Gibberellin Application Improved Bunch Rot Control of Vignoles Grape, but Response to Mechanical Defoliation Varied Between Training Systems

Control of late-season bunch rots is challenging in grape-growing regions with humid and warm summers, which favor the growth of many grape fungal and bacterial pathogens. Under these weather conditions, grape growers apply up to four or more synthetic pesticides for control of Botrytis cinerea (Weigle et al. 2020), a ubiquitous fungus that causes gray mold of ripening grape berries. However, late-season bunch rots are often a result of non-Botrytis microorganisms that are not controlled by fungicides. To optimize late-season bunch rot management, cultural and other nonchemical methods must be incorporated.

An integrated disease management approach is crucial for grape varieties with compact clusters (e.g., Chardonnay, Pinot noir, Riesling, and Vignoles), which are highly susceptible to late-season bunch rots (Vail and Marois 1991). Compactness leads to overcrowding of berries, which can tear them from the pedicel or lead to berry splitting. These wounds are then easily colonized by opportunistic fungi and bacteria. Compactness can also increase flattened areas of berries, where berry-to-berry contact reduces protective cuticle and epicuticular wax and increases the susceptibility of those berries to fungal pathogen invasion (Marois et al. 1986; Rosenquist and Morrison 1989). Retained floral debris after bloom can act as a substrate for Botrytis spp. and a principal source of inoculum for bunch rot development (Northover 1987). The effect of this debris, made up of dehiscent flower caps, filaments, and anthers, is more pronounced in compact than in loose clusters (Hed et al. 2009).

Several studies examined the potential of nonchemical treatments for loosening clusters on these varieties. For example, Hed and Travis (2005, 2006) and Nail (2009) evaluated bloom applications of horticultural oils (Ultra-Fine, SprayTech, and JMS Stylet) but with inconsistent results. Hed and Travis (2005, 2006) and Molitor et al. (2012) examined removal of the lower portion of the cluster with some success. However, implementation of this treatment is not economically feasible on many varieties. Foliar application of gibberellin (GA) is an inexpensive way to reduce bunch rot predisposition of seedless and seeded grape varieties with compact clusters (Christodoulou et al. 1968; Coombe 1959; Weaver and Pool 1971; Weaver et al. 1962). GA can reduce cluster compactness in different ways. First, it could reduce pollen germination (Weaver and McCune 1960), limiting fertilization, berry set, and cluster compactness. When applied prebloom, GA may also increase the length of the cluster rachis, thereby reducing compactness. Finally, GA application prebloom could stimulate vegetative growth, which might reduce resource allocation to the flowers or increase flower abortion (Caspari et al. 1998). Because of label restrictions, GA cannot be applied on all wine grape varieties and there is paucity of information on its effectiveness in reducing bunch rot. Moreover, there are contrasting findings on the effects of GA applications on berry development, fruitfulness, and yield (Blaha 1963; Hed et al. 2011, 2015; Nagao et al. 1997). Factors such as grape variety, rate, and timing of foliar application might, in part, explain different results.

Another nonchemical method for cluster loosening and subsequent bunch rot reduction is prebloom or trace bloom leaf removal in the fruit zone (Hed and Centinari 2018; Hed et al. 2015; Molitor et al. 2011; Palliotti et al. 2011; Poni et al. 2006; Sabbatini and Howell 2010). The removal of fully developed leaves just before or at the beginning of bloom limits the amount of carbohydrates allocated to inflorescences (Caspari and Lang 1996; Coombe 1959; May et al. 1969). This resource limitation can reduce the number of flowers able to set fruit. Clusters have fewer berries and they are less compact and less prone to late-season fruit rots. Removal of leaves in the fruiting zone is also crucial for improving fungicide spray coverage and microclimatic conditions around the clusters.

Potential loss in revenue due to decreased yield needs to be considered when applying prebloom leaf removal, although economic benefits from fruit with less rot might offset lower overall yield. Another factor that might limit prebloom leaf removal adoption is the cost of application. Fruit zone leaf removal by hand is labor intensive and can be cost prohibitive for interspecific hybrid grape varieties (Vitis spp.) because of their lower grape and wine price compared with the most widely cultivated Vitis vinifera varieties (e.g., https://nygpadmin.cce.cornell.edu/uploads/doc_66.pdf). However, hybrid varieties are largely grown in the Northeast and Midwest United States for their greater cold hardiness (Wolf 2008).

Mechanization could improve cost-effectiveness of prebloom fruit zone leaf removal, when the machinery is to be used to cover vast acreages or can be shared or locally contracted among smaller farms near each other. When compared with hand application, mechanized prebloom leaf removal (hereafter referred to as mechanical defoliation [MD]) had similar reductions in berries per cluster and bunch rot, and similar effects on fruit composition in several V. vinifera varieties (Diago et al. 2012; Hed and Centinari 2018; Intrieri et al. 2008; Tardaguila et al. 2010; VanderWeide et al. 2018, 2020). If this holds true also for Vitis interspecific hybrid varieties, MD could potentially increase adoption of this otherwise cost-prohibitive practice.

To date, it is unclear whether MD could be successfully applied to training systems used to grow hybrid grape varieties. Hybrid grapevines are commonly grown on a high-wire bilateral cordon (HWC) system (also known as Hudson River Umbrella), which allows shoots to grow with their natural orientation without any wire restriction (Reynolds and Wolf 2008). Work on MD has been almost exclusively conducted on varieties grown on vertical shoot position systems (Hed and Centinari 2018; Hed et al. 2015; Intrieri et al. 2008; VanderWeide et al. 2020), which force vertical shoot growth within sets of catch wires, forming a two-dimensional canopy. The HWC system is less labor intensive than vertical shoot position systems but it might be less suitable for MD because it results in a more three-dimensional canopy.

The goal of this study was to optimize recommendations for bunch rot control on hybrid grape varieties grown on two commonly used trellis systems which differ in canopy shape in a humid grape-growing region. We evaluated whether MD and foliar GA application at bloom, alone or combined, enhanced fruit health and composition on Vignoles (Vitis sp.) across 3 years through improvements of cluster architecture and fruit zone environment. We also determined whether vine responses to the treatments varied between training systems. We selected Vignoles because of its significant commercial value and cold hardiness (Bautista et al. 2008; Howell et al. 1991; Kaps and Odneal 2001) but extremely compact clusters highly susceptible to late-season bunch rots (Ferree et al. 2003b; Hed et al. 2009).

Materials and Methods

Experimental site.

The trial was conducted over three seasons (2016 to 2018) in two rows of 51 grapevines (about 122 m long each) within a mature vineyard of Vitis interspecific hybrid Vignoles, at the Lake Erie Regional Grape Research and Extension Center in North East, PA, U.S.A. (latitude 42°23′ N, longitude 79°86′ W). Vine spacing between and within rows was 2.8 by 2.5 m, with a west-to-east row orientation. The soil was classified as a sandy, gravelly loam in the Harborcreek series (USDA 2017). Weather (air temperature and rainfall) data were recorded every 30 min throughout the duration of the study by an onsite weather station (Rainwise Inc., Trenton, ME, U.S.A.). Growing degree days (GDD, base 10°C) were calculated from May to September as GDD = [(maximum daily temperature + minimum daily temperature)]/2] – 10.

Vines of one experimental row were trained to a four-arm Kniffin (4AK) system with two canes (1-year-old wood) or cordons (two-or-more-year-old wood) horizontally tied on each side of the trunk to a wire located at about 1.0 m from the ground and the other two to a wire at 1.3 m from the ground (Reynolds and Wolf 2008). Shoots that emerged from wood on the upper wire were trained upward within a single set of catch wires, and shoots that emerged from wood on the lower wire were trained upward within the catch wires (if buds were oriented upward) or allowed to grow downward (if buds were not oriented upward). Canopies of 4AK-trained vines have a two-dimensional shape. Shoots were hedged once or twice in midsummer to avoid shading and improve pesticide penetration in the fruit zone. Vines on the other experimental row were trained to a single-curtain, bilateral HWC with the training wire positioned at about 1.8 m from the ground. The two rows were separated by one buffer row of the same length.

In all plots, powdery mildew (Erysiphe necator), downy mildew (Plasmopara viticola), black rot (Guignardia bidwellii), and Phomopsis fruit rot (Phomopsis viticola) were controlled with standard fungicides applied to all vines with a Berthoud (Berthoud, Belleville, France) air-blast sprayer (in 2016 and 2017) or a Turbo-mist (Slimline Manufacturing, Penticton, BC, Canada) air-blast sprayer (in 2018). Applications were applied with the sprayer calibrated to deliver 467 liters/ha. Information on fungicides used, including formulation, rate per hectare, and date of application are reported in the supplementary materials (Supplementary Table S1). Botrytis-specific fungicides were applied at 467 liters/ha with a Berthoud air-blast sprayer in 2016 and a Friend covered-boom plot sprayer (Friend Manufacturing Corp., Gasport, NY, U.S.A.) at a pressure of 689.5 and 861.9 kPa in 2017 and 2018, respectively (Supplementary Table S2). Botrytis-specific fungicides were included every year in the trial to mimic a spray program used by grape growers in the region. B. cinerea causes the majority of rots on Vignoles at our site, whereas non-Botrytis rots occur more sporadically and were not addressed chemically to minimize pesticide applications.

Four treatments were established within each training system and applied to three-vine plots (a single post space, approximately 7.3 m long) in a 2-by-2-by-3 factorial design (two levels of MD, yes or no; two levels of GA, yes or no; and 3 years), with four replications for each treatment combination. Treatments were applied on the same vines every year from 2016 to 2018. Data were collected from the center vine and the inner halves of the two outer vines within the plot. MD was applied at trace bloom, when flowers began to open, on 17, 15, and 14 June in 2016, 2017, and 2018, respectively. Defoliation was performed with a tractor-mounted, pulsed-air leaf remover (Blueline Deleafer; Blueline Manufacturing Company Inc., Yakima, WA, U.S.A.) that operates by blowing compressed air into the fruit zone with enough pressure to shatter leaf tissue, while leaving inflorescences relatively undamaged (Hed and Centinari 2018; Tardaguila et al. 2010). The Blueline air sheer system has two pulse-air heads that are mounted at slightly different heights from each other to better accommodate the two fruit zones on the 4AK system. MD was applied to both sides of the canopy, one pass on each side, at a tractor speed of 1.3 kph. The air shear system head was positioned close to the canopy at the fruit zone height to remove leaves at basal nodes one to five. Foliar GA was applied at full bloom to both sides of the canopy, when about 50 to 80% of flowers were opened on 19, 23, and 18 June in 2016, 2017, and 2018, respectively. ProGibb 4% (in 2016 and 2017) or ProGibb LVplus (in 2018) (Valent Biosciences Corporation) was applied at a concentration of 25 ppm with a backpack sprayer at 414 kPa of pressure and approximately 935 liters/ha.

Leaf removal efficiency.

The percentage of leaf tissue removed with MD was measured immediately after the treatment was applied. All leaves from nodes one to five (primary leaves and leaves from lateral shoots present at those nodes) were collected from eight randomly selected shoots from each experimental unit of the noMD noGA vines, and subsequently weighed. In the MD, any leaves (intact and partially shattered) still present at nodes one to five after treatment implementation were collected from eight randomly selected shoots for each experimental unit and weighed. Leaf tissue removed by MD was estimated by subtracting the leaf weight at nodes one to five harvested from shoots in the MD treatment from the weight of the leaves removed at the same nodes from noMD noGA vines. This number was converted to leaf area (square centimeters) by multiplying grams of leaf tissue by 38.1 (1 g of leaf tissue = 38.1 cm² of leaf tissue). This relationship was determined by collecting and weighing a known area of leaf from 50 leaves.

Enhanced point quadrat analysis.

Enhanced point quadrat analysis (EPQA) (Meyers and Vanden Heuvel 2008) was performed around the onset of fruit ripening (i.e., veraison) to evaluate changes in fruit zone canopy porosity, and fruit and leaf sunlight exposure due to MD. We compared plots with MD and no GA to those with no MD and no GA. Point quadrat analysis (PQA) (Smart and Robinson 1991) data were collected on 29 to 30, 22 to 24, and 15 August in 2016, 2017, and 2018, respectively. PQA was used to characterize canopy structure in the fruit zone: a thin rod was inserted horizontally through the fruit zone perpendicular to the vine row at 20-cm intervals and the order in which clusters and leaves were contacted was recorded. These data were used to calculate the percentage of gaps or canopy insertions made with no leaf or cluster contact, leaf layer number, and percentage of interior, shaded clusters and leaves.

Photosynthetic photon flux of photosynthetically active radiation (PAR; micromoles per meter per second) was measured with a ceptometer (AccuPAR LP-80; Meter Group, Pullman, WA, U.S.A.) within 2 h of solar noon on 23, 30, and 23 August in 2016, 2017, and 2018, respectively, under sunny conditions. The in-canopy photon flux values were calculated as the ratio of within-canopy and ambient PAR measured above the canopy. For each experimental unit, the ceptometer was positioned in the canopy interior parallel to the row and at the fruit zone height with the sensors facing upward. For the 4AK system, PQA and PAR measurements were taken at the upper and lower cane level. Four photon flux measurements were recorded for each experimental unit, while one ambient PAR measurement was taken by holding the ceptometer in the row middle, above the canopy. Canopy structure and photon in-canopy flux data were analyzed using Canopy Exposure Mapping Tools, version 1.7 (available free of charge from J. M. Meyers). This software was developed to calculate the number of shade-producing contacts (occlusion layer number), the percentage of sunlight reaching the clusters (cluster exposure flux availability), and the leaves (leaf exposure flux availability) (Meyers and Vanden Heuvel 2008).

Yield parameters, cluster morphology, bunch rot, and fruit composition.

Vines were hand harvested on 20, 21, and 14 to 15 September, of 2016, 2017, and 2018, respectively. The incidence (percentage of clusters affected) and severity (percentage of the cluster’s area affected) of all bunch rots combined (gray mold caused by B. cinerea, non-Botrytis fungal rots [Penicillium, Aspergillus, and Rhizopus spp.], and sour rot [Acetobacter spp., yeasts]) were determined on 40 clusters from each experimental unit. Severity was rated using the Barratt-Horsfall scale (Horsfall and Barratt 1945) and converted to percentage of area infected using Elanco conversion tables (Redman et al. 1969). Determining with accuracy the individual contribution of B. cinerea versus non-Botrytis microorganisms within the same cluster is difficult for a variety with compact clusters such as Vignoles. Because our goal was to evaluate the overall effectiveness of the treatments against crop loss from late-season fruit rots, we did not rate each type of rot separately.

Of the 40 clusters rated for rot, 16 were individually harvested and stored at −20°C for further analysis. For each of the 16 frozen clusters, the number of berries was counted and the length of the main rachis was measured from the first lateral (below the cluster shoulder or wing, if there was one) down to the bottom of the rachis (Hed et al. 2015). Cluster compactness was determined as the number of berries on the main rachis divided by the length of the main rachis and expressed as berries per centimeter. The number of pieces of dehiscent floral debris retained in each cluster, a source of inoculum for bunch rot, was also counted (Hed et al. 2009).

Juice chemistry was determined for all 16 clusters as described by Hed and Centinari (2018). Total soluble solids (TSS) was measured using a refractometer (Reichert Abbe Mark II; Reichert, Inc., Depew, NY, U.S.A.) and pH was determined using a pH meter (Symphony B10P; VWR International, Radnor, PA, U.S.A.). Titratable acidity was measured with a manual titrator (Titronic universal titrator; SI Analytics, Mainz, Germany), where a 10-ml juice sample was diluted with 40 ml of distilled water and titrated with 0.10 N sodium hydroxide to an endpoint pH of 8.0 to 8.2.

Data analysis.

Statistical analysis was carried out using the SAS software package using MIXED procedure (version 9.4; SAS Institute, Cary, NC, U.S.A.). Data were subjected to a three-way analysis of variance for MD (two levels: yes versus no), GA (two levels: yes versus no), years (3 levels: 2016, 2017, and 2018), and their interaction effects. Data from the three seasons were combined over the years when treatment–year interactions were not significant.

Each training system was considered to be a different experiment; the 4AK and HWC vines were in separate rows and it was not possible to randomize them.

Results

Weather data.

Total rainfall during the growing season, May through September, was higher in 2018 (497 mm) as compared with the two previous seasons (2016: 415 mm and 2017: 386 mm). In-season rainfall pattern varied across years (Table 1). In 2016, rainfall from May through July was much lower than during the ripening period, August and September, when fruit is most susceptible to bunch rots. Conversely, in 2017, May and June were wet months but the rest of the season was much drier. The wettest season was 2018 and the month with the lowest precipitation was July. All three growing seasons had greater cumulative GDD than the 20-year average of 1,392 recorded at the site. The only cooler-than-average months during the 3-year trial were the months of May 2016 and 2017 (20-year average of 154 GDDs). Overall, 2017 was a cooler season (only 3% above 20-year average GDD accumulation) than 2016 and 2018 (12 and 19% above average, respectively), the differences occurring mostly from July through September, which were hot during 2016 and 2018 (Table 1).

Table 1. Summary of monthly total precipitation and growing degree day (GDD) accumulation during the 2016, 2017, and 2018 growing seasons at the experimental site

	Total precipitation (mm)			GDD (base 10°C)
Month	2016	2017	2018	2016	2017	2018
May	54.1	144.8	118.6	147	139	233
June	48.8	91.9	95.0	293	301	300
July	68.1	21.3	54.4	401	373	418
August	113.0	59.7	102.1	426	355	406
September	131.1	68.6	127.3	302	266	299
Total	415.1	386.3	497.4	1,569	1,434	1,656

View as image HTML

Effects of MD on fruit zone canopy structure.

The percentage of leaf area removed by MD from the first five basal nodes varied between 36.0 and 46.8% for 4AK-trained vines and from 31.4 to 44.4% for HWC-trained vines (Fig. 1). Leaf removal efficiency was highest in 2018 for 4AK (46.7%) and in 2017 for HWC (44.4%). Overall, MD worked slightly better on the 4AK trellis system, averaging 39.9% removal of leaf tissue in the fruit zone over 3 years, compared with 37.4% on HWC.

Download as PowerPoint

Mechanically defoliated vines (MD noGA) had less dense canopies (e.g., higher percent gaps, lower leaf layer numbers, fewer shade-producing contacts or occlusion layers, and greater leaf sunlight exposure) than noMD noGA vines regardless of training system; however, improvement in cluster sunlight exposure was significant only for the 4AK-trained vines (Table 2). There were few differences in EPQA parameters across years for 4AK-trained vines. There was a significant decrease in percent gaps and a corresponding increase in leaf layer number and occlusion layer number for HWC vines in 2018 compared with the previous 2 years. Furthermore, percentage of interior clusters was lowest, and leaf and cluster exposure flux availability was highest, in 2016. These trends across years for HWC-trained vines may be partially attributed to the shift from a simple cane pruning system in 2016 to cordon or spur pruning in 2017 and 2018.

Table 2. Enhanced point quadrat analysis characteristics of nondefoliated (noMD noGA) and mechanically defoliated (MD noGA) Vignoles vines trained on four-arm Kniffin (4AK) and high-wire cordon (HWC) systems^z

Parameters	Gaps (%)	Leaf layer (n)	Interior leaves (%)	Interior clusters (%)	Occlusion layers (n)	Leaf exposure availability (%)	Cluster exposure availability (%)
4AK
Treatment (T)
noMD noGA	9.6	2.13	58.7	43.3	2.42	48.4	47.1
MD noGA	22.2	1.31	3.4	26.2	1.49	66.4	63.0
P = (T)	0.001	<0.001	<0.001	0.064	<0.001	<0.001	0.020
Year (Y)
2016	18.9	1.74	25.3 a	26.9 b	1.96	54.1	56.3
2017	14.2	1.64	18.5 b	26.7 b	1.90	57.6	54.8
2018	14.6	1.78	22.3 ab	50.7 a	2.00	60.5	53.3
P = (Y)	0.430	0.618	0.056	0.056	0.865	0.184	0.934
P = (T × Y)	0.824	0.646	0.905	0.581	0.819	0.728	0.894
HWC
Treatment (T)
noMD noGA	6.36	2.31	29.6	49.8	2.54	68.2	64.3
MD noGA	13.58	1.91	27.8	37.5	2.13	72.2	63.8
P = (T)	0.007	0.022	0.516	0.199	0.031	0.043	0.907
Year (Y)
2016	13.9 a	1.78 b	25.8	25.2 b	2.00 b	76.6 a	76.6 a
2017	12.8 a	1.97 b	27.0	49.9 a	2.16 b	66.5 b	66.5 b
2018	3.2 b	2.59 a	33.2	55.9 a	2.83 a	64.5 b	64.5 b
P = (Y)	0.003	0.002	0.087	0.034	0.003	0.002	0.001
P = (T × Y)	0.420	0.558	0.398	0.491	0.725	0.932	0.932

^zMeasurements were taken close to the onset of fruit ripening in 2016, 2017, and 2018. MD = prebloom mechanical defoliation and GA = gibberellin. Treatment means within columns followed by different letters are different based on Tukey’s test (P ≤ 0.05).

View as image HTML

Effects of GA and MD on cluster morphology and architecture.

Both GA and MD vines had a lower number of berries and retained floral debris per cluster compared with noMD noGA for both training systems (Table 3). In the HWC system, the significant interaction between MD and GA was caused by the fact that GA more effectively reduced berries per cluster in the absence of MD (99.5 to 58.4 = 41.1 berries) compared with GA combined with MD (76.1 to 54.0 = 22.1 berries).

Table 3. Effects of treatments and years on yield components, cluster architecture, and juice composition for Vignoles vines trained on four-arm Kniffin (4AK) and high-wire cordon (HWC) systems^z

Treatment (T)	GA	Cluster weight (g)	Berry weight (g)	Berries/cluster	Comp (b/cm)	Rachis length (cm)	Debris (n/cluster)	TSS (Brix)	pH	TA (g/liter)
4AK
MD	…
No	No	107.7	1.20	86.5	10.4	8.05	8.55	20.7	3.21	12.7
No	Yes	96.2	1.49	62.4	7.89	7.59	5.88	21.7	3.25	11.9
Yes	No	88.9	1.24	69.6	9.99	6.73	6.36	21.8	3.26	10.7
Yes	Yes	84.6	1.54	52.9	6.52	7.86	4.95	22.4	3.30	11.1
P = (GA)	…	0.065	<0.001	<0.001	<0.001	0.362	<0.001	0.007	0.194	0.581
P = (MD)	…	<0.001	0.254	<0.001	0.150	0.157	0.004	<0.001	0.032	<0.001
P = (MD × GA)	…	0.393	0.991	0.072	0.232	0.034	0.217	0.409	0.863	0.069
HWC
MD	…
No	No	117.1	1.13	99.5	12.20	7.65	9.90	21.3	3.25	12.1
No	Yes	89.5	1.49	58.4	7.89	7.05	6.92	23.4	3.22	10.7
Yes	No	93.2	1.18	76.1	10.44	7.06	7.08	22.2	3.28	10.5
Yes	Yes	80.8	1.44	54.0	7.52	6.97	5.53	22.7	3.25	10.4
P = (GA)	…	<0.001	<0.001	<0.001	<0.001	0.676	0.031	0.011	0.239	0.043
P = (MD)	…	0.002	0.899	<0.001	0.058	0.700	0.049	0.042	0.754	0.022
P = (MD × GA)	…	0.160	0.120	0.012	0.313	0.404	0.153	0.203	0.746	0.113

^zMD = prebloom mechanical defoliation, GA = gibberellin foliar application at bloom, Comp = compactness (berries [b] per centimeter), TSS = Total soluble solids, and TA = total acidity. Treatment means were averaged across years because treatment–year interactions were not significant.

View as image HTML

Some cluster morphology parameters were affected by only one of the treatments either applied alone or in combination with the other. Regardless of training system, GA application, with or without MD, consistently decreased cluster compactness and increased berry weight compared with the other two treatment combinations without GA. MD, applied alone or with GA, always decreased cluster weight but GA effects alone were less consistent between training systems. Other effects on cluster morphology were less consistent across treatments and training systems. For example, MD decreased cluster compactness of the HWC-trained vines only. In the 4AK system, MD was more effective in reducing rachis length alone than in combination with GA.

Treatment effects on fruit composition.

Treatments positively affected fruit maturity across the years (Table 3). Treatment effects on TSS were consistent between training systems, because both GA and MD vines had higher TSS than noMD noGA. MD was effective in reducing total acidity (TA) for both training systems whereas GA was effective for HWC-trained vines only. Effects on fruit pH were not consistent or biologically relevant.

Treatment effects on bunch rot development.

Every year, most rots were in the form of gray mold (B. cinerea) but all rots were combined as a single bunch rot assessment.

Treatments differed in their effects on bunch rot incidence and severity between the two training systems (Table 4). Application of GA reduced bunch rot incidence and severity on both 4AK and HWC-trained vines in all years. MD reduced bunch rot incidence and severity in all years on 4AK. However, the effects of MD on HWC-trained vines were only significant for bunch rot severity, in 2 of 3 years (2017 and 2018).

Table 4. Impact of treatments on the intensity and severity of bunch rot for Vignoles vines trained on four-arm Kniffin (4AK) and high-wire cordon (HWC) systems^z

		Bunch rot incidence (% clusters)			Bunch rot severity (% cluster area)
Treatment	GA	2016	2017	2018	2016	2017	2018
4AK
MD	…
No	No	76.9	89.4	81.9	9.95	15.96	29.2
No	Yes	61.3	69.4	61.3	4.48	9.73	16.8
Yes	No	42.5	64.4	56.9	3.17	6.20	12.5
Yes	Yes	28.1	43.1	47.5	1.30	2.43	5.8
P = (GA)	…	0.033	0.002	0.022	0.024	0.008	0.001
P = (MD)	…	<0.001	<0.001	0.006	0.005	<0.001	<0.001
P = (MD × GA)	…	0.919	0.902	0.328	0.217	0.423	0.207
HWC
MD	…
No	No	68.1	80.6	84.4	6.6	16.1	29.8
No	Yes	35.0	50.0	65.6	2.0	5.5	12.7
Yes	No	64.4	61.9	73.1	3.3	7.2	18.2
Yes	Yes	48.8	34.4	46.9	2.1	2.9	7.8
P = (GA)	…	0.004	0.020	0.034	0.041	0.016	0.008
P = (MD)	…	0.455	0.130	0.132	0.229	0.050	0.073
P = (MD × GA)	…	0.205	0.883	0.688	0.205	0.243	0.432

^zMD = prebloom mechanical defoliation and GA = gibberellin foliar application at bloom. Treatment means were averaged across years as treatments × year interactions were not significant.

View as image HTML

Discussion

Cluster architecture and fruit zone environment can influence bunch rot development, especially during ripening, when fruit is most susceptible. Treatments that beneficially modify these parameters can be integrated into a bunch rot Integrated Pest Management program. Over 3 years, MD and foliar GA application, alone or combined, enhanced fruit health and composition on a grape variety with compact clusters through improvements in cluster architecture and fruit zone microclimate. The fact that bunch rot losses (severity) were lowest in 2016, when rainfall during ripening was highest and fungicide applications for fruit rot pathogens were lowest, suggests the role of other factors in bunch rot development. For example, demethylation inhibitors (tebuconazole and difenoconazole) applied in 2017 and Polyoxin D zinc salt applied in 2018 both have efficacy against fruit rot pathogens. These materials as well as a bloom application for Botrytis spp. (applied in 2018 only) were not included in the 2016 fungicide program. The longer cluster length and lower resulting compactness of clusters measured for all vines in 2016 (data not shown) may account for the lower rot levels in that year, despite weather favorable to the growth of bunch rot microorganisms. For HWC-trained vines, it may also relate to the more favorable fruit zone environment in 2016 compared with 2017 and 2018.

Bunch rot reductions induced by MD and GA were achieved through different mechanisms and effects were not consistent between training systems. An examination of noMD noGA vines alone suggests that incidence and severity of bunch rot over the course of the trial was not affected by training system, although we did not directly compare the two training systems because they were not replicated and randomized within the experimental plot. Vine responses to treatments, however, varied between the two training systems. MD significantly reduced bunch rot on 4AK but results were less consistent on HWC. This was not attributed to superior leaf removal efficiency or effects on cluster architecture, because the average percentage of leaf area removed by MD (4AK: 39.9% and HWC: 37.4%) and the reduction in number of berries per cluster (4AK: 19.5% and HWC: 23.5%) did not vary greatly between the two training systems.

Surprisingly, MD did not reduce cluster compactness (berries per centimeter) of 4AK-trained vines, likely due to the offsetting effect of a shortening of the rachis, as previously reported (Hed and Centinari, 2018). Reduction in bunch rot induced by MD on 4AK-trained vines might be attributed to improved fruit zone microclimate (e.g., higher temperature and solar radiation, greater air circulation, and improved fungicide penetration and cluster spray coverage) by means of fewer leaf layers, interior clusters and leaves, and occlusion layers than noMD vines.

MD affected cluster architecture of HWC-trained vines, including decreased number of berries per cluster and cluster compactness, but did not significantly improve bunch rot incidence, and effects on severity were lower than with GA. Improvements in fruit zone canopy structure were fewer than those reported for the 4AK-trained vines, and there were no differences in percentage of interior leaves and clusters, and cluster exposure flux availability, between noMD and MD vines by veraison. HWC-trained vines tend to have pendulous vegetative growth and it is possible that shoots growing downward shaded the fruit zone after MD was applied, explaining the lack of significant differences in cluster sunlight exposure. Together, these results suggest that MD benefits on bunch rot relate mainly to improved fruit zone microclimate (4AK) and, to a lesser extent, to modified cluster architecture (HWC).

Benefits of MD were extended to greater fruit maturity (higher TSS and lower TA); greater TSS of the MD vines compared with noMD noGA vines may relate to greater sunlight exposure of leaves (fewer leaf layers and occlusion layers), which increases their carbon assimilation capability (Keller 2015). A more porous canopy with a greater percentage of gaps suggests greater air temperature in the fruit zone (English et al. 1989), which can cause higher malic acid respiration and then lower TA at harvest. Application of GA also improved TSS for both training systems. Although GA vines were not defoliated, the increase in fruit maturity might be explained by lower crop load values (crop mass: vegetative mass ratio) of those vines.

Because training system appears to play an important role in the efficacy of MD, it is recommended that growers train their vines to more upright, two-dimensional trellis systems, and avoid the HWC system, if they want to maximize the returns from this technology. Our data also suggests that growers can apply MD for bunch rot control later in the season, around or after fruit set, as reported by other studies conducted in the eastern United States (Hed and Centinari 2018; Hickey and Wolf 2018). By applying MD at or after fruit set, potential yield reduction caused by prebloom application could be avoided while maintaining bunch rot control and fruit composition benefits.

Overall, GA had greater impact on cluster architecture than MD and the vine response to GA was less dependent on training system. The superior reduction in cluster compactness compared with MD can explain why GA vines more consistently lowered bunch rot severity every year for both training systems, despite the offsetting effect of producing heavier berries. The increased berry weight might be a compensatory result from lower number of berries per cluster, even lower than MD (Ferree et al. 2003a), or a direct consequence of GA application, as reported in seedless grape (Casanova et al. 2009; Dokoozlian and Peacock 2001).

Collectively, our results suggest that GA applications could be more effective than MD at reducing rot levels on HWC-trained vines. Although not estimated in this study, applying GA is much less expensive than removing leaves in the fruit zone either manually or mechanically. However, the current ProGibb label limits wine grape use to a single, prebloom application on 16 varieties of V. vinifera, and we are not aware of any formulation of GA that is currently labeled for use on Vignoles or any other seeded hybrid wine grape varieties.

Both MD and GA, implemented alone or together, reduced floral debris retention within the cluster when compared with noMD noGA vines. Dehiscent floral debris that remain within clusters after bloom can act as a substrate for the growth of Botrytis spp. within the cluster and play a role in the development of bunch rots, as previously reported (Fedele et al. 2020; Hed et al. 2009; Northover 1987; Wolf et al. 1997). Looser clusters (primarily by GA) and clusters exposed to better air circulation (by MD) can facilitate floral debris removal by weathering. However, in our study, we did not find a clear relationship between the amount of floral debris, cluster compactness, and bunch rot.

When we examine average cluster weight as a proxy for yield, the reductions in bunch rot induced by MD and GA did not always result in more, healthy (i.e., without rot), “sellable” fruit. For example, if we estimate the weight of fruit loss to bunch rot as absolute value (grams) rather than percentage, the 3-year average amount of healthy fruit of the noMD noGA vines is greater than that of the treatments for both training systems. This is largely because both GA and MD reduced cluster weight, an effect that could be at least partially remedied by leaving more buds at dormant pruning. When this analysis is broken down into individual years, the percentage of fruit with no rot increases in the treatments relative to the noMD noGA vines, because annual bunch rot pressure increases. Although applying MD or GA did not result in more sellable fruit in 2016 and 2017, similar or higher amounts of healthy fruit were estimated for MD and GA, either alone or combined, relative to noMD noGA in 2018, regardless of training system.

Acknowledgments

We thank the Pennsylvania Wine Marketing and Research Board and the Penn State College of Agricultural Sciences, for financial support for this project; W. Schultz, S. Paulson, and M. Wheeler for field and laboratory technical support; and R. Marini and M. Campbell (Penn State University) for their critical reading of the manuscript.