Plant material and study location.

This study was conducted in a non-irrigated commercial vineyard in Lexington, KY, with own-rooted ‘Vidal blanc’ planted in 2001 on a Maury silt loam soil [a fine, mixed, active, mesic Typic Paleudalf (U.S. Department of Agriculture, 2011)] with 2.4 × 4-m (vine × row) spacing. Row orientation was north to south. The vines were trained to a 1.8-m high bilateral cordon and were supplied with 38.5 kg nitrogen/ha annually. Pests were managed using an integrated pest management program according to University of Kentucky guidelines (Bordelon et al., 2006). The experiment was set up as a three × three factorial arrangement of pruning formula and cluster thinning treatments in a completely randomized design with four replications. Each experimental unit consisted of three vines. The same vines were used for all treatments in all years of the study. The treatments described below were applied beginning in 2005. Data collection started in 2006 to allow the vines to acclimate to the treatments described below.

Balanced pruning treatments.

The dormant canes were long-pruned to five nodes each to an upper limit of 65 nodes per vine and prunings were weighed on a top-loading scale. The number of nodes to be left on each vine was then determined by the weight of prunings removed. Three levels of balanced pruning severity were applied: 20 + 10, 30 + 10, and 40 + 10. The first number refers to the number of buds left for the initial 454 g of pruned biomass and the second number to the number of buds retained for each additional 454 g of biomass. Pruned biomass was also divided into dead and living canes and weighed separately. Separation into the two categories was determined by visual inspection of wood and bud color. Vine size was calculated as total biomass divided by within-row spacing or 2.4 m.

Cluster thinning treatments.

Three levels of cluster thinning severity were established at the modified Eichhorn-Lorenz scale stage 31 (3- to 5-mm berry size) in all years. The number of clusters per shoot and per vine was counted before thinning in both years. The cluster thinning treatments consisted of retaining one (thinned to cluster on first count node), two (thinned to clusters on first and second count nodes), or all clusters per shoot.

Phenology of key developmental events.

Air temperature data from the University of Kentucky Agricultural Weather Center were used to calculate the mean daily temperature. GDD during each year were calculated starting 1 Apr. using a base temperature of (10 °C) with no upper limit. This permitted the description of phenological stages by both calendar date and GDD (Table 1).

Table 1.

Phenological progression of own-rooted ‘Vidal blanc’ in 2006 and 2007, in Lexington, KY.

View Table

Shoot, cluster counts, and canopy architecture.

Two weeks after bud burst, shoots per vine and clusters per shoot and per vine were counted to measure canopy architecture variables as reported elsewhere (Kurtural et al., 2013). Briefly, the number of count shoots (borne from buds counted in the formula and retained at pruning greater than 5 mm distal to the base of the cane), non-count shoots (borne from basal buds less than 5 mm distal to the base of the cane or on wood older than 1 year and not counted in those left after pruning), and total shoots (count + non-count shoots) retained per vine were counted and recorded at this time. Indicators of canopy architecture measurements like count shoots per hectare, distance between shoots, and leaf layer number were measured as described by Smart (1985).

Yield components and berry composition.

Fruit yield and cluster numbers for all treatments were measured on a single vine within each experimental unit during harvest. During both years, yield was divided into marketable and unmarketable fruit (fruit displaying greater than 30% disease damage or insect herbivory). However, in both years, there was no unmarketable fruit. All treatments were harvested on the same date. Harvest date was determined by the cooperating winery based on a target juice pH of 3.5. Average cluster weight was calculated by dividing yield by cluster number. The RI for each treatment was then calculated by dividing the yield per vine by the dormant pruning weights collected.

A random 100-berry sample was collected at harvest from each replicate, placed in a polyethylene bag, stored on ice, and analyzed within 24 h. The samples were crushed by hand and the juice was placed in 100-mL beakers. A 5-mL portion of each sample was used to determine the percent total soluble solids measured as °Brix using a digital refractometer (Spec Scientific Ltd., Scottsdale, AZ) The juice pH was determined with a glass electrode and a pH meter (Model AR15; Fisher Scientific, Pittsburgh, PA). The acidity of each sample was determined by titrating to pH 8.2 with 0.1 N NaOH and expressed as mg·L^–1 tartaric acid (TA).

Node maturity and lethal injury to primary buds (LT₅₀) analysis.

Node maturity was derived as the ratio of nodes displaying periderm browning vs. those that did not for each experimental unit. Data were collected by determining the number of buds with periderm browning from 10 randomly selected shoots per node per experimental unit. Measurements were made on 25 Sept. 2006 and on 9 Sept. 2007 before a killing frost and on 11 Nov. 2006 and 24 Nov. 2008 after the first killing frost.

The lethal temperature to kill 50% of the primary buds (LT₅₀) was determined by subjecting cuttings from dormant canes to a controlled freezing stress as described by O’Daniel et al. (2012) in early November of 2006 and mid-February of 2008. Dormant canes (6 to 10 mm diameter) were randomly selected from pooled samples from all vines of a given treatment stored on ice at –4 °C. Selected canes were pruned to three buds and bundled in groups of four cuttings for four replications and nine temperature evaluation points for each treatment [16 cuttings (48 buds total)/pruning treatment/temperature]. Progressive freeze treatments began at 0 °C and ended at –40 °C. Temperatures in the chamber were lowered by –5 °C progressively for a 2-h ramp time and then held constant at that temperature for a 1-h soak time. Samples from each temperature treatment were removed from the freeze chamber every 3 h at the completion of each soak time and allowed to thaw for 48 h at 2 °C before evaluation. Buds were dissected by making lateral cross-section cuts across the dormant bud with a razor blade, exposing the primary bud, which was examined for oxidative browning, an indicator of lethal injury (Stergios and Howell, 1972). Numbers of live or dead primary buds were recorded. The temperature at which 50% lethal injury (LT₅₀) occurred was estimated by fitting a sigmoidal nonlinear regression model {logistic 4-P; Y = a +b/[1 + Exp(c-d*X)], where Y = % dead buds and X = temperature} to the data from each treatment using SAS Version 9.3 (SAS Institute Inc., Cary, NC) and then computing the predicted value from the regression equation. Estimated LT₅₀ values were compared among treatments by two-way analysis of variance (ANOVA) as described below.

Statistical analyses.

Data for all parameters were tested to verify if the assumptions of ANOVA were met using the Shapiro-Wilk test. Data that failed to meet the assumptions of ANOVA were log10 transformed and analyzed using the generalized linear model (GLM) procedure in SAS (Version 9.3; SAS Institute Inc.). The significance level was set at α = 0.05, and means were separated using Tukey’s honestly significant difference test. Pruning formula and cluster thinning treatments were then tested for polynomial trends across treatment levels using the GLM procedure of SAS. For the transformed data, when the ANOVA showed significant differences, the mean separation test was conducted on the transformed data but non-transformed means were presented for ease in discussion. Interactions between year and treatments were tested and, whenever these interactions were significant (P < 0.05), analysis was conducted separately for each year.

Canopy architecture.

There was an interaction of year and pruning on the number of count shoots per hectare (Table 2). There were fewer count shoots per hectare in 2007 compared with 2006. Generally, as the severity of pruning increased (fewer nodes retained per vine), there were fewer shoots per hectare. The distance between shoots exhibited an interaction of year by pruning formula. The shoots were further apart in 2007 compared with 2006. Generally, as the severity of pruning decreased (more nodes retained per vine), distance between shoots became closer. There was no effect of treatment or year on the leaf layer numbers. The cluster thinning treatments did not affect any of the canopy architecture variables measured.

Table 2.

Effect of pruning formula and cluster thinning on canopy architecture on own-rooted ‘Vidal blanc’ grapevine in 2006 and 2007 in Lexington, KY.^z

View Table

Yield components.

In 2006, the pruning formula treatments did not affect the number of clusters harvested, berry (data not shown) or cluster weight, or yield per vine (Table 3). However, in 2007, there was a linear response to the pruning formula applied where the number of clusters harvested increased with the decrease in the severity of pruning formula. The berry and cluster weight of ‘Vidal blanc’ did not respond to pruning formula treatments in 2007. There was a strong, linear response of clusters harvested per vine to cluster thinning in both years. As the severity of cluster thinning decreased (more clusters retained per shoot), there were more clusters harvested. Cluster thinning did not affect berry weight, but pruning formula interacted with cluster thinning on cluster weight in 2007. The combination of the 20 + 10 pruning formula and retention of one cluster/shoot resulted in the greatest cluster weight compared with the combination of the 40 + 10 pruning formula and the two+ clusters/shoot cluster thinning treatment, which had the smallest cluster weight. Yield per vine declined linearly as the severity of cluster thinning increased in 2006 and 2007. There was no effect of experimental year on yield, but, in both years of the study, applying the one cluster/shoot treatment almost halved the yield per vine compared with leaving two+ clusters/shoot.

Table 3.

Effect of pruning formula and cluster thinning on yield components of own-rooted ‘Vidal blanc’ grapevine in 2006 and 2007 in Lexington, KY.^z

View Table

Fruit composition.

There was an effect of experimental year on total soluble solids (TSS) of ‘Vidal blanc’ at harvest where it was slightly higher in 2007 than in 2006. However, neither pruning formula nor cluster thinning affected TSS or juice pH in either year (data not shown). The TA was affected by an interaction of year and cluster thinning. In general, the TA was higher in 2007 than in 2006. Furthermore, leaving two clusters/shoot in 2007 resulted in the highest TA at harvest compared with all year and any other cluster thinning treatment.

Node maturation, midwinter primary bud cold hardiness, pruning weight, and Ravaz Index.

There was an interaction of year and cluster thinning on node maturation when measured at pre-killing frost (Table 4). Retaining one cluster per shoot increased mature nodes per shoot in 2007 compared with all year and cluster thinning combinations. There was no effect of pruning or cluster thinning on node maturation per shoot when measured post-killing frost (data not shown). The lethal temperature to kill 50% of the primary buds (LT₅₀) when measured in winter was affected by the interaction of year and cluster thinning. Generally, primary buds displayed greater ability to withstand colder temperatures in 2006 compared with 2007. In 2006, there was no effect of pruning formula on LT₅₀ of primary buds of ‘Vidal blanc’, but LT₅₀ decreased linearly with the increase in cluster thinning severity. In 2007, there was a significant interaction of pruning formula and cluster thinning on the LT₅₀ of primary buds. The combination of 20 + 10 pruning formula and bearing one cluster/shoot had the lowest LT₅₀ compared with all other treatment combinations. Conversely, the combination of the 40 + 10 pruning formula and retaining two+ clusters/shoot had the highest LT₅₀ compared with all other treatment combinations.

Table 4.

Effect of pruning formula and cluster thinning on percent mature nodes post-harvest, midwinter primary bud cold hardiness, mature wood pruning weight, pruning weight, and Ravaz Index of own-rooted ‘Vidal blanc’ in 2006 and 2007 in Lexington, KY.^z

View Table

There was an experimental year and pruning, and experimental year and cluster thinning, effect on mature wood and dieback wood pruning weight of ‘Vidal blanc’ (Table 4). The mature wood pruning weight increased by approximately two times in 2007 compared with 2006. Conversely, the dieback wood pruning weight decreased by approximately three times in 2007 compared with 2006 (data not shown). There was a linear decrease in mature and dieback wood weight in both years of the study with decreasing pruning severity. In both years of the study, the two clusters/shoot treatment had the greatest mature wood pruning weight. However, the dieback wood pruning weight was not influenced by the cluster thinning treatments applied. The pruning weight (mature wood + dieback wood weight) was unaffected by the pruning formula treatments. The pruning weight, however, declined linearly with the increase in the number of clusters retained per shoot in both years of the study. The interaction of experimental year and pruning formula affected the RI of ‘Vidal blanc’. The RI was slightly lower in 2007 with the 20 + 10 pruning formula and with the 40 + 10 pruning formula in 2006. The RI was affected by the cluster thinning treatments in both years of the study, increasing linearly with the decrease in the severity of cluster thinning. Leaving two+ clusters/shoot resulted in the greatest RI in both years of the study.

Canopy architecture.

The study site was unseasonably cool in 2006, accumulating less than 2200 GDD. Furthermore, the study area also witnessed a historic freeze event in the spring of 2007 when air temperatures dropped below –5.5 °C for five consecutive nights on 7 Apr. through 12 Apr. 2007 preceded by an early, warm spring, which led to budbreak and leaf development a week before the dates in 2006. Therefore, there was significant shoot damage before the budbreak that generated the components of yield from the budbreak date reported in Table 1. Nevertheless, the number of count shoots per hectare and the distance between the shoots along the cordon responded in both years linearly to the pruning formula treatments as reported by Smart (1985) and Terry and Kurtural (2011). Severe pruning reduced the number of shoots exposed per hectare in both years as observed by Kurtural et al. (2006). Consequently, the distance between shoots along the cordon was further apart in the more severe pruning treatment, similar to that reported by Geller and Kurtural (2013). The effects of year and the interaction of year and pruning formula are explained by the freeze event that occurred in April of 2007. It was not surprising to see that distance between shoots was further apart in 2007 compared with 2006 as a consequence of secondary buds and non-count buds bursting, resulting in fewer count shoots per hectare in 2007. Although shoot density varied considerably depending on the severity of pruning, the leaf layer numbers were unaffected. This was a result of vegetative compensation occurring in warm climates where dormant pruning alone is a rough regulator of canopy architecture (Kurtural et al., 2013). The lack of a horticultural response to number of leaf layers likely stemmed from more sunlight available to the non-count positions on the cordon, generating many laterals and non-count shoots and a similar leaf area with the severe pruning formula when compared with the less severe pruning formula, noted by Bernizzoni et al. (2011) and Geller and Kurtural (2013).

Yield components.

Balanced pruning alone did not result in adequate and consistent yield control in a interspecific hybrid as others reported (Fisher et al., 1977; Kurtural et al., 2006) unless followed by cluster thinning. Balanced pruning provided yield control only when count shoots per hectare varied by 16% for each additional 10 nodes retained per fraction of 454 g of dormant pruning weight. Consequently, the number of clusters per vine increased by ≈10% among pruning formula in a linear response to the reduction of pruning severity. However, the berry weight was unaffected by balanced pruning in either year of the study. The lack of response in berry weight to pruning formula resulted in no effect on cluster weight in either year as reported elsewhere (O’Daniel et al., 2012). Therefore, balanced pruning alone was not sustainable for ‘Vidal blanc’. Similar conclusions were reported for ‘DeChaunac’ and ‘Chambourcin’ where unthinned vines have often suffered from low pruning weight retention and winter injury (Fisher et al., 1977; Kurtural et al., 2006).

In 2006 and 2007, the number of clusters retained on the vines explained 73% and 77% of the total variance in yield, respectively. This finding was corroborated by previous observations (Kurtural et al., 2006) of interspecific hybrids. Therefore, yield was primarily affected by the numbers of clusters retained on the vine. There was an effect of experimental year on %TSS and TA at harvest, likely a result of the late development of fruit resulting from the freeze in 2007.

Fruit composition.

Increasing canopy shade raises the pH and potassium in must and wine (Smart, 1985; Smart and Robinson, 1991). Canopy shading also leads to lower sugar accumulation and uneven ripening in clusters. Although canopy architecture was affected by pruning formula as described previously, there were no treatment effects on fruit composition. In a 2-year study of ‘Traminette’ on the same training system conducted in southern Illinois (Dami and Ji, 2008), which has a similar climate to Kentucky, °Brix was higher, pH was lower, and TA was intermediate to the levels observed in Kentucky. Because harvest was in early September both years as a result of site restrictions imposed by the winery for this study, berry ripening was perhaps limited, contributing to the differences with the other studies of interspecific hybrids.

Node maturation and primary bud cold hardiness.

Low temperatures, especially midwinter advective freezes, are the limiting factor in expansion of acreage and grapevine productivity for several grape-growing regions in the midwestern United States (Howell, 2001; O’Daniel et al., 2012). The cold midwinter temperatures in Kentucky are often the determining factor in deciding which cultivars to plant (Kurtural et al., 2008). The lack of a response by the pre-killing frost percent mature nodes to pruning formula application was previously reported for interspecific hybrids in Kentucky (O’Daniel et al., 2012). The present study reports an increase in percent mature nodes in both years when measured at pre-killing frost with decreasing crop level. This observation is important because practices that control crop size over multiple years would favor resistance to cold temperatures as reported by others (Howell et al., 1987; O’Daniel et al., 2012). When percent mature nodes were measured after the killing frost, there was a lack of treatment effect. However, whether pre- or post-killing frost, the percent mature nodes were above 60% in both years of the study. This was an artifact of the long growing season in Kentucky; periderm browning should not be an issue for growers of this cultivar if cluster thinning is applied and proper periderm browning is evident before occurrence of the killing frost in warm regions. Taken together with the LT₅₀ data, cropload management with pruning formula and cluster thinning was beneficial for primary bud cold hardiness of ‘Vidal blanc’. The several degree improvements in the primary bud cold-hardiness of ‘Vidal blanc’ with retention of one or two clusters per shoot would be important for the region. Previously, Howell et al. (1987) recommended a combination of 15 + 10 pruning formula while retaining one cluster per shoot. Our results, although they confirm this finding, also revealed that a higher bud load, using a 30 + 10 pruning formula with two clusters per shoot resulting in a larger crop, can be carried without much detriment to primary bud cold hardiness or percent mature nodes.

Pruning weight and Ravaz Index.

It was evident that fruit production depressed vine capacity as ‘Vidal blanc’ adjusted to the treatments (Keller, 2010). Perhaps the most interesting finding of this trial was the quadratic response of mature wood weight and pruning weight per meter of row to cluster thinning. Pruning weight would be expected to decline linearly as fruit production depressed vine capacity (Geller and Kaan Kurtural, 2013; Kurtural et al., 2006; Terry and Kurtural, 2011). Conversely, in both years of the study, retaining the correct number of clusters (two clusters/shoot in this case) improved mature pruning weight and total pruning weight. Furthermore, in both years, pruning weight per meter of row, although never below the recommended rate of 0.30 kg·m^–1 of row (Kurtural et al., 2006), was the greatest at two clusters/shoot. Other interspecific hybrids such as ‘Chambourcin’, ‘Seyval blanc’, and ‘Traminette’ have shown that vine size greater than 0.59 kg·m^–1 of row was a sign of excessive vine growth (Kurtural et al., 2006; O’Daniel et al., 2012; Reynolds and Wardle, 1994). This indicated that a pruning weight per meter of row needs to be between 0.30 kg·m^–1 and 0.59 kg·m^–1 of row, and this range was achievable at two clusters/shoot without any detrimental effects on pruning weight.

An ideal RI range of 5 to 10 kg yield per kilogram of pruning weight may be optimal while maintaining vine capacity (Kliewer and Dokoozlian, 2005; Terry and Kurtural, 2011) in a warm climate. The linear increase in RI seen in this trial was not surprising. As more crop was retained, RI would undoubtedly increase (Geller and Kaan Kurtural, 2013). However, the optimum RI value of ≈10.0 kg·kg^–1 was only achieved with the combination of 30 + 10 pruning formula and two clusters/shoot treatment in both years of the trial without any detrimental effects on primary bud cold hardiness or a severe decline in pruning weight per meter of row. This finding, coupled with the longer and warmer season witnessed in central Kentucky, indicated that a larger crop could be retained in this region compared with the Michigan recommendation of 15 + 10 pruning formula with one cluster per shoot (Howell et al., 1987).