Abstract
A Concord production trial conducted from 1998 to 2002 in Fredonia, New York, examined the effect of pruning level and canopy division on yield, vegetative growth, and fruit characteristics to a harvest target of 16 ± 0.5 Brix. Retained nodes per vine ranged from 56 to 383 on five single-wire (SW) trained treatments and from 90 to 260 on three Geneva double-curtain (GDC) trained treatments. Increasing retained nodes per vine increased vine yield and decreased the rate of juice soluble solids accumulation, which delayed harvest date. Increasing retained nodes above 260 nodes on SW training did not increase yield but additionally delayed juice soluble solids accumulation and harvest date. The relationships between retained nodes and yield, yield components, canopy growth, leaf area, lignified nodes of periderm, and fruit maturation were similar in both SW and GDC training with no advantage to the divided canopy system. Season had a greater effect on veraison titratable acidity and the rate of titratable acidity decline to harvest than did pruning or training treatments. Juice soluble solids and color were closely related, and since all treatments were harvested at a predetermined Brix level, there was no effect of season, pruning, or training on juice color at harvest.
Concord (Vitis labruscana Bailey) is an economically important grape variety for the New York (NY) grape industry. In 2003, 49% of the United States grape production east of the Continental Divide was in NY and 73% of the NY grape production was Concord (NASS 2004). Approximately 78% of NY Concord yield was for juice and jelly production and 22% was for wine production, comprising 52% of the winegrape crush in NY (NASS 2006).
Juice processors purchase Concord grapes based on yield and juice soluble solids irrespective of other juice characteristics, such as color, pH, or titratable acidity. The average price per tonne of Concord grapes has dropped from $263 in 2001 to $145 in 2004 (Shaffer and White 2006). The economic production goal for Concord growers is to consistently produce the highest yield that will mature to a minimum juice soluble solids concentration before the end of the commercial harvest season and to achieve this at minimum production costs. The challenge in achieving this goal lies in developing lower-cost mechanical techniques within the framework of sound viticulture concepts such as vine size, light interception, and leaf area to crop weight ratios.
Pruning practices have changed in the NY Concord industry to satisfy the “high yield at minimum Brix” goal. In the 1950s and 1960s, different balanced-pruning formulas were evaluated for effects on vine size, yield, and juice soluble solids (Kimbal and Shaulis 1958). Conservative formulas, such as retaining 22 nodes for each 500 g of dormant cane pruning weight (20 nodes per pound) with a maximum of 60 retained nodes per vine on standard vine spacing (2.7 m row x 2.4 m vine), were adopted because such formulas ensured an adequate exposed leaf area to fruit ratio to fully mature the crop in the cool NY climate. However, the conservative formulas also limited the maximum potential crop that could be matured in warmer and longer than average growing seasons. Early studies suggested evaluating pruning formulas from site to site to match crop load with the growth and ripening characteristics of the region. For example, research in Washington and Arkansas found 50+10 pruning (50 nodes per vine for the first pound of pruning weight [55 nodes/500 g] and 10 nodes for each additional pound [11 nodes/500 g]) more suitable for Concord production in regions with relatively longer growing seasons than NY (Clore and Brummund 1961, Cawthon and Morris 1977, Morris et al. 1984). The practice of fixed-node pruning at 80 to 100 nodes per vine has increased in the NY Concord juice industry because the additional nodes per vine yield higher crops, which can be fully matured in good seasons. However, the larger crops struggle to reach minimum commercial juice soluble solids standards in poor seasons. Despite the risk for overcropping, fixed-node pruning more closely fits with the production goals of higher yield, minimum Brix, and lower production costs.
After adoption of mechanical harvesting in Concord vineyards in the early 1970s, labor for dormant vine pruning became the highest production cost and interest in mechanical pruning increased. However, the specificity gained through manual balanced pruning in terms of retained node quantity and quality decreased with mechanization. Mechanical pruning or prepruning Concord research began in the mid-1970s with the intention of lowering production costs and maintaining appropriate node quantity and quality (Cawthon and Morris 1977, Pollack et al. 1977). Original techniques used downward mechanical shoot positioning 15 to 20 days after bloom, mechanical pruning to retain only those canes below the cordon, manual cane thinning, and mechanical scrubbing of green shoots originating above the cordon.
A sharp decline in Concord grape prices in the 1980s initiated evaluation of lower labor-cost hedge and minimal pruning methods (Pool et al. 1993). Commercially, high retained nodes and shoots per vine with minimal pruning led to high crop levels and dense canopy structures, both of which were implicated in causing delayed juice soluble solids accumulation (Lakso et al. 1996, Miller et al. 1993, Miller and Howell 1998, Morris and Cawthon 1981). Given the cool and short growing season in New York, commercial minimal pruning often led to fruit not reaching minimum juice soluble solids before the end of commercial harvest (Bates and Walter-Peterson 2008, Zabadal et al. 2002).
Geneva double-curtain (GDC) training was originally studied as a management tool to address excessive vine size (Shaulis et al. 1966, 1967). Under aggressive weed control, invigorating rootstocks, and conservative balanced pruning practices, Shaulis’ research showed the positive relationship between vine size and yield in NY Concord vineyards. It also illustrated the potential for excessive vegetative growth, internal canopy shading, and decreased bud fruitfulness. GDC training uses excessive leaf area produced by large vines by spreading out the leaf area along two trellis wires and intercepting more sunlight per unit land area. GDC training was intended to increase the fruit maturation rate by increasing the exposed leaf area to fruit weight ratio per unit land area and not to necessarily increase fruit yield per unit land area. Similar to the history of standard single-wire (SW) training, commercial use of GDC training in NY Concord vineyards has trended toward increased retained nodes per vine to maximize fruit yield at a minimum juice soluble solids standard.
The goal of this study was to investigate the effect of retained node number and canopy division on the yield, canopy development, and juice characteristics of Concord grapevines when harvest was based on a predetermined juice soluble solids concentration.
Materials and Methods
Vineyard description.
The experiment was conducted from 1998 to 2002 in a 0.7 ha Concord (Vitis labruscana Bailey) vineyard at the Cornell Vineyard Laboratory in Fredonia, NY (lat. 42°21′N; long. 79°18′W). The experimental block elevation was 231 m above sea level with a slope of 1 to 2% and a southeast aspect. The soil was well-drained Chenango gravel-loam with a surface soil pH of 5.5 and 2% organic matter.
The Lake Erie grape production region is characterized by cool and humid conditions. The 38-year (1964 to 2002) average GDD accumulation (base 10°C) from 1 Apr to 31 Oct recorded at the Fredonia Vineyard Laboratory site was 1373 GDD (Figure 1A⇓). Average precipitation accumulation for the same time period was 67 cm, with the rainfall evenly distributed throughout the growing season (Figure 1B⇓). Supplemental irrigation was not used in this experiment because it is rarely used for commercial NY Concord production. Thirty-year average Julian phenology dates for Concord grapevines at the same site are in Table 1⇓.
The seasonal GDD accumulations (1 Apr to 31 Oct, base 10°C) for the course of the experiment were 1650, 1664, 1462, 1625, and 1686 in 1998, 1999, 2000, 2001, and 2002, respectively (Figure 1A⇑). All five data-collection years were warmer than the long-term average, with 2002 the warmest and 2000 the coolest and closest to the long-term average. Above average spring temperatures in April and May led to earlier than average Concord budbreak in all five years and earlier than average bloom in four out of five years.
The seasonal precipitation accumulation (1 Apr to 31 Oct) in cm for the course of the experiment was 54.7, 56.9, 69.2, 52.1, 69.4 in 1998, 1999, 2000, 2001, and 2002, respectively (Figure 1B⇑). Seasonal precipitation was below average in 2001 and above average in 2000. The other three years had above average precipitation in the spring, which tapered to average or below average in the middle of the growing season.
The own-rooted Concord vines were 40 years old at the beginning of the experiment and planted at a row-by-vine spacing of 2.6 x 2.4 m with rows in the east-west direction. Vines were cordon trained to a trellis wire at 1.8 m. Floor, nutrient, pest, and disease management were according to commercial standards for western NY Concord vineyards (Jordan et al. 1980). No-till weed management was used by maintaining a 1.2 m-wide weed-free zone under the vines with pre- and postemergence herbicides and treating row centers with one glyphosate application at bloom. Ammonium nitrate fertilizer was surface broadcasted across the block in a single application around budbreak at a rate of 56 kg/ha of actual N. Fungicide and insecticide materials and application rates were done according to the NY and Pennsylvania pest management guidelines for grapes (Weigle 2006) and varied annually depending on seasonal weather conditions.
Viticulture treatments.
There were five pruning level treatments on single-wire (SW) training and three pruning level treatments on Geneva double-curtain (GDC) training. The five SW pruning levels were balanced pruning (SW Bal: manual balanced pruning, leaving 22 nodes per vine for each 500 g of dormant cane pruning weight), manual fixed-node pruning at 90 and 130 nodes per vine (SW 90, SW 130), mechanical hedging (SW 260: mechanical hedge pruning with a vine separation cut to target 260 nodes per vine), and minimal pruning (SW Min: mechanical horizontal pruning plane at 91 cm aboveground with additional vine separation cuts). The three GDC pruning levels were manual fixed-node pruning at 90 and 130 nodes per vine (GDC 90, GDC 130) and machine hedging targeting 260 nodes per vine (GDC 260). Pruning treatments, within training, were imposed in a randomized complete block design with three replicate blocks and three-vine plots per treatment-block combination. SW and GDC rows were not random in the vineyard; therefore, SW and GDC were analyzed as two separate pruning experiments. Each three-vine plot was bordered by three buffer vines on each side with similar pruning. Temperature and precipitation were recorded with a Campbell weather station at the Cornell Vineyard Laboratory in Fredonia, NY, and the data was retrieved from the Lake Erie Regional Grape Program weather database (http://lergp.cce.cornell.edu/weather.htm).
Vegetative measurements.
Prior to dormant cane pruning each year, lignified nodes per vine were counted. After rough pruning, dormant cane prunings were weighed and retained nodes were adjusted according to treatment descriptions. Prior to bud swell, three midcane nodes per vine were selected and flagged for shoot and leaf measurements to be taken later in the growing season. Prior to bloom, count shoots (shoots originating from a measured retained node), noncount shoots (shoots originating from base or latent buds), and blank nodes were counted on all treatment vines. At bloom and 650 GDD (~30 d postbloom), primary shoot length, lateral shoot length, leaves per shoot, and midrib leaf length of each leaf on a shoot were measured on each of the three shoots per vine selected before bud swell. Midrib leaf length was converted to leaf area by the regression equation [y = 0.9442x2 + 0.9309x, where x = midrib leaf length (cm) and y = leaf area (cm2), R2 = 0.9393, n = 454] (recalculated after Elsner and Jubb 1988). Shoot length and leaf area per shoot were averaged for the three count shoots per vine and then multiplied by the total shoot number to give total shoot length and total leaf area per vine.
Yield measurements.
Fruit from the eight treatments was harvested when juice soluble solids reached 16.0 ± 0.5 Brix. Starting at veraison, 150 to 200 berries per three-vine plot were collected weekly and analyzed for juice soluble solids to determine the fruit maturation rate and target harvest date. The berry sample was randomly collected from two vertical planes transecting the east and west side of each vine. Juice soluble solids were measured with a hand-held Leica refractometer (model 10423; Leica, Inc., Buffalo, NY) and the remaining fruit was frozen at −5°C for further juice analysis. When a treatment approached the 16.0 ± 0.5 Brix harvest target, a final 100-berry sample was collected from each individual vine (not the three-vine composite) and measured for berry weight and juice soluble solids. Cluster number and yield were measured by manually harvesting the remaining fruit.
Juice analysis.
Preharvest and harvest berry samples were further analyzed for juice pH, titratable acidity, and color. Juice pH was determined on 150 mL after thawing and homogenizing the samples (Corning 340 pH meter; Corning, NY). For color and titratable acidity, each 150 mL homogenized grape sample was brought to 60°C in a water bath, treated with 1 mL solution of 1% pectinase (CAS No. 9032-75-1; Sigma-Aldrich, St. Louis, MO) for 25 min, and filtered by gravity (12–25 μm particle retention; Whatman, Inc., Florham Park, NJ). For color, 5 mL of the gravity filtered juice sample was diluted with 95 mL of McIlvaine’s buffer (Cat No. LC16300-4, pH 3.2; LabChem Inc., Pittsburgh, PA) and suction filtered (Glass Fiber Filter, Type A/E, 1.0 μm particle retention; Pall Corp., East Hills, NY). Absorbance at 520 nm and 430 nm was measured with a Spectronic 20 Genesys Spectrophotometer (Spectronic Instruments, Rochester, NY) on the double-filtered juice samples against a 100% McIlvaine buffer standard at 22°C. Titratable acidity was determined by diluting 10 mL of the gravity filtered juice sample with 40 mL distilled water and titrating with 0.1 N NaOH solution until pH 8.1 was reached.
Crop-load calculations.
Crop load was calculated by dividing the total vine leaf area measured at 650 GDD by the total vine fruit fresh weight at harvest. Adjusted crop load was calculated by dividing the estimated light-exposed vine leaf area at 650 GDD by total vine fruit fresh weight at harvest. For estimating light-exposed leaf area, canopy dimensions (height, width, length) on all pruning treatments were collected and used to calculate the flat-canopy surface area (~10–11 m2 per vine on SW and 20 m2 per vine on GDC). This value was multiplied by 1.5 to represent the nonflat leaf orientation in each canopy; therefore, 15.5 m2 and 30 m2 light-exposed leaf area values were used to calculate the adjusted crop load in SW and GDC, respectively. If the actual total leaf area measurement at 650 GDD was less than the estimated exposed leaf area (i.e., actual <15.5 m2 on SW), then the lesser value was used.
Data analysis.
Statistical analyses were carried out using the SAS software package (version 8.02; SAS Institute Inc., Cary, NC). The SW and GDC vines were analyzed separately. Data were tested for homogeneity of variance using Levene’s test and subjected to two-way (pruning treatment x season) analysis of variance (ANOVA), using the general linear model and F test. Pruning treatments were also analyzed as one-way ANOVA for each season and post-hoc mean comparisons were done using Duncan’s new multiple range test. Seasons were also analyzed as one-way ANOVA for each pruning treatment.
Results
Treatment nodes and harvest juice soluble solids.
As expected, the five SW and three GDC pruning treatments affected retained nodes in each season. Average retained nodes ranged from 68 on SW Bal to 310 on SW Min and from 90 on GDC 90 to 250 on GDC 260 (Table 2⇓). There was no season effect within a pruning treatment for all but the lightest pruning levels in both SW and GDC. There was no pruning, training, or season effect on harvest juice soluble solids, and the 16.0 ± 0.5 Brix target was accomplished in all but three instances (Table 2⇓). SW Bal in 1998 and 2002 and GDC 130 in 1998 all outpaced the sampling procedure and averaged juice soluble solids slightly above the Brix target.
Yield and harvest date.
On SW trained vines, there was a pruning treatment effect on yield in 1999, 2001, and 2002 with a positive relationship between retained nodes and yield (Table 2⇑). In SW training, yield was lightest in SW Bal vines and heaviest in SW 260 vines. Although SW Min had greater retained nodes than SW 260 in 1999, 2001, and 2002, there was no yield difference between the two treatments, indicating that maximum yield potential through pruning had been reached. There was a similar pruning effect on GDC vines in each year of the experiment; however, it is unclear if maximum yield was achieved with GDC 260 pruning. SW Bal vines with the lowest retained nodes and lightest yield was the first treatment to reach 16 Brix on average (Figure 2⇓). Therefore, the SW Bal harvest date for each year was designated as day 0 and used for comparison with the other treatments (Table 2⇑). In general, increasing retained nodes delayed harvest date in both SW and GDC vines. Although there was no yield difference between SW 260 and SW Min, minimal pruning further delayed Brix accumulation and harvest date compared with SW 260 in three of five years.
Vegetative characteristics.
Pruning treatment and retained node number affected canopy development and structure through shoot numbers, shoot length, and leaf area. There was a positive relationship between retained nodes and count shoots per vine in both SW and GDC trained vines (Table 3⇓). However, light pruning (SW 260, SW Min, GDC 260) also increased the number of non-count shoots and blank nodes adding to the total shoot count and overall shoot density. On average, SW Bal had 32 shoots per m of canopy, SW 130 had 53 shoots per m and SW Min had 121 to 186 shoots per m. There was a pruning treatment effect on shoot length and total leaf area at bloom and on shoot length at 650 GDD for both training systems and for each season (Table 4⇓). Increasing retained nodes decreased shoot length at both measurement times but increased early season leaf area development. By bloom, SW Bal vines had developed 36% of 650 GDD total leaf area and SW Min vines had developed 63%. Bloom leaf area in GDC 90 and GDC 260 were 42% and 53% of 650 GDD total leaf area, respectively. By 650 GDD, there was only a pruning treatment effect on vine leaf area in two of five years in SW training and one of five years in GDC training. Therefore, heavy pruning, such as in SW 90 and GDC 90, led to canopies with longer shoots, larger leaves (data not shown), and lower shoot density than lightly pruned vines. However, there was little difference in total leaf area at 650 GDD.
Yield components.
Pruning treatment affected mean clusters per vine, cluster weight, berries per cluster, and berry weight in both SW and GDC vines (Table 5⇓). In general, increasing retained node number increased clusters per vine but decreased cluster weight, berries per cluster, and berry weight. Therefore, light pruning led to vines with greater nodes, shoots, and clusters, but the yield-compensating effects of fewer berries per cluster and lower berry weight buffered the impact of retained nodes on vine yield.
Crop load and adjusted crop load.
There was no pruning treatment effect on crop load in SW trained vines when total leaf area measured at 650 GDD was used in the crop-load calculation (Table 6⇓). In addition, all crop-load values were at or above the suggested acceptable crop-load range of 0.8–1.2 m2 per kg (Kliewer and Dokoozlian 2005). Lighter pruning with greater retained nodes led to vines with slightly higher yield and total leaf area but also a likely higher percentage of shaded leaf area. The adjusted crop load takes into account the exposed leaf area to fruit ratio, which was affected by SW pruning in three of five years with lighter pruning leading to decreased adjusted crop-load values.
Juice characteristics.
All treatments were harvested at 16 ± 0.5 Brix; therefore, neither pruning nor season had an effect on harvest juice soluble solids (Table 2⇑). However, pruning level did have an effect on the juice soluble solids accumulation rate from veraison to harvest and the relative harvest date (Figure 2⇑, Table 2⇑). Each season, all treatments were between 7 and 8 Brix at veraison and there were no differences among treatments. In general, increasing retained nodes in both SW and GDC vines decreased juice soluble solids accumulation rates, leading to a delayed harvest when compared to SW Bal vines. Neither pruning nor training had an effect on the rate of TA decrease from veraison to harvest within a season (data not shown). However, season did influence veraison TA and the rate of decline from veraison to harvest (Figure 3⇓). The warm 1999 season had veraison TA <30 g/L and finished the season with 10 g/L, on average. The cool 2000 season had veraison TA >30 g/L and finished the season with 13 g/L, on average. Since there was a decrease in TA with time from veraison to harvest, pruning treatments with later harvest dates tended to have lower harvest TA (Table 7⇓). Juice soluble solids and juice color were positively correlated in this experiment (Figure 4⇓); therefore, pruning treatment had no effect on harvest juice color because of the uniform Brix harvest (Table 7⇓). Lighter pruning consistently had lower mean color values but the effect was only significant on SW vines in 2002.
Discussion
The pruning treatments affected vine crop load and Brix accumulation rate by changing both canopy characteristics and crop size. In both SW and GDC training, less severe pruning led to greater retained nodes, higher yield, and decreased juice soluble solids accumulation rates. Increasing nodes per vine increased clusters per vine with a concomitant decrease in berries per cluster and berry weight.
Less severe pr uning also increased shoots per vine, resulting in higher shoot density but shorter shoots and smaller leaves. In two of five years, lighter pruning on SW vines led to more total leaf area; however, all vines developed sufficient leaf area to fill the allotted trellis space, and the additional leaf area with light pruning likely increased the proportion of shaded leaves. The adjusted crop-load calculation attempted to estimate the exposed leaf area:fruit weight and assumed the additional shaded leaf area had a neutral effect on fruit maturation. Kliewer and Dokoozlian (2005) indicated sufficient crop-load values of 0.8–1.2 m2 per kg to fully ripen V. vinifera crops on SW trellis. Concord crop-adjustment research in NY indicates a slightly lower crop load (1.2–1.5 m2 per kg) may be needed for maximum juice soluble solids (Bates 2006). Adjusted crop-load data from this pruning experiment averaged from 1.5 in SW Bal vines with the highest juice soluble solids accumulation rates to 0.91 in SW Min vines with the lowest juice soluble solids accumulation rates (Table 6⇑, Figure 2⇑). Comparing SW 260 and SW Min, both treatments had the same yield, leaf area, and adjusted crop load but SW Min had lower Brix accumulation rates and later harvest date in three of five years. This suggests the dense canopy structure of minimal pruning (greater shoots per vine and shorter shoot length at the same leaf area per vine) had an additional negative effect on Concord Brix accumulation under western NY conditions.
The lack of a yield response with GDC training may be explained by the moderate vine size in this experiment. Original research on GDC training only showed yield, fruitfulness, and juice quality responses when excessive leaf area on large vines (>0.56 kg pruning weight per m of canopy) was divided onto two trellis wires (Shaulis et al. 1967). Average pruning weight on SW 90 and GDC 90 vines was 0.56 kg pruning weight per m of canopy (Table 6⇑). Therefore, vine size in this experimental block was arguably not large enough to justify canopy division for increased juice soluble solids.
Regional comparison.
A Concord study in eastern Washington had some similar pruning treatments, the same harvest criteria, and the same juice analysis (Keller et al. 2004), although the NY experiment compared a wider range of retained nodes (56–383 nodes per vine in NY and 87–162 in WA). When compared to western NY, eastern WA is characterized by sunnier and drier conditions (seasonal precipitation: 8.3 cm WA, 67 cm NY) with warmer day and cooler night temperatures and earlier average bloom date (29 May WA, 15 Jun NY) but similar growing season GDD accumulation (1300 WA, 1373 NY). Concord vines pruned to 130 nodes and trained to a single trellis wire were compared to investigate the effect of location on juice characteristics (Table 8⇓). Coincidentally, the WA study reported later than average bloom for the 5-year study (1997–2001), while this NY study had earlier than average bloom (1998–2002), leading to no difference in bloom date between the regions. Comparing SW 130 node pruning, WA vines had greater yield because of greater cluster weight caused by more berries per cluster despite lower berry weight on the same number of clusters per vine. SW 130 node vines in NY and WA had similar total leaf area at 650 GDD, and the WA vines had a later harvest date on average. Linear regression analysis of yield and harvest date (Pr > F, 0.0022) from both regions had a parameter estimate of 2.0, indicating a 2-day delay to reach 16 Brix for every 1 kg yield increase per vine. This agrees with the NY observation of a 1.34-day delay in harvest for every 1 tonne yield increase per ha (3 day/ton/acre) in Concord (Bates 2006). Therefore, given a similar bloom date, leaf area, and crop size (despite difference in yield components), there was little difference in Concord Brix accumulation between eastern WA and western NY. It is important to note that Concord bloom in WA is, on average, 17 days earlier than in NY, giving WA the potential to ripen larger crops. If the 2 day/kg relationship is accurate, WA could potentially ripen 8.5 kg/vine more Concord fruit to 16 Brix than NY given average bloom dates. In general, WA juice tended to have lower TA, higher pH, and lower color, but the effect of location was only significant on juice pH over a 5-year period. These trends in juice characteristics at the same juice soluble solids may be a result of how GDD were accumulated (i.e., warmer days and cooler night temperatures in WA) and not the absolute GDD accumulation.
Conclusions
Over a 5-year period, a Concord pruning trial in Fredonia, NY investigated the impact of a 6-fold increase in retained nodes per vine on vine growth, productivity, and fruit maturation on two training systems. In both SW and GDC treatments, less severe pruning led to greater retained nodes, higher yield, and later relative harvest date. Differences in both yield and canopy influenced the exposed leaf area to fruit ratio and the Concord juice soluble solids accumulation rate from veraison to harvest. Retaining more than 250 nodes per vine did not translate into higher yields but continued to increase shoot number and canopy density and further decreased the rate of juice soluble solids accumulation. Given a similar bloom date, the effect of crop load on Brix accumulation was similar for both eastern Washington and western New York. The two-state comparison also illustrates the importance of bloom date on Concord ripening potential in any given season.
Footnotes
Acknowledgments: This research was supported by the Lake Erie Grape Processors, the New York Wine and Grape Foundation, the Viticulture Consortium-East, and National Grape Co-op.
The author thanks Robert Pool for establishing the experimental block in Fredonia; Tom Davenport from National Grape Co-op for research coordination; Tim Weigle for historical weather information; and Richard Dunst, Christine Cummings, Eileen Eacker, Paula Joy, Madonna Struzynski, Ted Taft, Mike Vercant, and Kelly Link for technical support.
- Received February 2008.
- Revision received May 2008.
- Copyright © 2008 by the American Society for Enology and Viticulture