlated strongly with daily solar radiation (r = 0.83, p < 0.001, n = 1284), maximum temperature (r = 0.80, p < 0.001), and average relative humidity (r = -0.79,p < 0.001).Multiple regression analysis showed that these three variables accounted for 88% of the variation in ET o (multiple R = 0.94, p < 0.001).

Irrigation water supply
Flow meters installed from 2013 through 2016 showed that the control treatment (irrigation at 100% ET c ) received between 689 mm and 826 mm irrigation water per year (Figure 2A).Flow meter readings, as a proxy for ET c of the 100% ET c vines, combined with ET o accumulated between weekly irrigation events, permitted the estimation of a K c (ET c / ET o ) for the fruit set to harvest period.The K c was rather consistent at 1.05 ± 0.09 (mean ± SE) during this period and over the four years.On average, across years and irrigation treatments, 10% (58 mm) of the irrigation water was applied from budbreak to fruit set, 53% (345 mm) from fruit set to veraison, 19% (117 mm) from veraison to harvest, and 18% (112 mm) after harvest (Figure 2B, Supplemental Table 1).The av-  erage water supply between fruit set and veraison ranged from 201 mm at 50% ET c irrigation to 391 mm at 100% ET c and 632 mm at 150% ET c .Similarly, water supply during fruit ripening varied from 74 mm at 50% ET c to 147 mm at 100% ET c .Because we attempted to even out differences in θ v at budbreak and after harvest by irrigating the deficit treatments more heavily, postharvest water supply ranged from 157 mm in the 50% ET c treatment down to 48 mm in the 150% ET c treatment.On an annual basis, reducing irrigation to 50% ET c from fruit set through harvest decreased water supply by 31% below the 100% ET c control; restricting the 50% ET c treatment to the preveraison period resulted in 30% water savings.The next most water-conserving strategy was 75% ET c from fruit set to harvest (20% savings); this treatment required less irrigation water after harvest than the 50% ET c treatments.Irrigating at 150% ET c before veraison, however, increased overall water supply by 20% over the control.

Soil and plant water status
Seasonal trends of θ v and Ψ s are shown in Figure 3, and Ψ s data, averaged for the preveraison and postveraison periods, are summarized in Supplemental Table 2.In most years, θ v declined through fruit set or somewhat later, changed depending on the irrigation treatment, then increased to prebloom levels following postharvest irrigation.Irrigating at 75% ET c , and especially, at 50% ET c , generally led to soil drying, irrigating at 100% ET c kept θ v constant, while irrigating at 150% ET c steadily increased θ v (Figure 3).Thus, the preveraison θ v was greatest in the 150% ET c treatment (applied from 2014) and least in the 50% ET c treatment.Moreover, the vines irrigated at 150% ET c the previous year started out with the greatest θ v the following spring, and although θ v decreased somewhat through fruit set, it then increased and remained near or above FC through harvest.The other treatments were irrigated after harvest and near budbreak to increase θ v to within 3 to 5% of FC.
Each year, the average preveraison θ v (or θ e ) was a nearly linear function of the irrigation water applied during that period (0.72 < r < 0.95, p < 0.001; see also Figure 4A).The correlation between postveraison θ v and water applied during ripening was not as strong (0.31 < r < 0.77, p < 0.11) due to carryover effects from the preveraison irrigation treatments.The average preveraison Ψ s correlated strongly with irrigation water supply during that period in all years (0.74 < r < 0.87, p < 0.001), but reached a plateau (Ψ s ≈ -0.4 MPa) at ~400 mm applied water (Figure 4B).As with θ v , correlations were weaker for postveraison Ψ s and water supply (0.31 < r < 0.79, p < 0.11).Moreover, θ v (and hence θ e ) and Ψ s correlated positively.Figure 4C shows the Ψ s versus θ e relationship for the 0 to 60 cm depth, but the correlation coefficients were nearly identical for the 0 to 30 cm and 0 to 90 cm depths, indicating that roots were taking up water over the entire measured soil profile.Except in 2016, when more irrigation water was applied than in the other seasons, Ψ s generally declined to -1.2 to -1.4 MPa when θ v reached ~12% and θ e ≤ 0.3 (Figure 3).While deficit irrigation, especially at 50% ET c , clearly resulted in lower plant water status, there was no further rise in Ψ s at θ v ≥ 16% and θ e ≥ 0.6 (Figure 4C).Even irrigating at 150% ET c , which raised θ v above FC (i.e., θ e > 1), did not increase Ψ s significantly above -0.4MPa.

Canopy size and light exposure
The canopy dimensions measured at veraison were generally similar to those measured preharvest and changes in postveraison irrigation had negligible effects on canopy size; thus, only the harvest data are presented here (Figure 5, Supplemental Table 3).Before veraison of 2014, the canopy was accidentally hedged ~70 cm above the vineyard floor, which decreased its surface area and volume below that of the other years.In general, the preveraison moderate deficit irrigation treatments led to slightly smaller canopies (i.e., lower surface area and volume) than the other treatments (Figure 5A, 5B).The canopy size rarely differed among the remaining treatments and the 150% ET c treatment did not increase canopy size above that of the control.Although statistically significant, the difference between the lowest and highest water supply was only 10% in terms of canopy surface area and 25% in terms of canopy volume.The irrigation-related differences in canopy size were accompanied by an 11% difference in trunk diameter (thinner at 50% ET c irrigation), which increased by 8% (3.7 cm to 4 cm) from 2013 through 2016 (p < 0.001).The trunk diameter in 2016 correlated positively with the average preveraison Ψ s (r = 0.64, p = 0.003).
The proportion of sunlight that reached the fruit zone decreased from an average of 8% in 2011 to 4% in 2012 and then to 2% thereafter.Moderate deficit irrigation before veraison generally more than doubled the amount of light in the fruit zone over the control (Figure 5C).Changing the irrigation rate at veraison did not alter light penetration.In each year but 2011, fruit zone light correlated inversely with the average preveraison Ψ s (-0.75 < r < -0.40, p < 0.05).
The cold-hardiness of buds and cane phloem and xylem remained unaffected by irrigation treatments between mid-October 2014 and late March 2015 (p > 0.3).

Yield and its components
Crop yields varied from 17 t/ha in 2012 to over 55 t/ha in 2014 (Table 2, Supplemental Table 3).Alternate bearing did not account for yield fluctuations: no negative correlations were observed between prior-year yield and currentyear yield of the 112 data vines (Supplemental Figure 2).The low crop in 2012 was a result of 38% fewer clusters per vine, 37% fewer berries per cluster, and 4% smaller berries than in 2011 (Table 2, Figure 6A, 6B).Yields recovered in 2013, mostly because of high cluster numbers per vine, but cluster weights remained low following the late frost that year.The heavy crop in 2014 was mostly due to larger than usual clusters (i.e., more berries per cluster).The 2015 crop was significantly lower than in other years except 2012, mostly as a result of lower cluster numbers.The 2016 crop was the second-highest among the six years, resulting from a combination of above-average cluster numbers, cluster weights, and berry weights (Table 2).Within years, between 22% and 59% of the yield variation could be attributed to the variation in cluster number per vine (0.47 < r < 0.77, p < 0.001), while the cluster weight contributed only between 1% and 14% to that variation (0.08 < r < 0.37, p < 0.001).The main driver of variation in cluster weight was the number of berries per cluster, while berry weight was comparatively unimportant.The average berry number per cluster ranged from 17 ± 1 in 2013 to 28 ± 1 in 2011 (p < 0.001).
Compared with yield variability due to other sources, irrigation effects on yield were small.On average, irrigating at 50% ET c before veraison decreased the harvest yield by 14% below that of the 100% ET c control (Figure 6A).However, preveraison irrigation at 75% or 150% ET c did not alter yield relative to the control, and differences in postveraison irrigation had no effect on yield.Irrigation influenced crop yield mainly by altering berry weight (Figure 6C).Moderate deficit irrigation before veraison led to a 6% decrease in berry weight compared with the control (p = 0.017), but the other treatments rarely impacted any yield components.The magnitude of the irrigation effect seemed to increase over time, with virtually no significant effects in the first two years, when θ v was not as well differentiated among treatments.In 2016, moreover, preveraison moderate deficit irrigation also reduced clusters per vine by 18% (p = 0.002).Except in 2011, the average berry weight correlated positively with the average preveraison Ψ s , and the correlation gradually increased from 2012 (r = 0.46, p = 0.016) through 2016 (r = 0.76, p < 0.001).A similar correlation was also found between berry weight and average preveraison θ v (0.41 < r < 0.82, p < 0.03).Correlations between berry weight and average postveraison Ψ s and θ v were also significant, but only if the preveraison 50% ET c treatments were included in the analysis, indicating that any apparent effects of postveraison water status resulted from differences established before veraison.
In four of the six years, the canopy volume correlated positively with yield per vine (0.40 < r < 0.52, p < 0.05).Fruit yield also increased in a curvilinear fashion as the annual irrigation water supply increased, but the relationship was different each year (Figure 7A), again demonstrating the dominant influence of seasonal weather over irrigation.Adding the sparse rainfall to the water supply did not change this relationship in any year.When the irrigation water supply was broken down by phenological phase, only the correlations for the period from fruit set to veraison were significant (Figure 7B).Therefore, the irrigation water use efficiency (yield per unit water applied; mean 6.9 ± 0.2 t/ML) was entirely dependent on preveraison irrigation, decreasing successively from 50% ET c to 150% ET c (p < 0.001).The irrigation water footprint (water applied per unit yield; 158 ± 5 m 3 /t) and the total water footprint (rainfall plus irrigation water per unit yield; 199 ± 6 m 3 /t) showed the opposite trend of increasing from 50% ET c to 150% ET c (p < 0.001).Data by year and treatment are shown in Supplemental Table 4.

Fruit composition
Like yield, fruit composition was dominated by the influence of the growing season, rather than the irrigation treatment.The least-mature fruit was harvested in the cool 2011 season and the most-mature fruit was harvested in the warm 2015 season (Table 2).The very high yield in 2014 did not prevent the fruit from reaching adequate maturity, albeit with low color density, by the time of harvest in mid-October.The irrigation treatments did not impact juice TSS, TA, pH, or red color in three of the six years (Supplemental Table 5).In 2013 and 2016, the only effect was a small decrease in TA with moderate preveraison deficit irrigation, compared to the other treatments.Nevertheless, in each year except 2015, there was a positive correlation between TA and the average preveraison Ψ s (0.38 < r < 0.85, p < 0.05).In 2015, the low yield in the 50/100% ET c treatment was associated with higher TSS and color; irrigating at 150% ET c before veraison reduced juice color.Differences in postveraison irrigation did not alter fruit composition in any year.
Greater TSS correlated with more intense juice color, both within years (0.58 < r < 0.82, p ≤ 0.001) and between years (r = 0.69, p < 0.001, range 13.6 to 21.1 Brix).Berry size, however, did not correlate with color density in any year (p > 0.1).Despite the high variation in crop yield within and between years, yield was generally not an important driver of fruit composition.The strongest correlation between yield (range 17.5 to 28.5 kg/vine) and TSS was found in 2011 (r = -0.55,p = 0.003), when vines cropped at >45 t/ha had lower TSS and color density than vines cropped at <45 t/ha.Highyielding vines that were irrigated at 50% ET c were just as likely to have fruit with low TSS and color as were vines that were irrigated at 100% ET c .Despite similar or even greater yields in 2014 (23.4 to 32.0 kg/vine) and 2016 (20.3 to 29.5 kg/vine), yield did not correlate with TSS in those warmer years (p > 0.2).Nevertheless, greater yields were associated with higher TA in the last three years (0.43 < r < 0.75, p ≤ 0.02), higher pH in 2011 only (r = 0.43, p = 0.02), and lower color only in the warm 2015 season (r = -0.71,p < 0.001).

Discussion
Our six-year study in a mechanically-pruned, high-yielding Concord juice grape vineyard in Washington's arid Yakima Valley, which is dependent on irrigation for crop production, demonstrated that (i) moderate (i.e., irrigation at 50% ET c , Ψ s < -1 MPa), but not mild (i.e., irrigation at 75% ET c , Ψ s 7.0 ± 0.2 6.5 ± 0.3 2.7 ± 0.2 11.1 ± 0.5 4.5 ± 0.2 a Means ± SE (n = 20 to 28); the year effect was always significant at p < 0.001.b nd, not determined.varying from 20% to 140% (Williams and Trout 2005).Moreover, θ v (and θ e ) correlated with midday Ψ s , indicating that deficit irrigation led to lower plant water status, which is a common finding in wine and table grapes as well (Permanhani et al. 2016, Scholasch and Rienth 2019, Mirás-Avalos and Araujo 2021).In most years, Ψ s reached a minimum of -1.2 to -1.4 MPa when θ v declined to ≤12% (θ e ≤ 0.3), which typically occurred with moderate deficit irrigation (replacing 50% ET c ).These results indicate that Concord vines experience moderate to severe water stress (sensu Mirás-Avalos and Araujo 2021) below 30% extractable soil water.
The correlation between preveraison Ψ s and trunk diameter and the thinner trunks of the 50% ET c vines are consistent with findings in wine and raisin grapes and suggest that the water-stressed vines produced fewer and/or narrower xylem vessels (Williams et al. 2010b, Munitz et al. 2018).
At the other end of the spectrum, and even with irrigation at 150% ET c , Ψ s plateaued near -0.4 MPa at θ v ≥ 16% (θ e ≥ 0.6), indicating that the water status of Concord is insensitive to soil moisture above 60% relative extractable water.> -1 MPa), water deficit between fruit set and veraison limits canopy size, berry size, and yield; (ii) moderate water deficit after veraison has no effect on canopy size, berry size, or yield; (iii) compensation for inadequate preveraison water supply is not possible by applying more water after veraison; (iv) moderate, but not mild, water deficit before veraison slightly decreases juice acidity; (v) mild water deficit from fruit set to harvest can conserve 20% of the total annual irrigation water with no reduction in yield; and (vi) water savings of ≥30% are possible with moderate water deficit, but reduce yield by >10%.
In general, θ v remained constant under vines that were The curvilinear relationship between θ v and Ψ s looks much like that reported for Thompson Seedless, where increasing irrigation from 100% to 140% ET c did not raise Ψ s (Williams and Trout 2005).Midday Ψ s values above -0.6MPa have been reported to indicate no vine water stress (Mirás-Avalos and Araujo 2021).The Ψ s variability at any given θ v likely resulted from fluctuations in vapor pressure deficit and light intensity (Williams andBaeza 2007, Suter et al. 2019), as plant water status is determined by the interplay between soil water supply and evaporative demand driving canopy transpiration.Therefore, irrigating to keep θ v close to FC wastes irrigation water even for large, mechanically-pruned and heavily-cropped juice grapes.Excess irrigation also favors excessive canopy and weed growth (Williams et al. 2010b, Keller et al. 2016), while allowing the soil to dry down somewhat facilitates vineyard access and weed control.
Over the four years that we monitored water supply and averaged across irrigation treatments, the vineyard received between 586 and 731 mm of irrigation water per year (689 to 826 mm for the 100% ET c vines).This amount is similar to values obtained for mature, fully-irrigated Thompson Seedless grapes in central California (Williams et al. 2010a), but substantially greater than the ~300 mm applied in deficitirrigated Cabernet Sauvignon winegrapes in eastern Washington (Keller et al. 2016).Winegrapes often have similar planting densities, but much smaller canopies, than juice or raisin grapes, and canopy size is a main driver of vineyard water use.The effect of canopy size is reflected in the K c estimated here for the fruit set to harvest period; application of irrigation water at amounts different from 100% ET c precluded estimation of a K c outside of this period.The estimated K c of 1.05 ± 0.09 is similar to that found for Thompson Seedless in California (Williams and Ayars 2005), but greater than those estimated for winegrapes in eastern Washington (Evans et al. 1993).Nevertheless, because of the high crop yield of our juice grapes, their irrigation water use efficiency was 2.5-times greater, while the irrigation water footprint and the total water footprint were >2.5-times lower than those of winegrapes in the same region (cf.Keller et al. 2016).
Our results indicate that deficit irrigation can be applied successfully in large juice grapes with no detrimental effects on crop yield and quality, so long as the preveraison deficit is mild and the postveraison deficit is mild to moderate.Irrigating at 75% ET c from fruit set through harvest reduced whole-season water use by 20% below the 100% ET c control, but decreasing the water supply from 100% to 50% ET c at veraison saved only 8% water on a seasonal basis.
Both deficit treatments generally kept Ψ s above -1 MPa (indicating mild water stress; Mirás-Avalos and Araujo 2021), and neither treatment altered canopy size, yield components, or fruit composition.In contrast, the 30% irrigation water savings achieved by irrigating at 50% ET c from fruit set through veraison (31% if the deficit continued through harvest) was associated with a Ψ s decrease below -1 MPa (indicating moderate water stress; Mirás-Avalos and Araujo 2021), a somewhat smaller and more open canopy, and a 14% reduction in yield below the control.In another study, conducted from 2011 through 2014 in a Concord vineyard in the same region, a comparable decrease in preveraison irrigation water supply reduced yield only in the second year (Stout et al. 2017).However, irrigation generally improved yield after the first year when nonirrigated Concord vines were compared with irrigated vines in Arkansas over four years (Morris et al. 1983) or in Ontario over five years (Reynolds et al. 2005).
Because the period from fruit set to veraison, when days are long and warm and the canopy reaches its full size, dominates the total seasonal irrigation water demand in arid climates (53% in this study), this is the period during which deficit irrigation potentially results in the greatest water savings.However, moderate water deficit at this time limits canopy size, berry growth and, consequently, crop yield (Williams et al. 2010a, Intrigliolo et al. 2012, Junquera et al. 2012, Levin et al. 2020), confirmed here by the curvilinear relationship between preveraison irrigation water supply and yield.The optimal deficit irrigation strategy must balance water conservation and yield sustainability.We propose that the irrigation strategy that best accomplishes this balance in highly-productive juice grapes in eastern Washington is 75% ET c , applied from fruit set through harvest.While severe water stress after veraison can lead to berry weight loss (Hardie andConsidine 1976, Keller et al. 2006), none of our postveraison irrigation treatments induced severe stress, and berry size remained unaltered.
Crop yields were variable both within and between years.Much of this variability was driven by differences in cluster numbers per vine.Variable cluster numbers could have been partly a result of the mechanical pruning strategy that did not control for bud numbers per vine.However, we did not count buds or shoots in this experiment.Extending observations over the prior six-year period in the same vineyard block (Keller and Mills 2021), bigger vines were more productive, as suggested by the correlation between canopy volume and yield per vine, and alternate bearing did not explain the interannual yield variation.While differences in water supply accounted for some of the yield variation, climate variability clearly induced more variation in clusters per vine and berries per cluster from year to year.For example, the unusually light crop in 2012 mirrored that in another irrigation study in a nearby Concord vineyard (Stout et al. 2017).The low cluster numbers in 2012 might have been caused by low bud fruitfulness due to reduced inflorescence initiation during the cool 2011 summer or by the early soil water deficit in 2011 (Vasconcelos et al. 2009, Levin et al. 2020).In addition, there was about one week of unusually cool temperatures (T mean < 15°C) during bloom in 2012, which likely reduced fruit set (Keller et al. 2022).In 2016, the decrease in clusters per vine under moderate preveraison deficit irrigation may have been a consequence of lower bud fruitfulness or lower shoot numbers resulting from longterm water stress (Williams et al. 2010a, Levin et al. 2020).
Although moderate preveraison water deficit (and lower preveraison Ψ s ) decreased berry size, fruit composition varied primarily as a result of seasonal differences and secondarily with crop load, similar to long-term irrigation studies with wine and raisin grapes (Intrigliolo and Castel 2010, Williams et al. 2010a, Junquera et al. 2012).Irrespective of the irrigation treatment, higher TSS, but not smaller berry size, was associated with more intense juice color.Up to the relatively low TSS (15 to 18 Brix) at which Concord grapes are typically harvested, anthocyanin production is linked to sugar accumulation (Hernández-Montes et al. 2021).Irrigation effects on fruit composition were minor and inconsistent, though lower preveraison Ψ s was associated with higher fruit-zone light and lower juice TA at harvest.The smaller canopy of the 50% ET c vines allowed greater light penetration into the fruit zone, which would have increased cluster temperatures and malate catabolism during ripening (Sweetman et al. 2014, Keller et al. 2016).

Conclusions
Though juice grape producers have traditionally tried to avoid plant water stress, the present study demonstrated that deficit irrigation can be applied successfully in large, mechanically-pruned Concord juice grapes (K c ≈ 1 between fruit set and harvest), as long as the irrigation strategy induces only mild plant water stress (Ψ s > -1 MPa) during the preveraison period.While moderate water stress (-1 MPa > Ψ s > -1.4 MPa) before veraison decreased canopy size, berry weight, crop yield, and juice acidity compared to nonstress conditions, moderate water stress after veraison, or mild water stress during either period, did not alter any of these variables.Our results suggest that presumed impacts of moderate postveraison water stress may often be due to carryover effects from stress that started before veraison.Trying to compensate for preveraison water stress by applying more irrigation water after veraison proved ineffective at preventing yield losses.Among the eight drip-irrigation treatments tested over up to six years, replacing 75% of ET c from fruit set through harvest was the optimal strategy: it reduced annual irrigation water use by 20% compared with the 100% ET c control, without altering yield or fruit composition.Reducing irrigation to 50% ET c increased water savings, but incurred a yield penalty with no gain in juice quality.While prebloom and postharvest water supply in arid climates should aim to avoid soil and plant water deficit to maximize vine productivity, mild preveraison water deficit and mild to moderate postveraison deficit are desirable for juice grape production.

Figure 1
Figure 1 Seasonal trends in growing degree days (GDD, base 10°C) in a Concord juice grape vineyard in southeastern Washington over six years.

Figure 2
Figure 2 Annual irrigation water supply (A) and proportion of water supply by phenological period (B) for mechanically-pruned Concord juice grapes planted in 2003 in southeastern Washington and irrigated at various fractions of crop evapotranspiration (ET c preveraison/postveraison). The 75/100 treatment was replaced with the 150/100 treatment in 2014; the 100/75 and 75/75 treatments were not applied in 2016.Bars in A show means ± SE; the sequence of treatments is identical in A and B (B, budbreak; S, fruit set; V, veraison; H, harvest).

Figure 3
Figure 3 Seasonal trends over six years of volumetric water content (θ v ) in the soil's top 60 cm and midday stem water potential (Ψ s ) of mechanically pruned Concord juice grapes planted in 2003 in southeastern Washington and irrigated at different fractions of crop evapotranspiration (% ET c ) applied preveraison/postveraison (S, fruit set; V, veraison; H, harvest).Data show means ± SE.

Figure 4
Figure 4 Associations over four years between irrigation water amount applied from fruit set to veraison and average preveraison relative extractable soil water (θ e ; r = 0.86) in the soil's top 60 cm (A) and average preveraison midday stem water potential (Ψ s ; r = 0.67) (B), and between instant θ e and Ψ s (r = 0.62) over the fruit set to harvest period (C) of mechanically-pruned Concord juice grapes planted in 2003 in southeastern Washington and irrigated at various fractions of crop evapotranspiration (ET c ). Symbol color indicates ET c and symbol shape indicates year (all p < 0.001).

Figure 5
Figure 5 Effects of irrigation treatment (in % ET c preveraison/postveraison) and growing season on preharvest canopy surface area (A), canopy volume (B), and fruit-zone light (C) of mechanically-pruned Concord juice grapes planted in 2003 in southeastern Washington.Bars show means ± SE.The 75/100 treatment was replaced with the 150/100 treatment in 2014; the 100/75 and 75/75 treatments were not applied in 2016.

Figure 7
Figure 7 Association over four years between annual irrigation water supply and crop yield (A) and between irrigation water supply from fruit set to veraison and crop yield (B) of mechanically-pruned Concord juice grapes planted in 2003 in southeastern Washington (all p < 0.02).

Figure 6
Figure 6 Effects of irrigation treatment (in % ET c preveraison/postveraison) and growing season on crop yield (A), cluster weight (B), and berry weight (C) in mechanically-pruned Concord juice grapes planted in 2003 in southeastern Washington.Bars show means ± SE.The 75/100 treatment was replaced with the 150/100 treatment in 2014; the 100/75 and 75/75 treatments were not applied in 2016.

Table 1
Weather conditions and timing of key phenological stages for the Washington State University Concord research vineyard in southeastern Washington from 2011 to 2016.Data were obtained from an AgWeatherNet station located ~500 m from the trial site.
a Average for 1989 to 2016.b GDD, growing degree days (base 10°C) accumulated from 1 April to 31 Oct. c GST, growing season temperature.d Average daily maximum temperature for the fruit set to veraison period.e Average daily maximum temperature for the veraison to harvest period.f Grass reference evapotranspiration.g DOY, day of year for 50% occurrence of phenological stages.

Table 2
Effect of growing season on yield and its components and on harvest fruit composition of mechanically-pruned Concord juice grapes planted in 2003 in southeastern Washington.TSS, total soluble solids.