Apparent hydraulic strategies of diverse winegrape cultivars depend on the choice of driving and response variables

Abstract

Background

Current and projected increases in the frequency and intensity of drought episodes require adaptation of irrigation management for crop plants. Such adaptation is complicated by incomplete knowledge of intra-species differences in physiological responses to drought stress. For instance, winegrapes have been variously classified as behaving isohydrically or anisohydrically depending on the cultivar and the environment. Aiming to determine whether cultivars should be irrigated differently depending on their hydraulic strategy, this study investigated the behavior of 30 winegrape cultivars with diverse geographic origins in a common garden experiment. Using controlled irrigation in an arid climate, we imposed two successive soil drydown cycles in each of two years and measured changes in canopy size, soil moisture, meteorological conditions, and plant water status.

Results

Though the canopy size varied fivefold across cultivars, their hydraulic strategy did not differ but changed depending on the extent of drought stress and the choice of driving and response variables. When soil moisture was expressed as relative extractable soil water (θ_e), all cultivars behaved isohydrically (constant predawn and midday leaf water potential, Ψ_leaf) as the soil dried from θ_e > 1 to 0.33, and anisohydrically (decreasing Ψ_leaf) at θ_e < 0.33. In contrast, when soil moisture was expressed as soil tension (Ψ_soil), or when predawn Ψ_leaf was used as a proxy for Ψ_soil, all cultivars gradually changed their midday Ψ_leaf response from anisohydric to isohydric as the tension increased.

Conclusion

At least in winegrapes, the choice of driving and response variables, and the range over which they are measured, strongly influences conclusions about plant hydraulic strategies and may alter modeling outcomes. This offers a benchmark for identifying water deficit and has implications for field data interpretation and irrigation management. Irrigation scheduling may be guided by common soil- or plant-based thresholds across cultivars, while the time to reach such thresholds depends on differences in canopy size among cultivars and on weather patterns and soil properties.

Key findings

• Variable choice matters: Conclusions about plant hydraulic strategy depend on whether soil moisture is expressed as volumetric water content or water potential, and on the degree of drought stress.

• Shared threshold versus gradual change: 30 winegrape cultivars differing fivefold in canopy size transitioned from isohydric to anisohydric behavior at a common threshold of volumetric soil water content, but changed gradually from anisohydric to isohydric behavior as water potential declined.

• Similar hydraulic strategy: All 30 cultivars showed similar changes in leaf water status in response to declining soil moisture, indicating limited within-species variation in adaptation to drought stress.

• No isohydric–anisohydric categorization: The 30 cultivars did not differ in hydraulic strategy; responses were modulated by drought severity and measurement definitions, undermining strict classification of hydraulic strategies.

Practical implications

• Irrigation management: Irrigation scheduling may use shared physiological thresholds across cultivars but should consider differences in vigor and hence canopy size that affect water use.

• Modeling: Physiological models must account for differences in variable choice and measurement range to avoid misleading outcomes.

• Breeding and adaptation: Winegrape cultivars may lack diversity in hydraulic strategies, suggesting limited potential for genetic adaptation to intensifying drought stress.Breeding and adaptation: Winegrape cultivars may lack diversity in hydraulic strategies, suggesting limited potential for genetic adaptation to intensifying drought stress.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-07837-4.

Introduction

Drought stress impairs plant growth and development, hence the yield and quality of crops and their products [1, 2]. This is because drought limits water and nutrient uptake, cell division and expansion, carbon assimilation and partitioning, and alters gene expression [3–8]. Therefore, the increasing frequency and intensity of droughts in many areas of the world [9, 10] make understanding drought tolerance an important step in crop improvement and management [11, 12].

Though plant responses to drought stress have been studied extensively, the argument that there are two distinct categories of plants in terms of physiological responses to drought stress remains controversial. The literature suggests that when subjected to drought stress, some plant genotypes (species, cultivars) maintain near-constant leaf water potential (Ψ_leaf) regardless of the soil water potential (Ψ_soil) and vapor pressure deficit (VPD), while others gradually lower their Ψ_leaf as Ψ_soil declines and with increased VPD [13–16]. The former genotypes have been classified as isohydric, and the latter as anisohydric. Critical to this categorization is the stomatal regulation of plant water loss. Isohydric plants are said to tightly regulate the opening of their stomatal pores, thus lowering the stomatal conductance (g_s) as Ψ_soil declines, whereas the anisohydric plants have less stomatal regulation [13, 17–19]. However, the literature presents some inconsistencies that demonstrate our incomplete understanding of the differences among genotypes in the response of leaf hydraulic and physiological traits to progressively intensifying soil water deficit.

In its current form, the classification of plants as isohydric or anisohydric is mostly based on the relationship between midday Ψ_leaf (Ψ_md) and predawn Ψ_leaf (Ψ_pd), the latter typically serving as a proxy for Ψ_soil, or between g_s and Ψ_leaf [17, 18, 20, 21]. However, not only do single metrics not define plant responses to stress [2, 22, 23], but tighter stomatal control also does not necessarily equate to greater regulation of Ψ_leaf [24–26]. Moreover, some reports suggest that an-/isohydric behavior between or within species may be more of a continuum than a dichotomy [25, 27].

Because a wide range of grape (Vitis vinifera L.) cultivars are available for use in the potential mitigation of the impact of drought stress [28–30], we decided to focus on a diverse panel of grape cultivars to reassess the basis on which the isohydric versus anisohydric behavior claim is made. Following their domestication in Southwest Asia around 11,000 years ago [31], grapevines were selected and bred for cultivation in environments ranging from cool to hot and dry to humid. Thus, it seems reasonable to expect that V. vinifera cultivars also vary in stomatal behavior and Ψ_leaf regulation in response to soil moisture deficit. Bartlett & Sinclair [29] reported that cultivars with greater vulnerability to xylem embolism and more negative Ψ_leaf tended to be associated with cooler regions, while those with traits that promoted tighter stomatal regulation were associated with warmer regions. However, the literature is inconsistent regarding the behavior of the same cultivar in different locations or even in the same location at different times [17, 18, 32–39]. Furthermore, some authors hinted that cultivars might switch their hydraulic strategy depending on soil moisture [22, 40]. But even the question of how we define soil moisture when studying the behavior of plants in drying soil remains unresolved. Volumetric soil water content (θ_v) and Ψ_soil are the two most widely used terms in characterizing plant responses to soil water deficit [41–43]. The former refers to the relative amount of water in the soil and the latter to the force by which water is held in the soil [44, 45]. Soil water retention curves are required to describe the relationship between θ_v and Ψ_soil. In addition, because θ_v is dependent on soil texture [46], there is a need to convert θ_v to relative extractable soil water (θ_e) to normalize the soil texture effect [22, 47, 48]. There are studies in which plant behavior was described purely based on θ_v [15, 35, 49]. But in most studies, neither θ_v nor Ψ_soil is measured, and Ψ_pd is used as a proxy for Ψ_soil [50, 51] and to define plant responses to soil drying [e.g., 13, 17, 18, 27]. However, Ψ_pd and Ψ_soil are not identical and not necessarily linearly related, and their association may depend on the extent of nighttime transpiration [51–54].

These inconsistencies suggest that V. vinifera cultivars may be both isohydric and anisohydric, depending on definitions and assumptions, leading to further discrepancies. For instance, Zhang et al. [22] reported that Merlot vines were anisohydric under high θ_e and became isohydric with decreasing θ_e. A similar transition from anisohydric to isohydric as the soil dried down was reported for Syrah and Garnacha [55], two cultivars that were thought to be known for their contrasting response to water stress [17]. By contrast, despite reporting a continuum of stomatal responses to water deficit among 17 V. vinifera cultivars, including Syrah and Garnacha, Levin et al. [27] found that all cultivars responded to the water deficit in an anisohydric manner. It may be telling that across 19 Vitis cultivars, the stem xylem anatomical traits were similar, and there was no correlation between these and apparent hydraulic traits [56]. In most published studies, moreover, vines were grafted to rootstocks (which are typically derived from interspecific crosses), but the extent to which the rootstock influences the physiological behavior (or presumed scion differences) is still unclear [57]. Finally, the disease status (e.g., infection by viruses, bacteria, or fungal trunk-disease pathogens) is rarely reported in physiological studies.

Our study aimed to determine whether winegrape cultivars or cultivar groups should be irrigated differently depending on a cultivar’s hydraulic strategy, assuming a continuum of physiological responses to soil drying across cultivars. Through a combination of common garden experiment and (replicated) gradient experiment, we investigated the response of 30 own-rooted and disease-tested V. vinifera cultivars to soil moisture deficit over two growing seasons. We used the arid local climate (< 200 mm annual precipitation) to simulate conditions that occur naturally in Mediterranean climates with ample winter precipitation and limited or no summer rainfall. Though our field experiment resembled that of Franks et al. [58], who followed two natural seasonal drydown cycles interrupted by rainfall in a eucalypt woodland, we used precise irrigation to control changes in soil moisture. We recorded the phenology and canopy size of each cultivar and repeatedly measured θ_v, plant water status, and leaf gas exchange, generating a data set of about 1200 coupled measurements. These measurements were used to calculate θ_e, Ψ_soil, and various traits commonly used to assess plant hydraulic strategies, which were then compared to determine if the different traits convey the same message. This paper describes responses related to canopy size and Ψ_leaf. Stomatal responses and cultivar rankings for hydraulic traits will be reported in a separate communication.

Materials and methods

Site and plant material

An experiment was conducted during the 2021 and 2022 growing seasons in a vineyard at Washington State University’s Irrigated Agriculture Research and Extension Center (46°17’49"N; 119°44’07"W; 365 m above sea level) in southeastern Washington, USA. The region has very low annual precipitation (< 200 mm) and high evapotranspiration (1040 mm; 20-year average for Roza.2 AgWeatherNet station; https://weather.wsu.edu). The soil is a Warden silt loam with a depth of about 50 to 100 cm above an impermeable hardpan [51]. The vineyard was planted in 2010 with 30 own-rooted, virus-tested V. vinifera winegrape cultivars originating from diverse European regions with varied genetic backgrounds and published hydraulic strategies ranging from isohydric to strongly anisohydric (Table 1). Vines were grown in north‒south-oriented rows and trained to bilateral cordons, which were trained horizontally 0.95 m aboveground and pruned to 15 spurs with two buds each, with loosely vertically positioned shoots. The vines were grown with a spacing of 2.7 m (between rows) by 1.8 m (within rows), with five vines per cultivar in three or four replicated rows (15‒20 vines per cultivar). The vineyard was drip-irrigated with four 2-L h^{− 1} emitters per vine and a shut-off valve for individual control in each row. A 1.2-m herbicide strip was maintained under the vines, and permanent but summer-dormant resident vegetation was grown between rows. Annual disease testing and rogueing were done to keep the vineyard free of viral and bacterial diseases. Other pest and disease control practices followed regional industry standards.

Table 1.

Winegrape cultivars in the Washington State University research vineyard, with color of the berry skin, country of origin, and published hydraulic strategy. Countries of origin were obtained from Foundation Plant Services, University of California, Davis (https://fps.ucdavis.edu/fgrvarieties.cfm)

Cultivar	Berry color	Country of origin	Hydraulic strategy	References
Albariño	White	Portugal/Spain	Isohydric	[59]
Aligoté	White	France	Unknown
Auxerrois	White	France	Unknown
Barbera	Red	Italy	Unknown
Cabernet franc	Red	France	Anisohydric	[33]
Cabernet Sauvignon	Red	France	An-/isohydric	[33, 36, 39, 60, 61]
Chardonnay	White	France	An-/isohydric	[33, 35, 38, 49, 62]
Chenin blanc	White	France	Unknown
Durif	Red	France	Unknown
Garnacha	Red	Spain	An-/isohydric	[17, 34, 37, 39, 49, 62–65]
Gewürztraminer	White	Italy	Unknown
Grüner Veltliner	White	Austria/Italy	Unknown
Lemberger	Red	Austria	Unknown
Malbec	Red	France	An-/isohydric	[49, 61]
Melon	White	France	Unknown
Merlot	Red	France	Anisohydric	[38, 62]
Monastrell	Red	Spain	Isohydric	[34, 66]
Muscat blanc	White	Greece	Anisohydric	[33]
Nebbiolo	Red	Italy	Unknown
Petit Verdot	Red	France	Unknown
Pinot blanc	White	France	Unknown
Pinot gris	White	France	Unknown
Pinot noir	Red	France	An-/isohydric	[38, 67]
Riesling	White	Germany	Anisohydric	[33, 62]
Sangiovese	Red	Italy	Anisohydric	[68]
Sauvignon blanc	White	France	Isohydric	[38]
Sémillon	White	France	Anisohydric	[40, 64]
Tempranillo	Red	Spain	An-/isohydric	[33, 62, 69]
Viognier	White	France	Unknown
Zinfandel	Red	Croatia	Anisohydric	[60]

Drydown cycles

In both experimental seasons, all 30 cultivars were irrigated similarly from budbreak through anthesis to avoid plant water stress. Irrigation was interrupted at fruit set to create a gradual soil moisture deficit. Because the cultivars represented in our experiment do not reach the same phenological stage at the same time ([70], Additional file 1), irrigation was interrupted when the late cultivars, such as Monastrell, reached about 50% fruit set. A single irrigation to restore soil moisture to near-field capacity was applied during the early or mid-ripening phase to prevent excessive drought stress, as some vines showed leaf wilting or yellowing symptoms. In 2021, the first drydown cycle started on 23 June and lasted 42 days. Following rewatering, the second cycle started on 17 August and lasted 57 days. In 2022, the first cycle began on 5 July and lasted 70 days, and the second cycle (after rewatering) started on 22 September and lasted 18 days. At the end of the season, all vines were irrigated to near-field capacity to avoid water stress and cold injury to the roots during winter.

Data collection

Weather data were obtained from the Roza.2 AgWeatherNet station located ~ 550 m from the vineyard (https://weather.wsu.edu). Daily and cumulative growing degree days (GDD) were estimated from daily maximum (T_max) and minimum temperatures (T_min), using a base temperature of 10 °C, between 1 April and 31 October. The daily average vapor pressure deficit (VPD_mean) between 1 April and 31 October, and the predawn VPD (VPD_pd, between 01:00 h and 03:00 h) and midday VPD (VPD_md, between 13:00 h and 15:00 h) averaged over the daily time windows of Ψ_pd and Ψ_md measurements, were calculated from the Tetens equation [71] and relative humidity (RH) data from the weather station:

1

Where T is the temperature in °C.

Plant phenology was assessed through visual observation of three to four groups of four plants per cultivar, conducted two to three times per week. Phenological stages were expressed using the modified Eichhorn‒Lorenz system [72]. Shoot growth before irrigation was suspended was assessed at anthesis by measuring the length, from the base to the growing tip, of two uniformly growing shoots on either side of the trunk of six randomly selected vines per cultivar, three of which were also used for physiological measurements. The mean value from the two shoots was used for data analysis (i.e., n = 6). To obtain a robust and seasonally integrated measure of canopy size, all annual canes (1-year-old stems) from the same six vines were counted and weighed during winter pruning, which occurred in February of the year following each experimental season.

Throughout the first drydown cycle in both 2021 and 2022, soil and plant water status were measured every 14 days. During the second cycle in 2021, measurements were taken on days 1, 15, 36, and 57 after the mid-season irrigation event. In the second cycle of 2022, data were collected only once, on the day after the irrigation event. The lower measurement frequency during the second drydown cycle in 2021 was due to the slow soil drying towards the end of the season, as photoperiod, temperature, and VPD decreased, and the much later start of the second cycle in 2022 resulted in almost no θ_v decline. On each day of measurement, data were collected from three vines per cultivar, with one vine sampled from each of the three replicate rows (i.e., n = 3). Our sample size was limited by the need to conduct all measurements within environmentally consistent time windows. A neutron probe (Hydroprobe Am/Be model 503DR, CPN International Concord, CA, USA) was used for θ_v measurements, which were conducted between 07:00 h and 09:00 h at 30 cm and 60 cm depth, using PVC access tubes installed ~ 90 cm from the trunk of each data vine and between two drip emitters. The higher of the two θ_v values per measurement was used in subsequent data analysis because Ψ_leaf generally equilibrates with the wettest portion of the soil [51, 53, 73]. To permit comparison of soil moisture across different soil types, θ_v was converted to θ_e (unitless), defined as the relative soil water content normalized to field capacity (θ_fc) and permanent wilting point (θ_pwp) as follows [22, 47]:

2

To calculate θ_e, we used the local θ_fc of 26% (v/v) [51], but we adjusted the θ_pwp to 7% (from the laboratory-measured 8% in ref. 51) because soil drying to 7% (rather than 8%) led to leaf abscission; thus, θ_fc ‒ θ_pwp = 19%. The first measurement in each drydown cycle was conducted immediately after irrigation when θ_v >θ_fc (i.e., θ_e >1). The Ψ_soil was estimated from θ_v data using the van Genuchten [74] procedure as described previously for this vineyard [51]. Note that θ_e differs from the normalized water content used by van Genuchten [74] in that the former expresses water content relative to the plant-relevant θ_fc and θ_PWP, while the latter uses saturated and residual water contents, which are soil properties and are higher and lower than θ_fc and θ_PWP, respectively.

A pressure chamber (model 615D, PMS Instrument Company, Albany, OR, USA) was used to measure Ψ_pd and Ψ_md on vines adjacent to each neutron probe access tube. The Ψ_pd measurements were conducted between 01:00 h and 03:00 h, and the Ψ_md between 13:00 h and 15:00 h (Pacific Daylight Time). Diurnal measurements performed in this vineyard indicated that Ψ_md remained near a minimum between 12:00 h and 16:00 h [75]. All Ψ_leaf measurements were conducted on fully expanded leaves between the 7th and 15th node from the growing tip of the main shoot, following the protocol described by Turner [76].

We also fitted the transition points Θ₁ and Θ₂ in the Ψ_pd—Ψ_md response curve. According to Knipfer et al. [77], Θ₁ marks the boundary between mild and moderate drought stress and corresponds to the Ψ_leaf at which stomatal closure occurs, whereas Θ₂ (Ψ_pd = Ψ_md) is the boundary between moderate and severe drought stress and corresponds to the Ψ_leaf at which leaf turgor loss occurs (i.e., turgor loss point). Because our field-measured data did not extend all the way to the published Θ₂ boundary [77], which was obtained with very small potted plants, we extrapolated our data to the 1:1 line using the average of the 10 lowest Ψ_md values each year, since the slope of the Ψ_pd—Ψ_md curve approached zero below Θ₁ as observed elsewhere [18, 30, 77].

Data analysis

To determine differences in canopy size among the 30 cultivars, data were subjected to analysis of variance via SAS^® OnDemand for Academics (SAS Institute, Cary, NC, USA). The effects of cultivar, year, and their interaction were determined through the general linear model procedure (Proc GLM). To check for temporal consistency of the differences in canopy size among cultivars, cultivar means of pruning weights were ranked each year. The ranks were compared by Spearman’s rank correlation (Proc Corr Spearman). Significant results are reported at a 5% error probability (i.e., P ≤ 0.05) unless otherwise stated.

To analyze the response of Ψ_leaf to drought stress, we tested several models and chose the best fit. The Ψ_leaf data showed a nonlinear relationship with both θ_e and Ψ_soil. The responses of Ψ_md to Ψ_pd were also nonlinear. The θ_e—Ψ_leaf response curve was similar to the photosynthetic light response curve [78, 79]. In contrast, the Ψ_soil—Ψ_leaf and Ψ_pd—Ψ_md response curves resembled exponential decay curves. Thus, to determine the response of Ψ_leaf to θ_e and Ψ_soil, or of Ψ_md to Ψ_pd, the data were subjected to nonlinear regression analysis through the Gauss-Newton method (Proc nlin) with an equation of the form:

3

Where Ψ_leaf was Ψ_pd or Ψ_md, Ψ_asy was the maximum or minimum threshold (asymptote) of Ψ_pd or Ψ_md, a was a curve parameter describing some shift along the x-axis, b was the slope, and c was either θ_e, Ψ_soil or Ψ_pd. The differences in parameter estimates (i.e., a, b, and c) between any two cultivars or cultivar groups were compared through a series of pairwise Gauss-Newton analyses. Comparisons were conducted between years, between cultivars with presumed divergent responses to soil moisture deficit (i.e., Garnacha and Sémillon, which are classically categorized as isohydric and anisohydric, respectively [17, 40], and between cultivars with contrasting canopy sizes. Whenever the difference between Garnacha and Sémillon was not significant (i.e., P >0.05), the data from the two cultivars were combined and used as the control against which all other cultivars were compared.

The θ_e threshold was the point at which a θ_e—Ψ_leaf response curve changed its characteristic. This point was calculated from the overall θ_e—Ψ_leaf response data as follows:

4

Where ln was the natural logarithm and b the slope of the θ_e—Ψ_leaf response curve.

Results

Weather and phenology

Precipitation and temperature patterns differed between the two years (Fig. 1A): 2021 was hotter and received half the total annual precipitation (121 mm) compared to 2022 (228 mm). The early growing season between April and June was much drier in 2021, with 6 mm of rainfall compared to > 100 mm in 2022. July and August were very dry, without significant rainfall in either year. However, more rain fell during the fruit ripening period in September and October of 2021 than in 2022. Compared to the 20-year (2001‒2020) average precipitation of 170 mm, 2021 was 29% drier, while 2022 was 34% wetter. The cumulative GDD from April to October was higher in 2021 (1675 GDD) than in 2022 (1607 GDD), mostly because of the warmer spring of 2021. However, both growing seasons were considerably warmer than the long-term average of 1466 GDD. The average VPD_mean for the growing season was higher in 2021 (1.24 kPa) compared with 2022 (1.04 kPa). In both years, the VPD_mean was 53% to 62% higher during the first drydown cycle than the second (Fig. 1B). In 2021, the VPD_mean for the first drydown period was 2.08 kPa, reaching a high of 4.06 kPa and a low of 1.07 kPa, compared to a VPD_mean of only 0.79 kPa with a 1.75 kPa high and a 0.36 kPa low in the second drydown. In 2022, the VPD_mean for the first drydown period was 1.6 kPa, with a 2.87 kPa high and a 0.37 kPa low, compared to a VPD_mean of 0.76 kPa with a high of 1.09 kPa and a low of 0.43 kPa in the second cycle.

Fig. 1.

Average monthly precipitation and cumulative growing degree days (GDD; A), and daily mean vapor pressure deficit (VPD_mean; B) during the 2021 and 2022 growing seasons in a vineyard in southeastern Washington, USA. Arrows indicate the start and end of each drydown cycle. 1–2021 and 1–2022 are the first drydown cycles and 2–2021 and 2–2022 are the second drydown cycles in 2021 and 2022, respectively. Weather data were obtained from the Roza.2 AgWeatherNet station located ~ 550 m from the vineyard

Compared with 2021, the cool spring of 2022 delayed the average date of budbreak across the 30 cultivars by 6 days and lengthened the average period from budbreak to anthesis by 21 days (Additional file 1). Cultivars varied by up to four weeks for key phenological stages from the earliest to the latest. For instance, there was a 10-day difference in budbreak between an early (Chardonnay) and a late (Monastrell) cultivar in 2021 compared to a three-week difference for the same cultivars in 2022 (Additional file 1).

Canopy size

The interaction between cultivar and year had no significant impact on winter pruning weight (P = 0.22), which we used as a robust proxy for canopy size. Only the main effects of cultivar (P < 0.001) and growing season (P < 0.001) were significant (Fig. 2). Moreover, Spearman’s correlation coefficient for the cultivar ranking in pruning weights between 2021 and 2022 was r_s = 0.93 (P < 0.001). Consequently, while the canopies were, on average, more than 40% smaller after the dry 2021 (0.84 kg per plant) than the comparatively wet 2022 (1.49 kg per plant), the differences in canopy size among the 30 cultivars were consistent across the two years. Under identical irrigation regimes, the most vigorous cultivars had almost fivefold heavier pruning weights than the least vigorous ones (Fig. 2). For instance, the 2-year average pruning weight of Cabernet Sauvignon was 1.9 kg compared to only 0.4 kg in Melon. However, Garnacha and Sémillon, two cultivars with a presumed divergent response to soil moisture deficit, had similar pruning weights (Fig. 2). Nonetheless, canopy size was not related to cultivar origin (Table 1), and even full siblings (e.g., Auxerrois, Chardonnay, and Melon) were distributed over the entire spectrum (Fig. 2). Pearson correlation showed that pruning weight was much more strongly correlated with shoot length (r_p = 0.69; P < 0.001) than with the number of shoots per plant (r_p = 0.29; P < 0.001) since the former was allowed to vary freely while the latter was restricted by annual winter pruning.

Fig. 2.

Canopy size, measured as winter pruning weight, of 30 winegrape cultivars grown in a vineyard in southeastern Washington, USA. All dormant canes removed during winter pruning of six randomly selected vines per cultivar were weighed following the 2021 and 2022 growing seasons. Bars show means ± SE. Means with different letters within years differ significantly according to Tukey’s HSD test (P < 0.05)

Seasonal trends of soil moisture and leaf water potential

Following irrigation to near-θ_fc at fruit set, the soil at both measured depths and across all 30 cultivars dried down much faster in 2021 than in 2022 (Fig. 3A), consistent with the temperature, precipitation, and VPD trends in the two years (Fig. 1). The most rapid soil drying occurred during a series of three 4- to 6-day heatwaves (T_max >35 °C) between late June and mid-July of 2021, with T_max peaking at 44.2 °C (Fig. 3A). Similar heatwaves also occurred in 2022, but they did not start until mid-July. In 2021, it took about three weeks for the soil to dry from θ_e ≈ 0.9 to θ_e ≈ 0.2 in the first drydown cycle compared to six weeks in 2022. Moreover, θ_e did not decrease below 0.15 in 2022, unlike in 2021, when θ_e fell below 0.1 before irrigation water was applied to end the first drydown cycle. Consequently, the second cycle lasted longer in 2021 than in 2022. In 2021, the midseason irrigation event occurred when all cultivars except Durif and Melon, the two cultivars with the smallest canopy, had visible symptoms of severe water stress (i.e., leaf wilting or yellowing, as shown in Additional file 2). In 2022, mid-season irrigation occurred when nearly half the cultivars (i.e., Albariño, Cabernet franc, Malbec, Nebbiolo, Petit Verdot, Pinot gris, Pinot noir, Sauvignon blanc, Tempranillo, Viognier, and Zinfandel) showed signs of severe water stress.

Fig. 3.

Changes in relative extractable soil water (θ_e) at two soil depths and daily maximum temperature (T_max; A), and leaf water potential (Ψ_leaf; B) during two drydown cycles in 2021 (top) and 2022 (bottom) in a vineyard in southeastern Washington, USA. Symbols show means for 30 winegrape cultivars ± SE (invisible due to the size of the symbols)

When the data from all 30 cultivars were pooled, the shape of the change in Ψ_leaf over the course of the drydown cycles differed markedly from that of the shift in θ_e (Fig. 3). Unlike the rapid decline of θ_e early on in each cycle, the decrease in Ψ_leaf only began to accelerate later in the cycle, when θ_e was beginning to bottom out. Consistent with the seasonal differences in weather patterns (Figs. 1 and 3B), both Ψ_pd and Ψ_md decreased earlier and to a greater extent in 2021 than in 2022 (Fig. 3B). At θ_e = 0.2, Ψ_pd was − 0.4 MPa in 2021 and − 0.5 MPa in 2022, while Ψ_md was − 1.0 MPa in both years. By the end of the first drydown in 2021, the average Ψ_pd had dropped below − 0.7 MPa and Ψ_md below − 1.3 MPa. In contrast, in 2022, the average Ψ_pd only dropped to −0.5 MPa and Ψ_md to −1.1 MPa at the end of the same period. The midseason irrigation event restored θ_e to near-θ_fc and returned Ψ_leaf to non-stress levels in both years. During the second drydown cycle of 2021, Ψ_leaf decreased only slightly but significantly (P < 0.001). Further, Ψ_pd and Ψ_md were not only higher (P < 0.001) but also closer together (P < 0.001) than during the first drydown cycle.

The average difference between Ψ_pd and Ψ_md (ΔΨ_leaf) across all cultivars was 0.52 MPa in the first cycle and 0.24 MPa in the second cycle. In 2022, ΔΨ_leaf was 0.43 MPa (first cycle) and 0.4 MPa (second cycle, when Ψ_leaf was measured only once). The ΔΨ_leaf for the pooled data from all cultivars correlated positively with VPD_mean (r_p = 0.43; P < 0.001), indicating a greater daily drop in Ψ_leaf at higher VPD (Additional file 3). This correlation was also significant for each cultivar (r_p >0.33; P < 0.05) except Cabernet Sauvignon, Monastrell, and Nebbiolo (Additional file 3). As shown in Fig. 4A, ΔΨ_leaf correlated more closely with VPD_mean at θ_e > 0.33 (r_p = 0.44; P < 0.001) than at θ_e < 0.33 (r_p = 0.15; P < 0.001). The negative correlation between Ψ_md and VPD_md (Fig. 4B) was also stronger at θ_e > 0.33 (r_p = −0.46; P < 0.001) than at θ_e < 0.33 (r_p = −0.25; P < 0.001). Consequently, Ψ_leaf was more responsive to VPD in wet than dry soil.

Fig. 4.

Relationship between the difference between predawn (Ψ_pd) and midday (Ψ_md) leaf water potential (ΔΨ_leaf) and daily average vapor pressure deficit (VPD_mean; A), and Ψ_md and daily maximum VPD at the time of Ψ_md measurement (VPD_md; B) above and below a relative extractable soil water (θ_e) threshold of 30 winegrape cultivars grown in a vineyard in southeastern Washington, USA. Each vertical group of data represents all cultivars on a single day of measurement over two drydown cycles each in 2021 and 2022. VPD_mean and VPD_md were extracted for each Ψ_pd and Ψ_md measurement from weather data obtained from the Roza.2 AgWeatherNet station located ~ 550 m from the vineyard. n = number of observations

Apart from some minor shifts, the general response of Ψ_leaf to either θ_e or Ψ_soil was the same across the two growing seasons when all 30 cultivars were pooled together (Additional files 4 and 5). In the cultivars with the largest canopy (e.g., Cabernet Sauvignon), but not in those with a small canopy (e.g., Melon, Durif), the θ_e—Ψ_leaf curve declined somewhat more steeply in 2022 than in 2021. However, the apparent lack of flattening of the Ψ_leaf versus Ψ_soil curves at low Ψ_soil in 2022 (Additional file 4 C, D) is a consequence of the less severe soil drying that year (Fig. 3), which resulted in fewer data points in the dry region compared to 2021. Because the Ψ_leaf responses to soil drying were consistent between the two growing seasons and the two drydown cycles (Additional files 4 and 6), the data from both years and both cycles were pooled for all subsequent analyses.

Response of leaf water potential to soil drying using two definitions of soil moisture

The overall response of Ψ_leaf to either θ_e or Ψ_soil across 30 cultivars was nonlinear. The Ψ_leaf response to θ_e followed an asymptotic regression curve (Fig. 5A). In contrast, the response of Ψ_leaf to Ψ_soil resembled an exponential decay curve (Fig. 5B). When Ψ_leaf was plotted against θ_e, the vines were isohydric at θ_e > 0.33 (i.e., the θ_e value at which the θ_e—Ψ_leaf response curve changed its characteristic as given by Eqs. 3 and 4) and anisohydric as the water deficit became more severe. By contrast, when Ψ_leaf was plotted against Ψ_soil, the vines initially behaved anisohydrically in response to soil drying and became gradually more isohydric as Ψ_soil declined. In addition, the cultivar-averaged ΔΨ_leaf in both the θ_e—Ψ_leaf and the Ψ_soil—Ψ_leaf relationships gradually increased from 0.3 MPa in moist soil to 0.62 MPa in dry soil. The variation around the average Ψ_leaf curves increased in both cases as the soil dried. For instance, the range of Ψ_leaf at θ_e > 0.33 was c. 0.4 MPa (Ψ_pd) and 0.7 MPa (Ψ_md), respectively, while it was c. 1.2 MPa for either Ψ_pd or Ψ_md at θ_e < 0.33. Over much of the soil moisture range, the variation in Ψ_leaf must have been dominated by other factors, such as temperature and VPD. Pearson correlation analysis showed that the pooled ΔΨ_leaf data were positively correlated with T_pd, T_md, daily average temperature (T_mean) (r_p >0.36; P < 0.001), and VPD_mean (Fig. 4A and Additional files 3 and 7). In contrast, both Ψ_pd and Ψ_md had negative correlations with T_pd, T_md, and T_mean, but the correlations with Ψ_md (r_p > −0.40; P < 0.001) were stronger than those with Ψ_pd (r_p < −0.28; P < 0.001). The pooled Ψ_pd data also showed a negative correlation with VPD_pd (r_p = −0.28; P < 0.001) and only for 17 of the 30 cultivars (Additional files 3 and 7). The pooled Ψ_md data correlated negatively with VPD_md (r_p = −0.43; P < 0.001). Among the 30 cultivars, 28 had a significant VPD_md—Ψ_md correlation (r_p = −0.31; P ≤ 0.05), but not Cabernet Sauvignon (r_p = −0.28; P = 0.06) and Monastrell (r_p = −0.23; P = 0.13).

Fig. 5.

Overall relationship of leaf water potential (Ψ_leaf) with either relative extractable soil water (θ_e) based on volumetric soil moisture content (A), or soil matric potential (Ψ_soil; B) across 30 winegrape cultivars grown in a vineyard in southeastern Washington, USA. Data were pooled for two drydown cycles each in 2021 and 2022

Cultivar response to soil drying

All 30 cultivars responded similarly to the intensifying soil water deficit after irrigation was interrupted. From a θ_e perspective, they were all isohydric at θ_e > 0.33 and increasingly anisohydric below that threshold (Additional file 8). From a Ψ_soil perspective, they were all anisohydric at Ψ_soil > −0.3 MPa and increasingly isohydric below that threshold (Additional file 9). Cultivars with large canopies (e.g., Cabernet Sauvignon and Tempranillo) dried the soil more rapidly and would have perceived the same level of water stress earlier than those with small canopies (e.g., Durif and Melon) (Fig. 6, see also Fig. 2). Nevertheless, there were no significant differences in the response of Ψ_pd and Ψ_md to θ_e, neither between cultivars with more than fivefold differences in canopy size nor between cultivars that have been used as “model” genotypes for differences in their hydraulic response to soil moisture deficit (presumed “isohydric” Garnacha and “anisohydric” Sémillon) (Fig. 7 and Additional file 10). Therefore, we pooled the Garnacha and Sémillon data for pairwise comparisons against all other cultivars. The Ψ_pd and Ψ_md responses to θ_e were similar for all cultivars except that Auxerrois had a slightly lower Ψ_pd at the same (high) θ_e than the Garnacha—Sémillon pair, and the slope of the θ_e—Ψ_md curve was slightly gentler for Chardonnay, Chenin blanc, and Petit Verdot, and slightly steeper for Pinot noir (Additional files 11, 12, and 13). However, although these differences were statistically significant, their absolute magnitude was very small, which precludes clear conclusions regarding differences in isohydricity among cultivars.

Fig. 6.

Comparison of relative extractable soil water (θ_e) by day of year among winegrape cultivars with large canopy (Cabernet Sauvignon and Tempranillo) and small canopy (Durif and Melon) during the first drydown cycle of 2021 (A) and 2022 (B) in a vineyard in southeastern Washington, USA. Symbols show the individual values of three replicated rows of each cultivar. The fitted lines are polynomial regressions

Fig. 7.

Relationship of leaf water potential (Ψ_leaf) to relative extractable soil water (θ_e) between winegrape cultivars with presumed divergent response to soil water deficit (“isohydric” Garnacha and “anisohydric” Sémillon; A) and different canopy size (large Cabernet Sauvignon and Tempranillo, and small Durif and Melon; B) in a vineyard in southeastern Washington, USA. Data for each cultivar were pooled for two drydown cycles each in 2021 and 2022. n = number of observations

Relationship between midday and predawn leaf water potential

The overall shape of the Ψ_pd—Ψ_md response across 30 cultivars (Fig. 8A) resembled that of the Ψ_soil—Ψ_md response (Fig. 5B), albeit with lower variability. The relationship between Ψ_md and Ψ_pd differed slightly but significantly (at 95% CL by Proc nlin Gauss-Newton method, Eq. 3) between the two years, as Ψ_md dropped to lower values during the hot and dry 2021 than in 2022 (Fig. 8A). Applying the water stress phases I and II and boundaries Θ₁ and Θ₂ from Knipfer et al. [77], Ψ_md decreased rapidly as Ψ_pd decreased in the mild stress phase (I), while in the moderate stress phase (II) Ψ_md changed gradually and became increasingly unresponsive to a further decrease in Ψ_pd. The average Ψ_pd at Θ₁ of the pooled data from the 30 cultivars and the two growing seasons was − 0.68 MPa (−0.67 MPa in 2021 and − 0.70 MPa in 2022). We measured no Ψ_pd values below the Θ₂ of −1.33 MPa from Knipfer et al. [77] even though there was leaf wilting and abscission in some cultivars (Additional file 2). Extrapolating our data using the average of the ten lowest Ψ_md values each year gave a Θ₂ of −1.55 MPa in 2021 and − 1.39 MPa in 2022. The 2021 and 2022 curves differed somewhat, but 40% more data were collected from dry soil in 2021 than in 2022. The Ψ_pd—Ψ_md response was similar for Garnacha and Sémillon (Fig. 8B), as well as for cultivars with fivefold differences in canopy size (Fig. 8C and Additional file 14) and was not significantly different between the Garnacha—Sémillon pair and any other cultivar except Pinot gris, which had a slope in the “wrong” direction (Additional files 15 and 16). All Ψ_pd—Ψ_md response curves (besides Pinot gris) were nonlinear with a gradually decreasing slope, which has been interpreted as a change from anisohydric behavior under non-stress conditions to increasingly isohydric behavior as the water stress intensifies.

Fig. 8.

Relationship of midday leaf water potential (Ψ_md) to predawn leaf water potential (Ψ_pd) between years across 30 winegrape cultivars (A), cultivars with presumed divergent response to soil water deficit (“isohydric” Garnacha and “anisohydric” Sémillon; B), and cultivars with different canopy size (large Cabernet Sauvignon and Tempranillo, small Durif and Melon; C) in a vineyard in southeastern Washington, USA. Data for each cultivar were pooled for two drydown cycles each in 2021 and 2022. The diagonal line is the 1:1 line. Phases I, II, and III are mild, moderate, and severe drought stress phases, respectively, while boundaries Θ₁ and Θ₂ are Ψ_pd at the stomatal closure and turgor loss points, respectively, from Knipfer et al. [77]. Data on the 1:1 line were extrapolated using the average of the 10 lowest Ψ_md values each year. n = number of observations

Discussion

This study provided a detailed assessment of within-species variation in the response of winegrape leaf water relations to gradual soil drying. Our two-year common garden experiment conducted in an arid climate tested a panel of 30 V. vinifera cultivars originating from climatically diverse regions and differing substantially in phenology and canopy size, providing several novel insights. First, the responses of Ψ_leaf to changes in soil water content varying from above field capacity (i.e., θ_e >1) immediately following irrigation to nearly the permanent wilting point (i.e., θ_e ≥ 0) after a prolonged drydown period showed that all cultivars were both isohydric and anisohydric, depending on the definition of soil moisture and the degree of water deficit. When soil moisture was expressed volumetrically (as θ_v or θ_e), all 30 cultivars were isohydric (maintaining near-constant predawn and midday Ψ_leaf) under initial soil drying and became anisohydric (dropping Ψ_pd and Ψ_md) as the water deficit intensified. In contrast, when soil moisture was expressed as soil tension or when Ψ_pd was used as a proxy for Ψ_soil, all cultivars progressively changed from anisohydric at high Ψ_soil or high Ψ_pd to isohydric at low Ψ_soil or low Ψ_pd. These results support the notion that the shape of plant response functions to soil moisture depends on the choice of θ_v or Ψ_soil as the driving variable [43]. Consequently, different definitions of soil moisture may translate into different conclusions about the hydraulic strategy of plants. Second, although they varied nearly fivefold in canopy size and up to four weeks in phenology, leading to variation in the rate of soil water depletion, all cultivars had a common θ_e (soil water content normalized for FC and PWP) threshold of 0.33 at which they began to perceive water deficit, and Ψ_pd and Ψ_md declined in concert. Moreover, the shapes of the Ψ_soil—Ψ_pd, Ψ_soil—Ψ_md, and Ψ_pd—Ψ_md response curves were similar across all cultivars. These findings do not support the idea of within-species variation in hydraulic strategy and suggest that irrigation management may rely on common soil- or plant-based thresholds across cultivars. Third, variation in VPD introduced variation in Ψ_leaf, and the difference between Ψ_pd and Ψ_md widened with rising VPD, especially in wet soil, supporting the notion that Ψ_md is mostly driven by atmospheric conditions when the soil is wet [55].

Like the Ψ_leaf responses, the differences among cultivars in canopy size and phenology were consistent over two years, even though the canopies grew considerably larger during the cooler and wetter of the two growing seasons. Because most shoot growth occurs between budbreak and fruit set in this climate [80], the variation in canopy size among the 30 cultivars was mostly established before our soil drydown period started. Therefore, the variation in growth capacity or vigor under non-limiting water supply is likely to have a strong genetic basis. Nevertheless, this variability did not translate into disparate hydraulic strategies. Because Ψ_pd and Ψ_md of all cultivars followed similar trajectories during soil drying, stomatal closure did not prevent Ψ_leaf from declining in any cultivar. A similar shift in the changes of Ψ_leaf under declining soil moisture also occurred in tropical rainforest tree species and diverse wheat genotypes, indicating phylogenetic conservatism in the hydraulic strategy of plants [81, 82].

The relationship between Ψ_pd and Ψ_md is often used to categorize plants as either isohydric or anisohydric [18, 20, 39, 58, 62, 77, 83]. The overall pattern of our Ψ_pd—Ψ_md data is consistent with that presented by those authors. Knipfer et al. [77] hypothesized that stomatal closure (i.e., g_s → 0) caused a sudden change in the slope of the Ψ_pd—Ψ_md curve from steep to gentle or flat (their boundary Θ₁). Though the change in our study was more gradual, our data support the idea of a triphasic relationship between Ψ_pd and Ψ_md, with a similar threshold (Ψ_pd ≈ −0.7 MPa) at the boundary Θ₁ between mild and moderate stress phases as that reported previously [77]. Our estimated boundary Θ₂ (Ψ_pd = Ψ_md ≈ −1.5 MPa) between the moderate and severe stress phases was somewhat lower than the − 1.33 MPa shown in Knipfer et al. [77]. While we note their high standard error (0.44 MPa), our approach of using the 10 lowest Ψ_md values for extrapolation might have contributed to the lower Θ₂. Nevertheless, the difference in Θ₂ across our two growing seasons underscores the importance of collecting data over multiple years and the full range of soil moisture. Notably, many cultivars in our experiment experienced leaf wilting and some leaf abscission, indicating their older leaves reached the turgor loss point even before the Θ₂ boundary was reached in the somewhat younger leaves measured here. In addition, the pronounced variation in canopy size among our 30 cultivars did not alter the general Ψ_pd—Ψ_md response curve. Cultivars with large canopies (e.g., Cabernet Sauvignon, Tempranillo) dried the soil faster than those with small canopies (e.g., Durif, Melon) and therefore experienced water stress earlier in the season, but at the same θ_e threshold. Consequently, reported differences in physiological responses to water deficit among different studies cannot simply be explained by differences in canopy size.

However, because ΔΨ_leaf was similar in cultivars with large canopies (and hence high whole-canopy transpiration) and those with small canopies (and hence low whole-canopy transpiration), it follows that the large cultivars should have higher whole-plant hydraulic conductance than the small cultivars. Indeed, earlier field experiments in this region found that hydraulic conductance correlated with canopy size within a single winegrape cultivar, and that the trunk diameter of juice grapes scaled with the planting distance and canopy volume, indicating hydraulic adaptation of the vines’ perennial structure to canopy size [84, 85]. The present study extends those findings more generally to diverse cultivars. Moreover, ΔΨ_leaf widened at higher VPD. This effect was especially notable in the difference of ΔΨ_leaf between the first and second drydown cycles in 2021, which coincided with a marked drop in temperature and VPD at the beginning of the second cycle. Previously, grape leaves were reported to become more resistant to cavitation over the course of a growing season, and a relatively constant or widening ΔΨ_leaf was observed in both grapevines and eucalypts as the season progresses [58, 86, 87].

In earlier studies, categorization of winegrapes as either isohydric or anisohydric was primarily based on visual differences in the area (hydroscape) between a 1:1 line and the Ψ_pd—Ψ_md relationship, or nominal differences in the slopes of the Ψ_pd—Ψ_md curves [27, 34, 49, 62, 77, 88]. Differences were often obtained from comparisons of field-grown or potted plants, differing widely in size, age, and rootstock used, with a range of soil types or potting substrates, and data collected within a single growing season. Many studies did not report any measure of soil moisture and simply assumed that Ψ_soil = Ψ_pd (but see ref [51]). Moreover, the health status (e.g., systemic infection with viruses, bacteria, or fungal trunk-disease pathogens) of the experimental plants was generally unknown. The reported differences and conclusions tended to be inconsistent among studies. Here, we used own-rooted plants, excluding rootstock as a source of variation, collected measurements over two growing seasons, used annual disease monitoring and rogueing to maintain plants clean of systemic pathogens, paired all physiological measurements with soil moisture measurements, and applied statistical models to determine the θ_e—Ψ_leaf, Ψ_soil—Ψ_leaf, and Ψ_pd—Ψ_md relationships as well as their slopes, asymptotes, and potential shifts in response curves between cultivars. This analysis revealed that the physiological responses to soil drying were somewhat influenced by the growing season, but not by the cultivar. However, the large number of measurements required to compare 30 cultivars limited the number of plants that could be sampled per cultivar (n = 3), and hence the statistical power to detect small cultivar differences. Though our study might thus have missed some genotypic variation, this effect would likely be small compared with other sources of variation, such as environmental influences and methodological choices.

The present results have several practical implications: (1) Contrary to initial expectations, irrigation scheduling in winegrapes may be guided by common soil-based or plant-based thresholds across cultivars. (2) The time to reach these thresholds depends on differences in canopy size among cultivars (faster for large canopies) and on weather patterns (faster in hot and dry weather) and soil properties (faster in light-textured soils). (3) Different definitions of soil moisture or proxies for plant water status can lead to inconsistent interpretations of the same cultivar’s drought response, highlighting the risks of oversimplified classifications. (4) As the hydraulic strategy and regulation of leaf water status seem to be broadly conserved across cultivars, efforts to improve drought resilience should focus on other traits, such as canopy size or rooting depth, rather than assuming that cultivar identity dictates hydraulic strategy.

Conclusion

Our common garden experiment, conducted using controlled irrigation in an arid climate with 30 field-grown, own-rooted winegrape cultivars from climatically distinct regions, provides several key insights: (1) Different cultivars respond similarly to soil water deficit despite large inherent variation in phenology and canopy size. Rather than supporting the notion of distinct categories or a continuum of hydraulic responses among cultivars, our data suggests that the general physiological response to drought stress is largely conserved across genotypes within V. vinifera despite more than ten millennia of selection and breeding in diverse climates. (2) The hydraulic strategy of V. vinifera is dependent on how soil moisture is expressed (or measured) and on the degree of water deficit. When volumetric soil water content is used as the driving variable, all cultivars are isohydric under well-watered conditions and become anisohydric as the water deficit exceeds a common threshold that depends on soil type. In contrast, when soil water tension or its frequent proxy, Ψ_pd, is used, all cultivars are anisohydric under low water tension and gradually become isohydric as the tension increases. (3) While plant vigor is not a good predictor of a cultivar’s physiological response to drought stress, vigorous cultivars dry the soil more quickly and thus experience water stress sooner than less vigorous cultivars.

Abbreviations

ΔΨ_leaf

Difference between predawn and midday leaf water potential

Θ₁

Boundary between mild and moderate drought stress

Θ₂

Boundary between moderate and severe drought stress

θ_e

Relative extractable soil water

θ_fc

Volumetric soil moisture at field capacity

θ_pwp

Volumetric soil moisture at permanent wilting point

θ_v

Volumetric soil moisture

Ψ_asy

Maximum or minimum threshold (asymptote) of Ψ_pd or Ψ_md

Ψ_leaf

Leaf water potential

Ψ_md

Midday leaf water potential

Ψ_pd

Predawn leaf water potential

Ψ_soil

Soil matric potential

cont

Continuation

cv

Cultivar

DOY

Day of year

Eqn

Equation

FC

Field capacity

GDD

Growing degree days

g_s

Stomatal conductance

h

Hour

IPCC

Intergovernmental Panel on Climate Change

PWP

Permanent wilting point

r_p

Pearson correlation coefficient

r_s

Spearman correlation coefficient

T_max

Daily maximum temperature

T_md

Temperature at the time of Ψ_md measurement

T_min

Daily minimum temperature

T_pd

Temperature at the time of Ψ_pd measurement

T_mean

Daily average temperature

VPD

Vapor pressure deficit

VPD_md

VPD at the time of Ψ_md measurement

VPD_mean

Daily average VPD

VPD_pd

VPD at the time of Ψ_pd measurement

Authors’ contributions

COO designed and executed the study, collected the data, conducted the statistical analysis, and wrote the first draft. MK designed the experiment, obtained the funding, contributed to data interpretation and writing, and supervised the study. The final draft was read and approved by both authors.

Funding

This study was funded by the National Institute of Food and Agriculture of the U.S. Department of Agriculture (Hatch project 7003737), the Washington State Department of Agriculture Specialty Crop Block Grant Program (grant number K2862), and the Washington State Grape and Wine Research Program.

Data availability

The datasets used and/or analysed in this study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.