Abstract
Grape phenolics are a relevant part of grape quality, and their metabolism in the berry may be modified by environmental factors. To determine the influence of different water regimes on the polyphenolic composition of berries, research was conducted for two years (2000 and 2001) on three-year-old Vitis vinifera cv. Merlot potted vines, comparing three water supply levels: (1) control (C), at ~80% of available water (aw); (2) moderate stress (M), at ~30% aw; and (3) severe stress (S), at ~15% aw. The treatments were applied in both years from veraison to fruit maturity. Predawn leaf water potential was reduced only in the S treatment in 2000, while in 2001 both M and S treatments had a lower value. Berry weight was reduced in M and S treatments, but no differences were observed in sugars, pH, and berry titratable acidity. Total skin and seed polyphenols were increased in S vines, but a probable increase of the structural complexity of phenolic compounds inside the berry (higher degree of polymerization) caused a lower extractability in winelike solution. Anthocyanins, which are monomers, were more concentrated and also more extracted in S berries. Water-stressed Merlot grapes (both S and M) will benefit from a longer maceration because a higher concentration of polymerized polyphenols could be extracted, stabilizing color and improving the mouthfeel properties of the resulting wines.
- ΨPD, predawn leaf water potential
- A, net photosynthesis
- aw, available water
- C, control
- ci, intercellular CO2 concentration
- E, transpiration
- EtOH, ethanol 12% solution
- gs, stomatal conductance
- M, moderate stress
- MeOH, methanol-HCl solution
- S, severe stress
- TB, tartaric buffer
Plant physiology and yield are modified when different water supply regimes are applied (Carbonneau and Deloire 2001). Water-stress tolerance has been shown to be strongly related to the genetic origin of the grapevine. At similar conditions of water stress Grénache was found to be more sensitive to water shortage than Syrah (Schultz 1996) and leaf area development was different between Manto Negro and Tempranillo (Escalona et al. 2003). Moreover, rootstock genotype affected gas exchange response of Müller Thurgau (Iacono et al. 1998). However, Gomez-del-Campo et al. (2002) reported similar values of photosynthesis between cultivars under water shortage, detecting only a different timing in dry matter partitioning. In terms of quality, Medrano et al. (2003) found no relation between water stress and berry composition in different cultivars.
In grapevine, the purpose of irrigation is to reduce and manage crop water deficit and thereby maximize yield and must quality in order to increase profit. Available water is the amount of water in soil that is usable for plant growth: it ranges from 30 mm m−1 of soil depth in sandy soil to 160 mm m−1 in clay (Smart and Coombe 1983). Williams and Matthews (1990) reported a reduction in berry growth when soil moisture was reduced from 90 to 25% of available water. Ginestar et al. (1998a) and Esteban et al. (2001) reported that water stress directly affected plant physiology, which in turn modified grape production and composition. There are conflicting conclusions as to the positive or negative effects of water stress on grape composition (Sipiora and Gutierrez-Granda 1998). Several studies mainly focused on the relationship among drought, accumulation of sugars, evolution of acidity, and pH (Goodwin and Jerie 1992, Koundouras et al. 1999). Williams and Matthews (1990), Ginestar et al. (1998b), Sipiora and Gutierrez-Granda (1998), and Esteban et al. (2001) revealed that water stress reduced berry size, thus increasing polyphenol concentration. Esteban et al. (2001) also demonstrated that the enhanced polyphenol concentration was due to berry size reduction and that an increased biosynthesis of these substances occurred. McCarthy and Coombe (1985) and McCarthy (1996) reported that water stress increased bound terpenes but did not affect free terpene concentration.
The level of water stress that may optimize vine growth and grape quality has not been quantified for all locations where grapevine is cultivated (Williams 2002). Given this open question, the aim of this trial is to evaluate the most suitable percentage of soil available water resulting in optimal and balanced growth of the plant and quality grapes.
Materials and Methods
A pot trial was carried out during 2000 and 2001 in Udine (Friuli region, northeastern Italy), in the experimental farm of the University of Udine, on three-year-old Merlot grapevines clone R8 grafted on SO4 rootstock clone 31 OP. The plants were pruned to two spurs, each bearing two buds. The shoots grew upward, and four clusters per plant were left. Clusters were harvested at berry set on half of the plants to assess the interaction of crop load and water stress. The 80-L pots were filled with a loamy soil of field capacity = 22.6% [(vol water/vol soil)*100] and wilting point = 12.2%. A completely randomized experimental plan was set up to compare different water supply levels: (1) control (C), at ~80% available water (aw); (2) moderate stress (M), with the plants kept at ~30% aw, and (3) severe stress (S), at ~15% aw. In the first year, the different water stress levels began on 21 July 2000, one week before veraison (27 July), and these levels were retained until harvest (13 September). In the second year, water regimes began on 25 July 2001, five days before veraison (30 July), and were retained until harvest (14 September). Water was daily supplied with a drip-irrigation system between 0900 and 1030 hr.
During the period of water stress, each pot was covered with a polyethylene film to avoid soil water evaporation. The plastic film was laid over the edge of the pots, without tightening: therefore, we estimated that root gas exchanges were not significantly changed and that possible anoxia of the roots did not occur. Pots were covered also with aluminum foil to avoid excessive substrate heating. Each treatment was applied to 16 plants (one plant = one replicate). Until the beginning of water stress, the vines were well irrigated. After the onset of water stress in 2000, the daily water supply per pot was 2.8 L, 1.0 L, and 0.5 L for C, M, and S treatments, respectively; during July and August water supply was increased, reaching 3.7 L, 2.4 L, and 0.9 L for C, M, and S treatments, respectively. This adjustment was necessary to match soil humidity targets. In 2001 the daily water supply per pot was 3.3 L, 1.8 L, and 0.7 L for C, M, and S treatments, respectively, and these amounts were retained until harvest. This change was dictated by the results of the first year of measurements. For the two years, ETo was evaluated with Penman Monteith equation and ETc value was estimated as 2.6 and 2.7 L pot−1 at the beginning of water stress in 2000 and 2001, respectively. The control level of 2.8 L per pot of water consumption was determined considering ETc and an average transpiration rate of 3 mmol H2O m−2 sec−1 for healthy, sun-exposed, and well-watered grapevines. At the beginning of water stress in 2000, leaf area was 1.90, 1.91, and 1.81 m2 plant−1 for C, M, and S treatments, respectively, without significant differences. During August, leaf area changed, reaching 2.60, 2.01, and 1.39 m2 plant−1 for C, M, and S treatments, respectively, with significant differences among all treatments. In 2001, leaf area was not measured but it was similar. The value of transpiration of the whole canopy was considered to be 50% of the previously mentioned maximum value of healthy leaves because of the different leaf age and shading. The total amount of water supplied during the period of water stress is shown in Table 1⇓.
Soil moisture.
Soil moisture (v/v) was measured from veraison to harvest by means of 370-mm steel rod probes using a time-domain reflectometry (TDR) cable tester (model 1502B; Tektronix, Beaverton, OR) at weekly intervals. Probes were vertically inserted in the pot soil 150 mm from the crown. Readings were taken once a week immediately before irrigation.
Predawn leaf water potential (ΨPD).
Leaf ΨPD was measured from veraison to harvest every two weeks at 0500 hr (4 hr before irrigation) with a Scholander pressure chamber (PMS, Corvallis, OR) supplied with a compressed air cylinder (Scholander et al. 1965); four young fully expanded sun-exposed leaves were measured for each treatment. Leaves were cut and inserted in the pressure chamber; pressurization started no later than 10 sec after leaf cut.
Gas exchange.
Net photosynthesis (A), transpiration (E), stomatal conductance (gs), and internal CO2 concentration (ci) were determined every two weeks from veraison to harvest with a portable gas-exchange analysis system (model LI-6400; LI-COR, Lincoln, NE). Measurements were performed between 1300 and 1500 hr (3 to 4 hr after irrigation). Six young fully expanded sun-exposed leaves were sampled for each measurement.
Xylem sap flow.
Xylem sap flow was determined with single heat-balance sap-flow gauges (Peressotti and Ham 1996). Three gauges on three different plants per each water regime were applied to the plant stem, and the data were collected in 2000 every 15 min from veraison to mid-October using a CR10X data logger (Campbell Scientific, Shepshed, Leicestershire, UK).
Berry weight and composition at harvest.
From veraison to harvest, 200-berry samples were randomly collected every week from the top, bottom, and center of the bunch, both from shaded and sun-exposed portions of clusters; thereafter berries were weighed and berry diameter was measured with a calliper. For each sample, 50 berries were crushed for 2 min (model Y47.R1; Moulinex, Ecully, France) and used to determine sugars (Brix), total acidity (g L−1 tartaric acid), and pH.
Skin and seed phenolic analytes.
At harvest skins and seeds (from 25 berries x rep) were carefully isolated from the flesh and extracted as follows: from skins, anthocyanin and total polyphenol concentration was determined; from seeds, only total polyphenol concentration was analyzed. Three extracting solutions were used in 2000: (1) tartaric buffer (TB), prepared with 5 g L−1 of tartaric acid buffered at pH 3.2 with 1N NaOH, simulates extraction during the first four hours of fermentation when no ethanol was present in must; (2) methanol-HCl (MeOH), prepared with 99% (v/v) of methanol (99% pure) and 1% (v/v) of HCl solution in water (36%), extracts nearly all the phenolics from the berries (Celotti et al. 1999); and (3) ethanol 12% (EtOH), prepared with 12% of ethanol (95% pure) in water (v/v) added with 5 g L−1 of tartaric acid buffered at pH 3.2 with NaOH 1N. In 2001 only the extractions with EtOH were repeated on skin and seed aliquots.
Total phenols were determined according to Ribereau-Gayon (1970), and anthocyanins were analyzed as reported by Ribereau-Gayon and Stonestreet (1965).
Results
Soil moisture.
In the first year (2000), after commencement of water regimes, soil moisture decreased in S and M treatments (Figure 1A⇓). For both S and M treatments, soil moisture reached values that were lower than 30% aw target; this situation was maintained until 24 August. Soil moisture then increased in M plants, reaching values close to 30% aw during the last period of berry ripening. Soil moisture did not increase for S plants, which remained at ~15% aw. The trend of humidity of C plant pots showed a reduction during the second and third week of August followed by a recovery.
In the second year (2001) soil moisture (Figure 1B⇑) dropped after onset of water stress in all treatments. From the fourth week of August soil moisture values reached and maintained their target levels until harvest. On 14 September 2001 a severe storm partly destroyed the covering of the plants; soil moisture subsequently increased in the final five days before harvest.
Leaf water potential.
In the first year of trial (2000), ΨPD was measured four times during the treatment period. From the onset of the soil moisture deficit period, S plants demonstrated reduced ΨPD in response to water shortage, but M and C plants maintained a level that was not different from that measured before imposition of the deficit (Figure 2A⇓).
For all water stress treatments, the presence of clusters reduced ΨPD and the interaction water stress/presence of clusters was found to be significant; the presence of clusters in S vines significantly reduced ΨPD (Table 2⇓), but no difference was recorded for C and M plants. In 2001, consistent with the results of the first year, S plants displayed lower ΨPD and differences occurred between M and C plants (Figure 2B⇑).
Gas exchange.
Photosynthesis (A), internal concentration of CO2 (ci), transpiration (E), and stomatal conductance (gs) were affected by water shortage. In 2000, two weeks after treatments began, A was reduced as a function of water shortage: C and S plants displayed reduced A during the observed period, but M plants maintained A at ~7 μmol CO2 m−2 sec−1 (Figure 3A⇓). E exhibited a behavior similar to A, but, as reported for photosynthesis, transpiration of M plants remained high at about 3 mmol H2O m−2 sec−1 until the final measurement, equaling that of C plants (Figure 4A⇓).
In 2001 the first date of measurements revealed a value of A of 0.9 μmol CO2 m−2 sec−1 for S plants, which was the lowest value registered during the two years of trial (Figure 3B⇑), together with a high value of ci (Figure 3D⇑). From the second measurement ci values followed water regimes with lower values for S plants and higher values for C plants. After a first phase, an increase of gs (Figure 4D⇑) and A (Figure 3B⇑) was obtained in M plants. In 2001 M plants showed an adaptation to water shortage primarily in early September, close to harvest.
Xylem sap flow.
Xylem sap flow was measured during the first year (Figure 5⇓). Until 24 August 2000, when temperatures were high (Figure 6A⇓), sap flow was intense in C plants and lower in S and M plants, showing values that were 2-fold and 8-fold higher in M and C plants, respectively, compared with S plants (Figure 5⇓). From the beginning of September, when temperatures dropped, xylem sap flow was strongly reduced in C plants and M and S plants reached comparable levels. M and S plants maintained their sap flow levels for 15 more days, but after harvest (13 September 2000), xylem sap flow was substantially reduced for all treatments. In comparing xylem sap flow with transpiration, it was possible to ascertain that the correlation between the two methods was good (R2 = 0.6795**).
Berry weight, sugars, titratable acidity, and pH.
Berry diameter (Figure 7⇓) allowed calculation of skin/pulp ratio, which may be important for determining wine quality. In 2000, no differences among treatments were found in mid-July before the beginning of water stress. From the second half of July onward, S plants displayed reduced berry diameter but M plants were not different than C plants until early September. The same samples were measured for their berry weight (Table 3⇓); a normal pattern of development was observed and the reduction in berry weight was directly related to the intensity of water stress. At harvest the largest berries were on C plants, intermediate on M, and smallest on S.
During the second year (2001) differences between S and M versus C plants occurred in berry weight (Table 3⇑): berry size was reduced by water stress. A similar difference was observed in berry diameter (Figure 7B⇑), but did not reach statistical significance. Brix, titratable acidity, and pH did not reveal differences among treatments.
Phenolics and anthocyanins.
The different solutions proved to extract total polyphenols and anthocyanins with a different strength: MeOH was the most effective in all cases, followed by EtOH and TB. This fact had diverse effects on samples obtained in different water regimes.
Phenolic extraction increased throughout ripening, as reported in a related paper (Sivilotti et al. 2004). In 2000, total polyphenol concentration in skins was higher in C berries as compared with S and M berries (Table 4⇓) when skins were extracted in TB or EtOH, while S berries in MeOH released a greater concentration of total polyphenols. In 2001, EtOH extracted less polyphenols in C skins, an opposite trend compared with the previous year.
Anthocyanins were extracted in a higher concentration in S berry skins when MeOH or EtOH was used (year 2000), which was confirmed in 2001 by the EtOH extraction. Anthocyanins did not show significant differences among treatments when extracted with TB (year 2000), but the concentration in S skins was slightly lower (Table 4⇑).
In the seeds sampled in 2000, EtOH extracted greater concentrations of total polyphenols from C seeds, but no differences were assessed when the extraction was performed with TB. With MeOH, the highest concentration of total polyphenols was extracted from S seeds (Table 5⇓). In 2001, EtOH extraction did not show any difference among treatments (Table 5⇓).
Discussion
Soil moisture was reduced as water was withheld. The measured data did not consistently match the treatment description: based on soil-moisture data, daily irrigation of the pots was adjusted, but an interaction between soil water availability and plant transpiration could account for the high variability of data (Figure 1A,B⇑).
In both years, from the beginning of water stress, soil moisture was reduced until harvest according to the applied water regimes, and differences among treatments were assessed. In 2000, S and M plants reached similar values of soil moisture until the beginning of September (Figure 1A⇑), probably because of higher transpiration of M plants (Figure 4A⇑). In 2001, the air temperature from the end of July to mid-August after the onset of water stress was about 5°C higher than the same period in 2000 (Figure 1A, B⇑). Winkel and Rambal (1993) reported that higher temperatures are responsible for increased transpiration (Figure 4A,B⇑). M and S plants showed a lower transpiration with higher temperatures (2001 versus 2000) and also higher values of soil moisture during the first period of water stress in 2001. This fact could be attributed to the increased amount of water that was supplied to the M and S plants in 2001 compared with 2000. Water supply was much higher than ETc, thus resulting in a slower reduction of the soil moisture in M and S plants. Another explanation could be that M and S plants probably had a “stress memory” from the previous year, they had a smaller root system, and so they reduced water uptake and leaf water potential (Figure 2B⇑), forcing the plants to close stomata to avoid dehydration. After this first period of physiological adaptation (from 25 July to 15 August 2001), soil moisture reached water-regime target level. In both years, when the temperatures started to decrease, soil moisture recovered, notably in M plants.
Leaf water potential drops when a severe stress is applied to the plants; ΨPD is not reduced linearly with the reduction of water availability (Jensen et al. 1998, Carbonneau and Deloire 2001) because several mechanisms act, such as osmotic adjustments, thus maintaining the level of ΨPD to high values (when water stress is not physiologically damaging the plant). There are different opinions about the threshold that discriminates severe and moderate water stress. Hardie and Considine (1976) observed that stress symptoms begin to appear whenΨPD approaches −0.4 MPa, and Smart and Coombe (1983) for the same condition indicated a value of −0.7 MPa. Ojeda et al. (2002) classified a medium water stress as ranging from −0.4 to −0.6 MPa of ΨPD. Davies and Mansfield (1993) explained that during moderate water deficit, plants could improve the root absorption efficiency, thus preserving a better hydraulic condition inside the plants. This suggestion could explain why M and S plants reached the same values of soil moisture until the end of August when the supplied water was double in M compared with S (Figure 1A⇑). In 2001 the values of ΨPD in M and S plants were reduced by water stress (Figure 2B⇑). The behavior of M plants could be attributed to the higher temperatures that occurred during 2001 as compared with 2000 (Figure 6A, B⇑), thus aggravating the water stress effects.
The presence of fruit intensified effects of water stress. A possible explanation of this behavior is that when water supply becomes too limited, clusters become the strongest sink for water (Smart and Coombe 1983): reproductive functions prevail over vegetative functions, and so the level of ΨPD decreases.
Gas exchange was more sensitive to water stress than ΨPD (Figures 3⇑ and 4⇑). A similar influence of water stress on gas exchange was reported by several authors (Ginestar 1998a, Escalona et al. 1999). Comparing the two years of trial, gs and E (Figure 4⇑) were much more affected by water stress than A and ci (Figure 3⇑), suggesting that stomatal mechanisms are important in order to avoid dehydration and to control water balance inside the leaves. In the second year, E (Figure 4B⇑) was reduced in M plants as much as A (Figure 3B⇑) and ci (Figure 3D⇑); thus, it may be that grapevines could “remember” the water-supply history (that is, the previous year water shortage), seeking to avoid irreversible damage to photosynthetic activity. Furthermore, plants could regulate stomatal conductance and transpiration according to temperature. In 2001, when temperatures were higher in the first part of the water stress period, transpiration and ΨPD for M plants were lower than in 2000. These findings could suggest that water stress effects on vine physiology are amplified as temperatures increase. How these mechanisms work is still to be discovered.
Xylem sap flow measured in the trunk (in this case) permitted transpiration to be measured on the whole plant. Escalona et al. (2001) reported high levels of correlation between xylem sap flow and gas exchange measurements, assessing that xylem sap flow could be a useful tool in detecting water stress in plants. However, the correlation was poor at high transpiration rates: as transpiration increased, the range of leaf efficiency was wider and the difference between the most efficient and the least efficient leaf was higher.
Berry growth via cell expansion can be reduced in response to water shortage (McCarthy and Coombe 1999). McCarthy (2000) showed that short-term moderate stress applied after veraison does not affect berry growth. That could explain why no differences in berry weight were observed between M and C plants during the first period of water stress in 2000 (Figure 7⇑). Berry weight is affected when a long-term moderate water shortage is imposed (Matthews et al. 1987), which could explain why a reduction of berry weight was detected only near harvest.
Brix, titratable acidity, and pH (Table 3⇑) showed no response to water deficit treatments in this study. The literature shows disagreement about the effect of water stress on these variables. Kounduras et al. (1999) and Ginestar et al. (1998b) reported that water stress affected berry size, thus increasing sugar level. Goodwin and Jerie (1992) explained that water stress could account for lower sugars and higher acidity in stressed plants, when water stress was applied at veraison. These conflicting findings could be due to differences among environments, water supplies, and weather conditions of the studies.
An increase in phenolics has been demonstrated when water stress occurred during grape ripening (Ginestar et al. 1998b, Esteban et al. 2001). We found similar results when skins or seeds were analyzed with MeOH. The literature reflects little knowledge concerning the real extractability of phenolics during winemaking. TB and EtOH are solutions that can simulate the beginning and the end of the fermentation process, respectively. By using TB it was possible to assess that no phenolics were extracted from seeds (Table 5⇑). An explanation of this finding is that the presence of mucilaginous matter on the seed surface prevented phenolics from being extracted. The presence of a minimum amount of alcohol could dissolve this layer, thus allowing phenolic extraction as shown by the high level of extraction by EtOH data (Table 5⇑). Seed total polyphenols were extracted at a significantly lower concentration from EtOH in S berries. This finding may indicate that must quality could be higher in response to S because seed tannins (the majority of total polyphenols are tannins) are usually more bitter and astringent than skin tannins (Brossaud et al. 2001, Peleg et al. 1999). Extraction with MeOH removed a higher concentration of total polyphenols in S seeds than did EtOH, which may be due to a higher degree of phenolic polymerization in S berries, which was likely due to a more prolonged ripening.
In skins, the extractability of anthocyanins (which are monomers) in TB was higher in C berries than in S berries. This may have been due to a tighter berry cell-wall structure in stressed plants resulting from faster senescence and a lower tissue hydration. In this case, the extraction of substances such as anthocyanins was poor when a weak solvent was used. EtOH extraction of anthocyanins, which were more concentrated in S berries, indicates that stressed grapes would benefit from a longer period of pomace contact; in fact, the pigment release from the skins to the must is easier when a higher amount of alcohol occurs in the must. Another positive aspect of long maceration of S berries is the resulting lower extraction of seed tannins (Table 5⇑). However, a higher concentration of anthocyanins is not sufficient to obtain a stable color in wine: pigments must be stabilized by copigmentation (Boulton 2001).
Carbonneau and Deloire (2001) reported that phenolics need to be more polymerized in order to equilibrate with a high osmotic concentration; thus the reduction of berry size could account for a limited phenolic extraction even if the amount per berry may be the same and the concentration higher. That could explain why S plants had higher values of total phenols together with a lower extraction through the use of TB or EtOH (Tables 4⇑ and 5⇑).
Conclusions
Deficit irrigation appears to be an important tool for improving grape winemaking quality. A moderate vine stress level allowed the plants to maintain leaf functions similar to the control plants. More severe vine stress strongly affected physiological parameters, but resulted in improved berry quality, particularly phenolic concentrations. It is likely that berry quality would be impaired by extreme water stress. An optimal level of stress might be attained by maintaining the available water between 30% and 15%. Our results indicate that water-stressed Merlot grapes could benefit from a longer maceration, because a greater concentration of polymerized polyphenols could be extracted, stabilizing color and improving the mouthfeel properties of the resulting wines.
Footnotes
Acknowledgments: This research was supported by the MIUR, Italian Ministry of Education, University and Research, COFIN 2000 project. The authors would like to thank the staff of the experimental farm, University of Udine, for their cooperation.
- Received October 2003.
- Revision received March 2004.
- Revision received August 2004.
- Copyright © 2005 by the American Society for Enology and Viticulture