Abstract
Background and goals Hot, dry conditions can exacerbate late-season berry dehydration, reducing yield and wine quality. Late-season dehydration occurs as mesocarp cells die, releasing water that is lost to berry transpiration or backflow to the canopy. We tested whether short pulses of increased irrigation could reduce late-season berry dehydration by interrupting stress-induced signals for cell death.
Methods and key findings Conventional irrigation was compared with an early- and a late-pulse treatment, where irrigation was increased by ~40% in the 2 wk immediately before or after the expected onset of cell death (~90 days after anthesis). We imposed each treatment on five vines of Cabernet Sauvignon growing in an experimental vineyard in Davis, California in an atypically hot (2022) and cool (2023) growing season. We monitored vine water potentials, berry cell death and shrivel index, and concentrations of putative signals for cell death (i.e., hydrogen peroxide [H2O2]) and markers of cellular damage. The late-pulse treatment significantly reduced the rate of cell death and the magnitude of berry shrivel at harvest, but only in the hot season (p < 0.05). Conversely, the early-pulse treatment did not affect the rate or timing of onset for cell death or shrivel (p > 0.05). H2O2 levels increased with cell death but were not affected by irrigation, indicating that other mechanisms produced the treatment effects on cell death.
Conclusions and significance These findings suggest the onset of cell death is robust to irrigation, but a short pulse of supplemental irrigation soon after onset can slow the rate of cell death and mitigate late-season berry dehydration under hot growing conditions.
Introduction
In many winegrape cultivars, berries are susceptible to dehydration late in the growing season, causing them to shrink as they approach phenolic ripeness (Fuentes et al. 2010). Late-season dehydration typically reduces yield and income per hectare by 10 to 30%, and severe cases can also affect wine and berry quality (Keller 2010, Bonada et al. 2013a, Šuklje et al. 2016, Chou et al. 2018). The processes that cause late-season dehydration are part of normal ripening but are exacerbated by stress, including drought and heat (Coombe and McCarthy 2000, Bonada et al. 2013a, Xiao et al. 2018a). California wine regions are expected to become hotter and drier in the near future, with the number of hot (>35°C) days predicted to increase by up to 3 wk by 2040 (Livneh et al. 2015, Rasmussen et al. 2016). Thus, interventions that reduce late-season dehydration have the strong potential to reduce the effects of climate change on growers.
Several processes contribute to late-season berry dehydration, which typically begins ~90 to 100 days after anthesis (DAA) (Tilbrook and Tyerman 2008, 2009) (Figure 1A). Berries block the sugar transport tissue (phloem), which is their main water source after veraison (Figure 1B and 1C) (Keller et al. 2015, Bondada et al. 2017), and initiate cell death in the mesocarp, which destroys cell membranes and releases cellular contents (Krasnow et al. 2008, Tilbrook and Tyerman 2008) (Figure 1B and 1C). Water released from dead cells is lost to berry transpiration or backflow, where the water is drawn to the canopy through the xylem by the vine water potential gradient, causing berries to shrink (Choat et al. 2009, Tilbrook and Tyerman 2009) (Figure 1A and 1C). Several lines of evidence point to cell death as a cause of late-season dehydration. The onset of cell death and dehydration closely coincide (Figure 1A) (Tilbrook and Tyerman 2008, Bonada et al. 2013b). Cell death and shrinkage are correlated across cultivars (Fuentes et al. 2010) and exhibit coupled responses to experimental treatments. Finally, both variables increase in response to heat and drought treatments (Bonada et al. 2013b) and decrease following exogenous antioxidant applications (Shekharappa 2022). Moderate cell death and shrinkage are a normal and beneficial part of ripening. Cell death helps to release juice during crushing and to produce new flavor compounds by combining reactants that were previously separated by cell membranes, while shrinkage concentrates sugar and flavor compounds (Keller 2010, Bonada et al. 2013a). Indeed, in a survey across grape varieties, all 16 winegrape cultivars that were tested exhibited some cell death (ranging from 5 to 60% of mesocarp cross-sectional area [mean = 35%]) as well as shrinkage at harvest (Fuentes et al. 2010), though some cultivars were able to limit and decouple shrinkage from cell death by blocking the xylem and impeding backflow (Tilbrook and Tyerman 2009, Knipfer et al. 2015). Severe cell death (>60%) and shrinkage can cause significant declines in yield and are associated with declines in quality, including more overripe flavors in berries and wine (Bonada et al. 2013a, 2013b, Chou et al. 2018).
Examples demonstrating the course of berry cell death over ripening and relationships with berry shrinkage. A) An example trajectory for berry cell death and mass (data replotted from Tilbrook and Tyerman 2008). B and C) Berry hydraulics and anatomy prior to and after the onset of rapid cell death (i.e., at points X and Y in panel A). Winegrape berries undergo low levels of cell death in the mesocarp after veraison, which rapidly accelerates ~90 days after anthesis (DAA) (A, blue points). Before this point, berries grow when water inputs through the phloem (B, 1) exceed water losses to berry transpiration and backflow to the canopy (B, 3), allowing water to accumulate in the intact mesocarp cells (B, 2). After this point, cell death releases the water stored in the mesocarp cells (C, 2), which is readily lost to berry transpiration and xylem backflow (C, 3) without the resistance from intact cell walls. Phloem inputs are also reduced or eliminated (C, 1). These processes cause the berries to dehydrate and shrink (A, black points). Cell death can be detected with fluorescein diacetate, which stains living cells green (B and C) (stained images are from this study).
Both heat and water stress exacerbate cell death and shrinkage (Bonada et al. 2013a, 2013b). Heat stress typically accelerates development and initiates cell death earlier in the season (Bonada et al. 2013b, but see Xiao et al. 2018a), while water stress typically increases the rate of cell death (Bonada et al. 2013b, Xiao et al. 2018a, Shekharappa 2022). Both stresses increase evaporative demand and backflow (Bonada et al. 2013b, Xiao et al. 2018a). The trigger for cell death is hypothesized to be hypoxia, since berry respiration strongly depletes oxygen inside the mesocarp (Xiao et al. 2018a, 2018b). Water stress is hypothesized to exacerbate hypoxia and cell death by increasing the production of reactive oxygen species (ROS) in the mesocarp (Xiao et al. 2018a). ROS, especially hydrogen peroxide (H2O2), typically accumulate in the mesocarp at veraison and serve as signals for ripening, while excessive ROS are known to damage DNA, proteins, and cell membranes and trigger programmed cell death in other tissues (Pilati et al. 2007, 2014, Carvalho et al. 2015). To counteract ROS damage, water-stressed tissues upregulate the antioxidant system, which can increase respiration rates and oxygen consumption, potentially exacerbating hypoxia (Carvalho et al. 2015). Consistent with this hypothesis, greater cell death was associated with higher respiration rates and more severe hypoxia in water-stressed grape berries (Xiao et al. 2018b). Water stress also downregulated the expression of VvBAP1 in the mesocarp, a gene shown in other tissues to upregulate antioxidant activity and reduce H2O2 concentrations and H2O2-induced cell death (Cao et al. 2019). Further, applying the amino acid gamma-aminobutyric acid (GABA), which increases antioxidant activity, to water-stressed grape berries reduced berry H2O2 concentrations and cell death (Shekharappa 2022). Altogether, these findings suggest that interventions that reduce ROS-induced cell death could help mitigate late-season dehydration under future, drier conditions.
Increasing irrigation over the entire period from veraison to harvest reduces berry shrinkage and increases yield (Mendez et al. 2011, Rogiers and Holzapfel 2015, Keller et al. 2016). However, the connection to cell death suggests that shorter pulses of irrigation timed to coordinate with key developmental events could mitigate shrinkage with less water. Without the resistance from intact cell membranes, only a small water potential gradient is needed to induce backflow, and preventing these gradients requires extensive irrigation (Tilbrook and Tyerman 2009, Williams 2010, Rogiers and Holzapfel 2015). In this study, we hypothesized that applying short pulses of additional irrigation near the typical onset of cell death would alleviate water stress and reduce ROS production and the induction of cell death. Reducing cell death, in turn, would limit backflow and make shrinkage less sensitive to canopy water status. This strategy could require less irrigation to mitigate late-season dehydration than maintaining high canopy water potentials from veraison to harvest.
Here, we tested whether irrigation pulses delay the onset or reduce the rate of berry cell death and shrinkage by reducing ROS activity. We compared an “early” irrigation pulse (applied over the 2 wk before the expected onset of cell death [~90 DAA]) and a “late” pulse (applied over the 2 wk after the expected onset) to conventional irrigation. The irrigation treatments were conducted on field-grown Cabernet Sauvignon vines in a hot (Winkler V) growing region over two seasons with contrasting climate conditions (Jones et al. 2010). Cabernet Sauvignon is the second most-produced winegrape cultivar in California and requires long hang times to reach phenolic ripeness, which increases the risk of late-season dehydration (CDFA 2023). We compared trajectories for vine water status, berry cell death and shrivel, and ROS activity, which was measured from berry concentrations of H2O2 and malondialdehyde (MDA), a lipid peroxidation product that serves as an indicator of ROS damage to plasma membranes (Pilati et al. 2007, Morales and Munné-Bosch 2019). We hypothesized that both treatments would reduce ROS activity and damage and the rate of cell death and shrinkage, and that the early treatment would also delay their onset. Overall, we expected this work to test a less water-intensive strategy to reduce late-season dehydration and provide insight into the mechanisms driving berry cell death.
Materials and Methods
Plant material and irrigation treatments
The study used mature (10-yr-old) Cabernet Sauvignon (clone #5) vines grafted to 101-14 rootstock growing in an experimental vineyard on the University of California, Davis campus (38°31′N; −121°45′W). Vines were cane-pruned and trained to a vertical shoot-positioned trellising system with 2.13 m × 3.35 m vine × row spacing and NE-SW row orientation. Canopy and pest management followed standard commercial practices. Soils are deep (>3-m rooting depth), well-drained Yolo loam.
The study was conducted in 2022 and 2023. In 2022, we assigned 15 vines from one row to a control, early, or late irrigation treatment (n = 5 vines per treatment) by randomly selecting five vine positions for each treatment. The vines next to each experimental vine received the same irrigation and vines from different treatments were separated by at least two vines (map provided in Supplemental Material).
Early treatment vines received a pulse of additional irrigation in the 2 wk prior to the expected onset of cell death (i.e., 90 DAA) and the late treatment vines received this pulse at expected onset and in the 2 wk after. All vines received the same irrigation outside of their treatment periods (Figure 2). Temperature differences between years produced differences in the treatment dates (Table 1) and the amount of irrigation applied (Supplemental Table 1) since 2022 was an unusually hot year occurring at the end of a severe multiyear drought, while 2023 was exceptionally cool and wet (Supplemental Figure 1). Prior to the treatments, the irrigation regime was that all vines were irrigated weekly at 25% replacement of evapotranspiration (ETc) from mid-April to anthesis and at 40% ETc after anthesis, except in anticipation of forecasted heatwaves (i.e., 3 or more consecutive days with maximum daily temperatures >38°C), when irrigation was increased to 50% ETc. ETc was calculated following the methods from Williams (2014), as the product of reference evapotranspiration measured by the nearby California Irrigation Management Information System (CIMIS) weather station in Davis, CA (https://cimis.water.ca.gov/), and from published crop coefficients for vineyards with similar trellising systems and vine by row spacing.
Weekly irrigation amounts applied to the control treatment over 2022 (A) and 2023 (B) and daily irrigation amounts applied to each treatment over the course of the experimental periods in 2022 (C) and 2023 (D). Only the control treatment is shown in (A) and (B) for visual clarity, and differences in irrigation amounts between treatments are shown in (C) and (D). Vertical dashed lines in (A) and (B) indicate the start of the early pulse treatment (cyan) and the end of the late pulse treatment (blue); “a”, “v”, and “h” represent anthesis, veraison, and harvest, respectively. Irrigation was applied three times per week during and after the treatments and weekly the rest of the year. As a result, the weekly irrigation amount shown for this period in (A) and (B) is divided among three irrigation events per week in (C) and (D). DAA, days after anthesis.
Key dates, annual growing degree days (GDD), and mean vapor pressure deficit (VPD) during the irrigation treatments for each year of the experiment. GDD were calculated as the cumulative difference between daily mean temperature and a base temperature of 10°C from 1 April to 31 Oct (Jones et al. 2010). The 10-yr average ± standard error for GDD at this site is 2308°C ± 33. Climate data were obtained from the California Irrigation Management Information System (CIMIS) weather station in Davis, CA (https://cimis.water.ca.gov/).
From the start of the treatments until harvest, irrigation frequency was increased to 3 times/wk to facilitate a response in vine water status during the irrigation pulses. The irrigation amount during the treatments was calculated from the ETc in the 2 wk before the early treatment rather than updated each week, to avoid large inconsistencies in irrigation between the early and late treatments. The percentage of ETc applied was chosen to maintain midday stem water potentials (Ψmd) within the range of moderate-to-severe water stress (−1.1 to −1.4 MPa), to match conventional practices for red wine varieties (Rienth and Scholasch 2019). The vines began the early treatment near the threshold for severe stress (−1.36 MPa) in 2022 and for mild stress (−0.96 MPa) in 2023, so we irrigated at a higher percentage of ETc in 2022. Altogether, in 2022, we irrigated the control vines at 55% ETc for the first week of the early treatment and at 40% thereafter, as Ψmd shifted away from the threshold for severe stress, to avoid overcompensating (Supplemental Table 1, Figures 2 and 3). We irrigated the early and late treatment vines at 66% ETc during their treatment periods. In 2023, we irrigated the control vines at 20% ETc and the early and late treatment vines at 40% ETc during their treatment periods (Figure 2). In the 4 wk encompassing both treatments, the early and late treatment vines received 20% and 27% more irrigation, respectively, than the control vines in 2022, while both treatments received 37% more irrigation than the control vines in 2023 (Supplemental Table 1 and Figure 2).
Vine water status
Vine water status was determined from weekly measurements of gas exchange, predawn water potential (Ψpd), and Ψmd. Measurements were made 1 and 2 days after irrigation applications in 2022 and 2023, respectively. At each time point, sun-exposed, recently fully expanded leaves were sampled from positions located 8 to 12 nodes below the shoot tip. Two leaves were sampled per vine in 2022 and one leaf was sampled per vine in 2023, due to unexpected equipment limitations in 2023 (n = 5 to 10 leaves per treatment). Predawn leaves were sampled from 0400 to 0500 hr. Excised leaves were immediately placed into humidified foil-laminate bags which were then placed into a humidified plastic bag inside an ice-filled cooler, with a cardboard barrier placed between the bag and ice to prevent freezing damage. Leaves were then transported to the lab, refrigerated, and measured for water potential with a pressure chamber (Model 1505D, PMS Instrument Company) within 1 hr of sampling. The cut ends of the petioles were observed through a stereo microscope (Omano OM2300S-GX4) to ensure the measurements captured water loss from the xylem. Previous work found that these storage conditions produce only small changes in water potential (i.e., a decline of 0.05 MPa/hr), compared to immediate measurements (Feng et al. 2025). Midday measurements were made from 1200 to 1330 hr. Leaves were first measured for gas exchange with a LICOR 6800 photosynthesis system using a fan speed of 10,000 rpm, CO2 concentration of 400 μmol/mol, light intensity of 1900 μmol/(m2sec), and ambient air temperature and vapor pressure deficit (VPD). The leaves were then wrapped in humidified, foil-laminate bags and allowed to equilibrate on the vine for 20 min, then excised to measure stem water potential (Ψstem), following the procedure for predawn measurements.
Berry sampling
Four sunlit clusters per vine on the evening sun-exposed (west) side of the canopy were marked for repeated sampling. We repeatedly sampled from the same clusters and selected clusters from similar microenvironments to better distinguish trends over time from error due to intravine variability in microenvironment and cluster development. These clusters should also be the warmest and thus, the most prone to late-season dehydration. One berry/cluster was sampled for fluorescein diacetate (FDA) staining once or twice per week (n = 20 berries per treatment). In 2022, one berry/cluster was also sampled twice per week for ROS analyses (n = 20). Size, color, and softness were used as criteria to select berries that were relatively advanced in maturity, to reduce random variability due to berry age. Berries were excised with the pedicel attached and transported on ice to the laboratory. Sampling began ~2 wk after veraison and continued until harvest, when we measured berry chemistry and, in 2023, berry weight and yield.
To decide when to harvest, a handheld refractometer was used to monitor total soluble solids (TSS) in the field. In 2022, we harvested when TSS reached 27 Brix, following common industry practices for Cabernet Sauvignon from warm regions of California. However, in 2023, cool conditions made ripening especially slow, with one of the latest harvests on record in California. We allowed berries to ripen until rain was forecast for early October, then harvested at 24 Brix to avoid confounding effects from a second pulse of water. We sampled 100 berries per vine from randomly selected clusters on both sides of the canopy. We alternated sampling positions between clusters to include berries from the fronts, backs, tops, middles, and tips. The berries were then crushed and the juice measured for TSS, titratable acidity (TA), and pH. In 2023, we also measured 100-berry weights prior to crushing and harvested and weighed all clusters to measure total yield per vine.
Cell vitality staining
The FDA staining technique (Krasnow et al. 2008) was used to quantify berry cell vitality. We first made a transverse cut with a razor blade to remove 1 to 2 mm from the pedicel end of the berry to locate the seeds. The berry was then cut in half longitudinally between the seeds, keeping the blade as close to the center of the berry as possible. One half of each berry was then placed in a humidified petri dish and refrigerated; the other half was crushed and the juices homogenized with that of the other berries from the same vine. Juice samples were measured for osmotic potential with a VAPRO 5600 vapor pressure osmometer. The refrigerated berry halves were stained with a 9.6-μM FDA solution that was prepared by adding 2 μL of a 4.8-mM FDA stock solution (in acetone) to 1 mL of a sucrose solution with the same osmotic potential as the juice samples, to avoid cell death from osmotic shock. Berry and solution osmotic potentials should be within 0.5 MPa to avoid osmotic shock; it was previously found that variability in osmotic potential between berries from the same vine was smaller than this range and homogenization did not affect vitality staining (Krasnow et al. 2008). The berry halves were blotted with a Kimwipe (Kimtech) that was dipped in the staining solution to remove cellular debris, then the halves were sectioned again to obtain a 2- to 3-mm thick longitudinal cross section without seeds. The staining solution was then pipetted onto the thin sections so that the entire surface was submerged and allowed to take up the solution for 30 min before imaging.
Stained berry sections were visualized with fluorescent microscopy using a Leica DM4000B microscope equipped with a Leica DFC 320 camera and Leica GFP filter cube. The camera was operated with Leica Application Suite Software (ver. 7.4). Camera settings were held constant for all images. Each berry was photographed at 1.4× magnification in four images that were later stitched together with Adobe Photoshop (ver. 23.5.3).
Image analysis
Percent cell vitality and berry perimeter (P) and area (A) were measured with ImageJ software (Schneider et al. 2012). The images were first renamed to blind the measurements to treatment. The berry images were then traced by hand to measure P and A. Percent cell vitality was measured by converting images to grayscale with ImageJ and manually adjusting the binary segmentation threshold separating living (white) from dead (black) areas until the shape of the dead areas in the segmented images matched that of the original images. We manually traced seeds and empty spaces created by removing the pedicels to exclude these areas from the percent cell vitality calculations, though berries were generally sectioned to avoid seeds. P and A were used to calculate the berry shrivel index (ShI) from Fuentes et al. (2010):
where R is the A/P ratio for an individual berry and Rmax and Rmin are the maximum and minimum ratios, respectively, across the entire data set. ShI ranges from 0 to 1, with 0 representing a completely shriveled berry, and 1 a fully turgid berry. We also tested defining Rmax and Rmin separately for each irrigation treatment and did not find an impact on our results.
ROS and lipid peroxidation
Berries sampled for ROS analysis were frozen in liquid nitrogen immediately after they were brought to the lab and stored in a −80°C freezer. The mesocarp tissue was isolated by peeling the frozen berries with a scalpel, then gently breaking and deseeding the frozen pulp. The four berries sampled per vine were then homogenized and cryogenically ground to a fine powder (n = 5 vines per treatment). H2O2 concentrations were measured with an Amplex®Red Hydrogen Peroxide/Peroxidase Assay kit, following Pilati et al. (2007). This fluorometric assay quantifies the absorbance of 10-acetyl-3,7-dihydroxyphenoxazine, which fluoresces red when oxidized in the presence of H2O2 and peroxidase. For each sample, 0.2 g of powder was dissolved in 0.5 mL of 50-mM phosphate buffer (pH 7.4) and kept on ice for 5 min. The mixture was then centrifuged at 20,000 g for 15 min, after which 0.4 mL of supernatant was extracted, mixed with 0.4 mL of a 2:1 (v/v) chloroform:methanol mixture, and centrifuged at 12,000 g for 5 min. 50 μL of supernatant was then combined with 50 μL of Amplex®Red working solution (prepared according to manufacturer instructions) in a black-walled 96-well plate with a transparent flat bottom. Absorbance was read using an Epoch-2 microplate reader at 560 nm and converted to H2O2 concentrations (nmol/g fresh weight [FW]) by generating a standard curve following manufacturer instructions.
MDA was measured following previous methods (Sun et al. 2011). Frozen powder (0.25 g) was homogenized for 10 min in 0.1% (w/v) trichloroacetic acid and centrifuged at 10,000 g for 5 min. 1 mL of the supernatant was then combined with 4-mL 20% (w/v) trichloroacetic acid containing 0.5% (w/v) thiobarbituric acid, then the mixture was heated for 15 min at 95°C then immediately cooled in an ice bath for 5 min. After centrifugation at 10,000 g for 10 min, 100 μL of supernatant was plated into a black-walled, clear-bottomed 96-well plate and absorbance was read at 532 and 600 nm. MDA concentration (nmol/g FW) was then calculated as (A532 − A600)/ε × 1010, where ε is the molar extinction coefficient (155,000 1/(M cm)).
Statistical analysis
We first determined the onset and rate of berry cell death, shrivel, ROS accumulation, and lipid peroxidation for each treatment by fitting piecewise linear regressions between DAA and percent cell vitality, ShI, and H2O2 and MDA concentrations, using the ‘segmented’ package in R (ver. 4.3.1) (Bonada et al. 2013a). The onset is the date these variables transitioned from relatively constant to rapidly changing, defined as the fitted breakpoints of the piecewise regressions, and the rate is the slope of these variables after onset. The regressions were fit to the mean treatment values on each sampling date, in accordance with previous studies evaluating the effects of watering treatments on cell vitality and berry shrivel (Bonada et al. 2013a, Xiao et al. 2018a, Shekharappa 2022). Each year was analyzed separately. We tested for significant differences between treatments by comparing 95% confidence intervals for the fitted onset dates and rates. Notably, cell vitality and ShI sharply increased and H2O2 concentrations decreased in early treatment berries collected on 22 Aug 2022 (99 DAA), possibly due to sampling less mature berries. We repeated the analyses without this sampling date and confirmed that the results were the same.
Second, we used analysis of variance (ANOVA) to test each year for treatment differences in percent cell vitality, ShI, and H2O2 and MDA concentrations on the last sampling date (i.e., harvest), when these variables would be expected to have the greatest effect on wine quality. ANOVA was also used to test for treatment differences in berry chemistry (i.e., TSS, TA, and pH), yield, and 100-berry masses at harvest.
Third, we used standardized major axis regressions to test for treatment differences in correlations between ShI and percent cell vitality with the ‘SMATR’ package in R, to determine whether cell death is associated with shrivel for this cultivar.
Fourth, ANOVA was used to test for treatment differences in water potentials and gas exchange. Each date was tested separately since we did not expect our pulsed treatments to produce linear, monotonic trends in vine water status.
Lastly, we tested whether the vines in each treatment responded to the irrigation pulses. ANOVA was used to test for differences in water potential and gas exchange between dates within each treatment, and Tukey’s honest significant difference (HSD) test was used to compare each pairwise combination of dates.
Results
Vine water status
The irrigation pulses significantly improved water status for the vines in each treatment in the hot 2022 growing season, but contrary to expectation, did not produce significant differences in vine water potential or gas exchange between treatments in either year (Table 2 and Figure 3). In 2022, mean daily Ψstem ranged from −0.34 to −0.61 MPa at predawn (Ψpd) and from −1.07 to −1.45 MPa at midday (Ψmd), which is considered moderate-to-high water stress in commercial vineyards (Figure 3A). In the early treatment vines, Ψmd became significantly higher during and in the week after the early irrigation pulse than at the beginning of the experiment (Tukey’s HSD test, p < 0.05) (Table 2). In the late treatment vines, Ψmd became significantly higher in the week after the late irrigation pulse than at the beginning of the experiment (p < 0.05) (Table 2). Ψmd did not significantly change for the control vines (p > 0.05) (Table 2). Conversely, Ψpd significantly increased after the early irrigation pulse for both the early and control treatment, suggesting this increase was not due to the irrigation pulse, while Ψpd did not significantly change in the late treatment vines (Supplemental Tables 2 and 3). However, treatment differences in Ψmd and Ψpd were not significant on any measurement date (ANOVA, p > 0.05) (Figure 3A). Water stress was lower in 2023, with water potentials in the range of mild-to-moderate stress (Ψpd = −0.12 to −0.40 MPa and Ψmd = −0.74 to −1.05 MPa). Ψmd and Ψpd significantly increased at the beginning of the late irrigation pulse for all treatments and were not significantly different between treatments on any measurement dates (Table 2, Supplemental Table 3, Figure 3B).
Midday stem water potentials (Ψmd) from the 2022 and 2023 experimental periods. Values are means ± standard errors (2022: n = 10; 2023: n = 5). For each treatment, letters indicate dates with significantly different water potentials, based on Tukey’s honest significant difference test. Note that each year and treatment were analyzed separately, so letters do not indicate differences between these variables. DAA, days after anthesis.
Stem water potential (Ψstem) in the three irrigation treatments in 2022 (A) and 2023 (B). Point colors indicate irrigation treatments; each point is the mean of n = 5 to 10 leaves. Error bars are standard errors. The pairs of vertical dotted lines in each panel show the dates for the irrigation pulses. DAA, days after anthesis.
Gas exchange showed similar responses to the irrigation treatments. In 2022, stomatal conductance (gs) and photosynthesis (An) did not significantly change over time in the control or early treatment, while An became significantly higher in the late treatment in the week after the late irrigation pulse than at the beginning of the experiment, in accordance with Ψmd (Supplemental Table 4, Supplemental Figure 2A and 2C). However, gs and An were only significantly different between treatments at the beginning of the experiment, when they were higher in the late treatment which had not yet received additional irrigation (Supplemental Figure 2A and 2C). In 2023, gs significantly declined over the first few weeks of the experiment and increased at harvest for all treatments, while only the late treatment vines showed a significant increase in gs and An in the week after the late irrigation pulse (Supplemental Table 4, Supplemental Figure 2B and 2D). An did not significantly change for the other treatments (data not shown). Gas exchange was not significantly different between treatments on any date (p > 0.05) (Supplemental Figure 2B and 2D).
Berry cell death
Cell death followed the typical pattern for grape berries, starting slowly after veraison, then rapidly accelerating ~90 DAA (Figure 4). The late irrigation pulse which was applied in the 2 wk at and after the expected onset of rapid cell death reduced the rate of cell death, but only in the hot 2022 growing season (Table 3 and Figure 4A). In 2022, the slope of the relationship between percent cell vitality and DAA was significantly less negative in the late irrigation treatment (mean ± standard error = −0.90 ± 0.11 %/days) than in the control treatment (−2.06 ± 0.22 %/days) (Table 3 and Figure 4A). However, in the cool 2023 growing season, the late and control treatment slopes were not significantly different (mean ± standard error = −1.03 ± 0.15 %/days [late treatment] and −0.94 ± 0.11 %/days [control treatment]) (Table 3 and Figure 4A). The rate of cell death was also significantly faster in 2022 than in 2023, but only for the control treatment vines (Table 3). Cell death rates were not significantly different between years for the early or late treatments.
Berry cell vitality in the three irrigation treatments in 2022 (A) and 2023 (B). Cell vitality is the percentage of berry cross-sectional area comprised of living cells, as indicated by fluorescein diacetate staining. Point colors indicate irrigation treatments; each point is the mean of n = 5 to 19 berries. Error bars are standard errors. Colored solid lines indicate the best-fit piecewise regression for each irrigation treatment. The pairs of vertical dotted lines in each panel show the dates for the irrigation pulses. DAA, days after anthesis.
Fitted parameters for piecewise regressions between berry cell vitality, berry shrivel index (ShI), and hydrogen peroxide (H2O2) concentrations and days after anthesis. Onset is the breakpoint at which the slope of the relationship significantly changes, Slope 1 is the slope prior to the breakpoint, and Slope 2 is the slope after the breakpoint. Numbers in brackets are 95% confidence intervals. Bolded values with different lowercase letters are significantly different. N/A, H2O2 concentrations were not measured in 2023.
Contrary to expectation, the early irrigation pulse applied in the 2 wk before the expected onset of rapid cell death did not affect the onset or rate of rapid cell death (Table 3 and Figure 4). The expected onset date closely approximated the actual onset date. Onset occurred at 83 ± 4, 89 ± 8, and 92 ± 1 DAA in the late, early, and control treatments, respectively, in 2022, and at 89 ± 2, 87 ± 2, and 83 ± 3 DAA in the late, early, and control treatments, respectively, in 2023 (Table 3 and Figure 4). The slope of cell death in the early irrigation treatment was also not significantly different from the other treatments (mean ± standard error = −1.21 ± 0.53 %/days in 2022 and −1.29 ± 0.11 %/days in 2023) (Table 3 and Figure 4). Finally, there were no treatment differences in percent cell vitality at harvest in either year (ANOVA, p > 0.05).
Berry shrivel
The late irrigation pulse reduced berry shrivel at harvest, but only in the hot 2022 growing season (Figure 5). ShI was largely constant until ~90 DAA, when it began to decline, indicating shriveling. Unexpectedly, the onset dates and slopes for shriveling were not significantly different between treatments in either year (Table 3 and Figure 5). The onset dates closely corresponded to those for rapid cell death. In 2022, the onset was 89 ± 2, 86 ± 4, and 93 ± 2 DAA and the slope was −0.021 ± 0.005, −0.011 ± 0.008, and −0.017 ± 0.005 1/days in the control, early, and late treatments, respectively (Table 3). In 2023, the onset was 87 ± 1, 96 ± 3, and 97 ± 4 and the slope was −0.013 ± 0.002, −0.019 ± 0.006, and −0.013 ± 0.005 in the control, early, and late treatments, respectively (Table 3). Finally, ShI in 2022 was significantly higher at harvest in the late irrigation (0.41 ± 0.03) than in the control (0.30 ± 0.04) treatment (p < 0.05), though in 2023 these differences were not significant (p > 0.05) (Figure 5).
Trajectories of berry shrivel index (ShI) for the three irrigation treatments in 2022 (A) and 2023 (B). ShI is calculated from the ratio of berry perimeter-to-area, and smaller numbers indicate greater shriveling. Point colors indicate irrigation treatments; each point is the mean of n = 5 to 19 berries. Error bars are standard errors. The pairs of vertical dotted lines in each panel show the dates for the irrigation pulses and the asterisk indicates dates with significant differences between treatments (analysis of variance, p < 0.05). DAA, days after anthesis.
ROS and lipid peroxidation markers
H2O2 accumulation in the mesocarp began at a similar time as rapid cell death and shrivel but the onset and rate of H2O2 accumulation were not significantly different between irrigation treatments (Table 3 and Figure 6). H2O2 accumulation began at 85 ± 4, 94 ± 7, and 88 ± 1 DAA at a rate of 0.027 ± 0.009, 0.011 ± 0.02, and 0.029 ± 0.004 nmol/g/days) for the control, early, and late treatments, respectively.
Berry hydrogen peroxide (H2O2) concentrations in the hot 2022 growing season. H2O2 is a reactive oxygen species with an important role in plant cell death. Point colors indicate irrigation treatments; each point is the mean of four berries homogenized for each of n = 4 to 5 vines. Error bars are standard errors. The pairs of vertical dotted lines in each panel show the dates for the irrigation pulses. DAA, days after anthesis.
In contrast, MDA did not accumulate in the pulp; instead, concentrations fluctuated over the course of ripening (Figure 7). Piecewise regressions did not identify a statistically significant breakpoint in the relationship between MDA concentrations and DAA (p > 0.05). There were also no significant differences in H2O2 or MDA concentrations at harvest (ANOVA, p > 0.05).
Berry malodialdehyde (MDA) concentrations in the hot 2022 growing season. MDA is a by-product of lipid peroxidation and serves as an indicator of reactive oxygen species-induced plasma membrane damage. Point colors indicate irrigation treatments; each point is the mean of four berries homogenized for each of n = 4 to 5 vines. Error bars are standard errors. The pairs of vertical dotted lines in each panel show the dates for the irrigation pulses. FW, fresh weight; DAA, days after anthesis.
Berry chemistry and yield at harvest
The irrigation treatments also did not significantly affect berry chemistry or yield at harvest. TSS were lower on average in the late treatment, consistent with less berry shrivel, but these differences were not significant (i.e., TSS in the late treatment in 2022 and 2023 = 25.6 ± 0.6 Brix and 23.0 ± 0.6 Brix, respectively; TSS in the control treatment in 2022 and 2023 = 26.7 ± 0.5 Brix and 24.2 ± 0.2 Brix, respectively) (Table 4). There were also no significant differences in TA or pH (all p > 0.05) (Table 4). Yield and 100-berry weight were measured in 2023 and were higher on average in the late treatment, but these differences were not significant (p > 0.1) (i.e., 100-berry weight in the late and control treatments = 120.7 ± 3.0 g and 119.7 ± 2.2 g, respectively; yield in the late and control treatments = 11.3 ± 1.0 kg and 11.0 ± 0.2 kg, respectively) (Table 4).
Mean ± standard errors for total soluble solids (TSS), pH, titratable acidity (TA), 100-berry mass, and yield per vine at harvest for each treatment and growing season. N/A, yield and berry mass were not measured in 2022.
Relationships between berry shrivel and cell death
ShI was significantly correlated with percent cell vitality (r2 = 0.55, p < 0.001), indicating that cell death is an important driver of berry shrivel in Cabernet Sauvignon (Supplemental Figure 3). The slopes of this relationship were not significantly different between treatments, indicating that irrigation did not affect the sensitivity of shriveling to cell death (standardized major axis regression, p > 0.05).
Discussion
Altogether, we found that a pulse of additional irrigation applied immediately after the onset of cell death significantly reduced the rate of cell death and the extent of berry shrivel at harvest, but only in the hot 2022 growing season (Table 3, Figures 4 and 5). We increased irrigation by 20 to 27% in the early and late irrigation pulse treatments in 2022 and by 37% in 2023 (Supplemental Table 1, Figure 2). The irrigation pulses significantly increased Ψmd in both treatments and An in the late treatment in 2022, though not enough to produce significant differences in Ψmd or gas exchange among treatments (Table 2 and Figure 3). Rootstock and soil factors could have dampened the irrigation response since 101-14 has low hydraulic conductivity after veraison and soil compaction has been observed at this site (Alsina et al. 2011). Unexpectedly, the early pulse treatment did not delay the onset or reduce the rate of cell death or shrivel (Table 3, Figures 4 and 5). Irrigation pulses also did not affect cell death or shrivel in the cool 2023 growing season. Further, the onset of cell death and shrivel coincided with the accumulation of ROS, but treatment differences in H2O2 and MDA, an indicator of ROS-induced plasma membrane damage, did not explain the differences in cell death or shrivel (Table 3, Figures 6 and 7). The mechanisms producing differences in cell death and shrivel without treatment differences in Ψmd are unclear. However, cell death could have responded to water uptake instead of water potential per se, since the small but significant increases in Ψmd and An after the late irrigation pulse suggest that the vines increased their water uptake. While the results from this study are promising, translating these findings into commercial viticultural practices would require larger-scale agronomic experiments that test the effects of “late” irrigation pulses on yield and quality, while incorporating environmental variation both within and across sites.
Previous studies have shown that droughted vines exhibit faster rates but a similar onset for berry cell death than conventionally irrigated vines, while this study is the first to find that pulses of additional irrigation can slow cell death compared to conventional practices (Bonada et al. 2013a, Xiao et al. 2018a). However, the effects of temperature in our study diverged from previous work. Heat treatments significantly advanced the onset and increased the rate of cell death in some studies (Bonada et al. 2013a) but had no effect in others (Xiao et al. 2018a). In this study, cell death was significantly faster in 2022 than 2023 for the control treatment vines, while onset was similar (Table 3). For comparison, there were 2384°C growing degree days (GDD) in 2022 and 2108°C GDD in 2023; the 10-yr average ± standard error is 2308 ± 33°C. Further, previous studies also found that heat treatments made cell death rates more similar between irrigated and nonirrigated vines (Bonada et al. 2013a, Xiao et al. 2018a), while this study only found a significant difference between irrigation treatments in 2022 (Table 3). These discrepancies could reflect differences in water status. Vine water potentials were similar between heat and control treatments in the previous studies, while lower winter rainfall and higher evaporative demand made water potentials significantly more negative in 2022 than in 2023 in our study (across all treatments, mean Ψmd ± SE in 2022 and 2023 = −1.25 ± 0.01 and −0.90 ± 0.01, respectively). Thus, irrigation could have been more beneficial in 2022, when the berries were experiencing more stress.
We hypothesized that the irrigation pulses would alleviate vine water stress and reduce ROS production in the berries, which in turn would reduce respiratory demand and hypoxia-induced cell death (Xiao et al. 2018a, 2018b, Cao et al. 2019). Previous work showed that berry H2O2 concentrations and cell death were higher in water-stressed Syrah berries than in well-watered Syrah berries, and that applying GABA (an amino acid that upregulates the antioxidant system) to the berries reduced H2O2 concentrations and cell death (Shekharappa 2022). Water stress also downregulated the expression of the VvBAP1 gene in the mesocarp, which was shown to activate the antioxidant system and reduce H2O2 concentrations and cell death in other tissues (Cao et al. 2019). However, contrary to these findings, the irrigation treatments did not produce significant differences in mesocarp H2O2 or MDA in this study. This discrepancy could reflect methodological differences. The measurements in Shekharappa (2022) combined skin and mesocarp, and skins generally exhibit higher H2O2 concentrations and different H2O2 dynamics than the mesocarp (Pilati et al. 2014). H2O2 production could also be responsive to the water stress imposed in Shekharappa (2022) and not to the small differences in water status between well-watered vines imposed in this study. Both studies reported gs, which ranged from 0 to 50 mmol/(m2sec) in the water stress treatment and from 100 to 400 mmol/(m2sec) in the well-watered treatment over the 5-wk experimental period in Shekharappa (2022), while gs ranged from 80 to 320 mmol/(m2sec) in the control treatment and from 40 to 280 mmol/(m2sec) in the late treatment over the 2-wk late irrigation pulse in the present study (Supplemental Figure 2). Instead, the late irrigation treatment could have reduced cell death through ROS-independent mechanisms. The signaling molecules nitric oxide (NO) and Ca2+ and many phytohormones (including jasmonic acid, ethylene, and salicylic acid) respond to plant water status and can trigger cell death through both ROS-dependent and -independent pathways (Baek et al. 2004, Talukdar 2019, Devireddy et al. 2021, Lau et al. 2021). Indeed, jasmonic and salicylic acid peak in the mesocarp at the onset of berry shrinkage (Ribalta-Pizarro et al. 2021). Future work measuring the effects of irrigation pulses on berry hormone levels and expression of genes known to regulate antioxidant and hormone activity and programmed cell death would provide insight into the underlying mechanisms.
The mechanism that reduced cell death in the late irrigation treatment does not appear to depend on absolute vine or berry water status, since water potentials were not different between treatments. Instead, this mechanism could depend on vine water uptake since the late irrigation pulse produced a small but significant increase in vine water potentials, indicating that the vines became more hydrated. Increased water uptake could have slightly increased the phloem or nighttime xylem water transport to the berries or slightly reduced daytime xylem backflow (Coombe and McCarthy 2000, Keller et al. 2015). However, these effects would have been expected to increase berry hydration and osmotic potentials, and berry osmotic potentials were highly similar between treatments (i.e., mean osmotic potential from the beginning of late treatment onward = −4.73 ± 0.07 MPa for both the late and control treatments; data not shown). Additional water inputs could have increased berry transpiration and thereby cooled the fruit and reduced cell death, but this is unlikely since irrigation pulses have inconsistent effects on berry transpiration (Keller et al. 2015, 2016). Instead, irrigation could have influenced berry development through nonhydraulic mechanisms such as root-to-shoot hormonal signaling. Plants transport thousands of molecules from roots to shoots that could serve as signals of soil water availability (Shabala et al. 2016, Cai et al. 2023), and in grape, root-to-shoot hormone transport can mediate environmental impacts on canopy physiology through water transport. Warmer conditions increase cytokinin transport in the transpiration stream, which promotes vegetative growth (Nikolaou et al. 2003, Zhang et al. 2016). These findings suggest that it is plausible for water influx into the roots to alter berry cell death through root-to-shoot signaling, but to test this hypothesis, further work is needed to experimentally isolate the effects of xylem sap from irrigated and control vines on cell death.
While the findings from this study are based on a small number of individual vines and need to be confirmed with larger-scale agronomic experiments across a range of sites, climates, and management regimes before making recommendations to growers, they do suggest that short pulses of additional irrigation timed at the expected beginning of cell death could reduce cell death and late-season dehydration in hot years. GDD in 2022 were closer to the 10-yr average than in 2023, which suggests that the findings for 2022 could better represent typical responses at this site. The significant correlation between cell death and ShI shows that Cabernet Sauvignon lacks adaptations to prevent xylem backflow (Supplemental Figure 3), since cultivars that limit backflow maintain berry volume despite cell death (Tilbrook and Tyerman 2009, Knipfer et al. 2015). Thus, mitigating late-season dehydration will require interventions to reduce cell death. The onset of cell death was fairly consistent between years (mean = 88 DAA in 2022 and 83 DAA in 2023, across all treatments), making it feasible for growers to time a supplemental irrigation pulse at ~85 to 100 DAA. Targeting this narrow window could reduce the irrigation needed over the rest of the period from veraison to harvest, potentially providing a water-use efficient strategy to reduce the impacts of late-season dehydration on yield, income, and quality. To provide optimal water-use efficient irrigation guidelines, future work should evaluate how irrigation amount and frequency affect cell death and dehydration.
Conclusion
In summary, we found that the “late” irrigation pulse in the 2 wk after the onset of berry cell death significantly reduced the rate of cell death and berry shrivel at harvest, but only in an atypically hot growing season. Conversely, the “early” pulse applied in the 2 wk before typical onset did not affect the rate or onset of cell death in the hot or cool growing season. The onset of cell death and shrivel coincided with accumulation in H2O2 (an ROS with important roles in ripening and programmed cell death) but there were no treatment differences in H2O2, suggesting that the relationship between irrigation and cell death is driven by ROS-independent mechanisms. Future work at an agronomic scale is needed to test whether timing a supplemental irrigation pulse to this period (~85 to 100 DAA) can mitigate late-season dehydration and reductions in yield, income, and quality under future, hotter conditions.
Supplemental Data
The following supplemental materials are available for this article in the Supplemental tab above:
Supplemental Table 1 Total irrigation (L/vine) applied to each treatment in each year over the experimental periods and in total, over the entire year.
Supplemental Table 2 Predawn stem water potential (Ψpd) from the 2022 experimental period. Values are means ± standard errors (n = 10). Letters indicate dates with statistically similar water potentials, based on Tukey’s honest significant difference test. Each treatment was tested separately.
Supplemental Table 3 Predawn stem water potential (Ψpd) from the 2023 experimental period. Values are means ± standard errors (n = 5). Letters indicate dates with statistically similar water potentials, based on Tukey’s honest significant difference test. Each treatment was tested separately.
Supplemental Table 4 Stomatal conductance (gs) on each sampling date from the 2022 and 2023 experimental periods. Values are means ± standard errors (2022: n = 10; 2023: n = 5). Letters indicate dates with statistically similar water potentials, based on Tukey’s honest significant difference test. Each year and treatment were analyzed separately. DAA, days after anthesis.
Supplemental Figure 1 Maximum daily temperatures (A) over the experimental periods and monthly precipitation (B) for 2022 and 2023. Irrigation treatments were applied in 2022 from 25 July to 22 Aug and in 2023 from 21 Aug to 15 Sept. Harvest took place in 2022 on 26 Aug and in 2023 on 3 Oct. Data were collected from the California Irrigation Management Information System (CIMIS) weather station in Davis, CA (https://cimis.water.ca.gov/).
Supplemental Figure 2 Stomatal conductance (gs) and photosynthesis (An) rates for the three irrigation treatments over the 2022 (A, C) and 2023 (B, D) experimental periods. Point colors indicate irrigation treatments; each point is the average of n = 10 (2022) and 5 (2023) leaves per treatment. Error bars are standard errors and asterisks indicate dates with significant differences between treatments (analysis of variance, p < 0.05). The pairs of vertical dotted lines in each panel show the dates for the irrigation pulses. DAA, days after anthesis.
Supplemental Figure 3 Correlation between berry shrivel index (ShI) and percent cell vitality. Point colors indicate treatment and each point is the mean for each combination of treatment and date. Shapes indicate the year. The solid line indicates the best-fit linear regression.
Supplemental Material Map showing the irrigation treatment assigned to each vine position examined in this study. The control and early- and late-pulse irrigation treatments are described in the Materials and Methods. The two vines adjacent to each experimental vine served as buffers and received the same irrigation as the experimental vine. All unlabeled vine positions received the same irrigation as the control treatment.
Data Availability
All data underlying this study are included in the article and its supplemental information.
Footnotes
This work was supported by the American Vineyard Foundation [#AVF-2739], the University of California, Davis College of Agricultural and Environmental Sciences and Department of Viticulture and Enology, and generous donations from the Rossi family to the department. We are also grateful to Dr. Ryan Stanfield for help with microscopy techniques, Dr. Rosa Figueroa Balderas with measuring berry chemistry, and Guillermo Garcia Zamora with vineyard management.
Ritter-Jenkins A, Serrano AS, Corella Caballero RI, Gao T, Knipfer T, Shackel K et al. 2026. Irrigation timing affects grape berry cell death and late-season dehydration. Am J Enol Vitic 77:0770007. DOI: 10.5344/ajev.2026.25048
By downloading and/or receiving this article, you agree to the Disclaimer of Warranties and Liability. If you do not agree to the Disclaimers, do not download and/or accept this article.
- Received October 2025.
- Accepted January 2026.
- Published online April 2026
This is an open access article distributed under the CC BY 4.0 license.
References
Vol 77 Issue 1
Jump to section
Related Articles
Cited By...
More from this TOC section
Similar Articles
