Effect of frost damage and pruning on current crop and return crop of Pinot Noir

Abstract

In October 2006, much of the wine-growing area in Tasmania was affected by a series of some of the worst frost events in more than 30 years. Widespread damage left vineyards with blackened shoots and the prospect of a considerably smaller crop, with later maturing bunches from secondary buds contributing to poorer quality wine. In a commercial, spur-pruned Pinot Noir planting, treatments intended to encourage and manipulate secondary bud-burst are imposed and effects on yield recorded. Treatments are imposed 10 days after the frost and include: (i) an untreated control (control) with all damaged tissue left in place; (ii) frost-damaged tissue removed (light pruning); (iii) frost-damaged tissue removed and original spur trimmed back to one bud (medium pruning); and (iv) original (damaged) shoot removed back to compound bud on the spur (heavy pruning). Pruning treatment responses for season 2006–07 show that heavy pruning reduces the current crop with no useful gain in uniformity of ripening. The medium and heavy pruning treatments also reduce pruning weights at the end of the season and all post-frost pruning treatments result in a smaller inflorescence primordia size in dormant buds dissected at the beginning of commercial pruning. In the 2007–08 vintage, the untreated control and the medium pruning treatment have significantly lower bunch numbers than the other treatments. There is also a significant effect of terrain elevation on total yield and the number of bunches in the frost year, with increasing damage lower in the inversion. This gradation in damage does not have carry-over effects into the second season. The results indicate that none of the pruning treatments tested have clear benefits for current or subsequent season production, and that a prescriptive approach to pruning should be avoided when the level of frost damage is inconsistent across vines.

Keywords:

• Vitis vinifera
• frost damage
• management after frost

Introduction

Spring frosts are a significant hazard to vineyards in most cool climate regions. Damage to young shoots, as a result of spring frost, can severely reduce the current crop, as well as reduce the crop of following seasons. Direct impacts on flowering and vegetative growth in both the current and subsequent year may influence longer-term vine balance, with the indirect effects of a single frost damage event potentially continuing into a third year (Trought et al. 1999).

The mechanism of frost damage is still not completely clear, but there is a fairly general consensus that extra-cellular ice formation results in rapid removal of water from cells and severe dehydration results. The consequent disruption of membrane function and enzyme systems causes widespread necrosis of the affected tissue and results in the typical ‘frost burn’ symptom (Allard et al. 1998).

Although −2^oC is generally taken as the critical temperature for damage to non-dormant tissue across a wide range of temperate crop species, tissues vary in susceptibility and the potential damage in any season depends on the time of the frost and the phenological stage of the crop (see review by Rieger 1989). It has been suggested that the management response to a damaging frost should take into account stage of development, severity of the damage, length of the growing season, and age and cultivar of the vines (Trought et al. 1999; Creasy et al. 2002; Trought et al. 2003). Formal research on field frost damage or response to damage is invariably limited by difficulties in planning for a chance event and providing untreated undamaged control plants. The discussion papers by Trought and co-workers are two of the very few publications on field management of frost-damaged plants of any species.

The expanding cool climate wine area in Tasmania is not subject to regular frost. A few vineyards on high-frost risk sites have installed various protection systems but growers generally rely on site selection as a primary means of avoiding serious losses. In spite of this, vineyards were hit by a series of frost events between 16 and 22 October 2006, as were vineyards in other cool climate areas of southern Australia. Most frost-prone vineyards, with temperature recording systems, reported temperatures below −2^oC on the night of 16 October, with a second similar event on 21 October. On both nights, below zero temperatures were recorded in several vineyards, which had never previously suffered frost damage and on some susceptible sites, temperatures were too low for protection systems to be effective. Damage was widespread but varied in severity from minor tip burn to an almost complete removal of both primary and secondary buds, evident as large numbers of ‘blind’ buds later in the season.

With a very short lead time, this trial was planned and developed in an attempt to answer questions from growers about what, if anything, they should do to minimize both immediate and longer-term effects on fruit yield and quality. Lack of information and experience, combined with an uncertainty of the fruitfulness of secondary buds, the reliability of fruitset in flowers from secondary buds, disease threat and the likelihood of ripening the crop, meant that few questions could be answered confidently.

Possible management strategies for reducing the damage by frost used by growers in Tasmania are the same as those discussed by Trought et al. (2003) and Creasy et al. (2002), who suggested that management options included taking no action, removing only dead material and removing all (including the green) shoots. The current study aimed to investigate the effect of such strategies on frost-damaged vines in Tasmania, based around varied levels of pruning severity.

Materials and methods

A 20-year-old spur pruned commercial Pinot Noir (clone D5V12) vineyard (42°45′S, 147°23′E) near Richmond, in southern Tasmania, was used for the trial. The vineyard ‘falls’ towards the north-east with a uniform 9% slope, ranging in terrain elevation from 60–100 m. Inter-rows are maintained as adventitious pasture. The area had never before suffered significant frost damage, and there was no temperature-recording system in the vineyard. Vines ranged in phenological stage from swollen bud to shoots of up to 25 cm, which corresponds with Eichorn Lorenz Scale EL03 to EL15 (Mullins et al. 1992). Rows were approximately parallel with the slope with the worst affected vines at the lower end, reflecting a typical atmospheric inversion temperature profile and hence damage (Perry 1998). The worst-affected vines at the base of the slope were showing severe damage with around 50% necrosis of shoots, including necrosis of inflorescences, and lesser-affected vines at the top of the slope exhibited moderate damage with around 30% necrosis of shoots and inflorescences (Fig. 1).

Fig. 1  Frost damage on 20-year-old Pinot Noir vines (clone D5V12) in the Coal River Valley, Southern Tasmania. This level of damage was consistent with 50% damage, with necrosis of shoots, leaves and inflorescences

Three pruning treatments and an untreated control were imposed on 31 October 2006, 10 days after the last frost, on five-vine plots: (i) an untreated control (control) with all damaged tissue left in place, (ii) frost-damaged tissue removed (light pruning); (iii) frost-damaged tissue removed and original spur trimmed back to one bud (medium pruning); and (iv) original (damaged) shoot removed back to the compound bud (heavy pruning) as shown in Fig. 2. The trial was set up in a single row as a nine replicate, randomized complete block to account for variation in damage according to expected inversion temperature.

Fig. 2  The untreated control and three pruning treatments imposed after frost damage occurred

Vine parameters, including shoot number and inflorescence number were measured prior to harvest. Vineyard manager's notes were taken for each treatment, with particular attention to dates of flowering and veraison. Total yield, bunch number, bunch size and berry number were measured at harvest in both the year of damage (9 March 2007) and the subsequent year (18 March 2008). Total bunch number was partitioned into bunches that were clearly ripe, soft and coloured (ripe), and clearly green, unripe and hard (green). Pruning weight and bud fertility were measured at pruning in the year of damage. Bud fertility was measured by dissecting the third bud from the fourth shoot from the crown on the left arm, from two vines per plot at commercial pruning. Using a dissecting microscope and transverse sectioning of the bud, the number of inflorescence primordia were recorded, and each primordia was allocated to either small (<500µm) or large (≥500µm) size categories. For determining the size of inflorescence prinmordia, a sample of buds was taken, and the apex and intact inflorescence primordia were removed by dissection under a light microscope (Nikon SM2-1B) and fixed in 2.5% gluteraldehyde. These were then dehydrated using a water–acetone–carbon dioxide series, critical point dried (Polaron CPD E3000), mounted on aluminium stubs with carbon tabs and sputter coated (Edwards 5150B) for examination under an environmental scanning electron microscope (Philips XL30 FEG). Using the general linear models package of SPSS, results were subject to analysis of variance and treatment effects separated using least significant difference (calculated at P=0.05) after Steel and Torrie (1980). For data collected in both years, separate analyses of variance were carried out for the two sets of data.

To explore the impact of frost severity with regard to relative terrain elevation, we fitted a Generalized Linear Model (McCullagh & Nelder 1989) using the R statistical environment (R Development Core Team 2009). The response data were bunch numbers, so a Poisson distribution with a logarithmic link function was chosen to account for stochastic variation. That is, we fitted the model:

where y _i is the observed bunch number, x _ii is the pruning treatment applied and x _2i is the elevation of the ith vine, i=1,2,…, n, to estimate the coefficients ß_0, ß₁ and ß₂. We used R to assess the degree of overdispersion in our initial model, and applied the correction suggested by Gelman and Hill (2007). Critical appraisal of the final fitted model using graphical regression diagnostics found the model satisfactory.

Results

For bunch number there was a significant treatment effect in both years, as shown in Table 1. In the 2006/07 season, there was a significant (P=0.005) treatment effect on the total number of bunches and the number of ripe bunches, with the treatment where vines were pruned back to a compound bud (heavy pruning) having significantly fewer ripe bunches than the other treatments. This same treatment also resulted in a significantly (P=0.021) lower total yield at harvest, with the mean yields being control 673 g (SD=307.43), light pruning 780 g (SD=332.46), medium pruning 672 g (SD=369.89) and heavy pruning 499.78 g (SD=117.96). There were no significant treatment effects on the number of green bunches in 2006/07.

Table 1  Mean number of bunches per vine for the three pruning treatments —frost-damaged tissue removed (light pruning); frost-damaged tissue removed and original spur trimmed back to one bud (medium pruning); and original (damaged) shoot removed back to compound buds on the spur (heavy pruning) —and an untreated control (control) in 2006/07 and 2007/08

Treatment	Total number bunches 2006/07	Total number bunches 2007/08	Ripe bunch number 2006/07	Green bunch number 2006/07
Control	17.42	29.5	9.60	7.82
Light pruning	18.50	38.4	10.27	7.44
Medium pruning	16.16	32.6	8.44	7.71
Heavy pruning	12.69	37.4	5.76	6.93
LSD (P=0.05)	3.47	6.28	2.55	ns
ns=Not significant.

In the following season (2007/08), the total number of bunches was significantly higher in the treatments with only damaged tissue removed and with the damaged shoot removed back to the spur. In this year the total fruit yield was also significantly higher in treatments comprised of light pruning and heavy pruning, with the mean yields being control 3894 g (SD=2064.1), light pruning 4124 g (SD=1559.4), medium pruning 3394 g (SD=2032.1) and heavy pruning 4650 g (SD=2135.6).

Relative elevation was found to be a highly significant covariate in the frost year 2006/07 (P<<0.01). The fitted model (1) is shown in Fig. 3a. The model predicts an 8.4% increase in bunch number per metre of elevation with a 95% confidence level for this estimate of [1.062, 1.106]. By contrast, Fig. 3b shows bunch number versus elevation for the same vines in a typical year in which no frost was experienced, 2007/08.

Fig. 3  (a) Fitted model of actual bunch number per vine plotted against relative elevation in the frost year 2006/07. The solid line is the expected bunch number under the control treatment, which is not significantly different from the light and medium pruning treatments. The perforated line is the expected bunch number under the heavy pruning treatment (see text). (b) Actual bunch number against relative elevation in the subsequent frost-free season, 2007/08. Elevation is not a significant predictor for bunch number in that season. The solid line is the expected bunch number under the control treatment, which was not significantly different from the medium pruning treatment. The perforated line is the expected bunch number for the heavy pruning treatment, which was not significantly different from the light pruning treatment

There were no significant (P>0.05) treatment or block effects on bunch characteristics, including total weight, berry weight or berry number, in either year. Mean bunch weight for 2006/07 was 40.4 g (SD=6.13) and in 2007/08 it was 120.5 g (SD=36.80). Mean berry number per bunch for 2006/07 was 60.5 (SD=11.6) and for 2007/08 it was 101.0 (SD=34.7) and 2006/07 mean berry weight was 0.69 g (SD=0.16) and 1.19 g (SD=0.20) in 2007/08. These figures are consistent with the hypothesis that in a frost-prone season fewer bunches are produced, the bunches that emerge produce fewer berries, and the berries that emerge are of lower weight.

There was an apparent effect of treatment on flowering date in the frosted season (2006/07). In the vineyard, Pinot Noir vines of the same clone, which had not been affected by frost, reached 50% flowering (when flower caps had fallen from 50% of flowers on the majority of bunches) on 30 November 2006. Field observations suggested that 50% flowering was recorded 7 days later on 7 December on the control vines, the light pruning treatment resulted in 50% flowering a further 3 days later on 10 December, and medium and heavy pruning treatments an additional 9 days later, on 19 December. There was no apparent effect on date of veraison according to the vineyard manager's observations.

Although the post-frost pruning treatments did not have a significant effect on shoot number per vine, there was a significant effect on pruning weight, with the two most severe pruning treatments having significantly lower pruning weights than either the untreated control or the light pruning treatment, where just the damaged tissue of the green shoot was trimmed (Table 2). There was no significant treatment carry-over effect on pruning weight or shoot number in 2007/08 (data not presented).

Table 2  Mean pruning weight, shoot number, inflorescence number per bud and mean number of large inflorescences per bud (determined by dissection) for the three pruning treatments—frost-damaged tissue removed (light pruning); frost-damaged tissue removed and original spur trimmed back to one bud (medium pruning); and original (damaged) shoot removed back to compound buds on the spur (heavy pruning) —and an untreated control (control) in 2006/07

Treatment	Pruning weight (g)	Shoot number	Mean inflorescence number per bud	Mean number of large inflorescences
Control	0.84	40.22	1.9	0.75
Light pruning	0.74	42.60	1.5	0.20
Medium pruning	0.56	32.65	1.5	0.15
Heavy pruning	0.48	29.6	1.3	0.05
LSD (P=0.05)	0.17	ns	ns	0.46
ns=Not significant.

Bud dissections performed before commercial pruning in winter 2007 (Table 2) showed no effects on mean inflorescence primordia number per bud but a significant effect on inflorescence primordia size. The untreated control had significantly more inflorescence primordia classified as ‘large’ than any of the three pruning treatments. When actual bunch size was measured in the following season, 2007/08, there was no significant difference between treatments.

Discussion

Although it was not possible to record temperatures during any of the potentially damaging frost events, the regression between elevation and bunch number in the frost-damaged season is a strong indication of a direct inverse relationship between damage and air temperature. There is little published information on atmospheric inversion structure during localized frost events in Tasmania, but Wilson (2001) reported a near linear temperature increase of approximately 0.25^oC per metre, measured vertically, at a comparable site in southern Tasmania in 1999.

Visually, the most severe damage at the base of the slope was consistent with published and anecdotal reports of frost damage at the same stage of development. For example, Gardea (1987) (cited in Trought et al. 2003) reported that 50% of individuals, out of a population of Pinot Noir vines subjected to low temperatures between −1.2°C and −1.7°C between the second and fourth leaf stage, expressed frost-damage symptoms. The effect of the frost on bunch number as the primary yield determinant did not carry over into the following season, confirming that there was no residual direct effect of frost damage, nor any indirect effects resulting from altered crop load or other influences on vine growth.

The levels of damage near the base of the slope were similar to those discussed by Trought et al. (2003) and Creasy et al. (2002), who suggested that management options included taking no action, removing only dead material and removing all (including the green) shoots. Three of the four treatments in the present trial follow these suggestions closely, with the fourth being a more severe treatment, pruning all shoots back to the original compound bud. Each of these treatments would be expected to remove differing proportions of the total numbers of buds and thus offer differing potential for secondary or renewed primary (Sánchez et al. 2009) bud development.

The heaviest pruning treatment reduced yield compared with all other treatments, with the effect being on ripe bunches, suggesting an effect on the numbers of remaining primary shoots. Stergios and Howell (1977) (cited in Mullins et al. 1992) reported that primary latent buds can be considerably less hardy than secondary buds and it appears either that both primary shoots and dormant secondary buds were damaged in the one frost event, particularly lower in the inversion, or a second later event caused damage to the breaking secondary or renewed primary buds.

In the 2006/07 season, the heavy pruning treatments also reduced the pruning weight and number of large inflorescences but showed an increase in the total number of bunches, compared with the unpruned control, and total yield was not affected. Smaller cane diameters may be the reason for this apparent reduction in per bud fruitfulness, as Sánchez and Dokoozlian (2005) reported that buds on thicker canes tend to be more fruitful. Cane lengths were not recorded however, and although the decreased pruning weight for the same shoot number suggests a smaller cane diameter, it cannot be confirmed.

At the end of the season in which the damage occurred, there was a significant effect of treatment on the proportion of inflorescences classified as large in the dissections of dormant buds. This difference probably reflects differences in time of initiation, with earlier initiation occurring in the unpruned controls. The bunch structure is generally considered to be fully determined at the onset of dormancy (May 2004), with no increase in primary branching through the following stages of development. Srinivasan and Mullins (1981) described development of the bunch as spring growth commences, but noted that the extent to which small inflorescences at the start of winter dormancy contribute to reduced bunch size the following season is not known. The present results, with no treatment effects on bunch size the following year in spite of this marked difference during dormancy, suggest that inflorescence development did continue during dormancy with smaller bunches continuing to increase flower numbers up to a size equivalent to the ‘large’ inflorescences.

From a practical point of view, all treatments (including control) in 2006/07 produced a ‘second’ crop, which did not reach full maturity. Consequently, average sugar levels were below the usual limit for table wine production (average 8.5°Be) and the number of green bunches at harvest was unaffected by pruning treatment. Heavy pruning did reduce the crop; apparently by reducing the number of bunches from primary shoots able to reach full maturity.

At harvest, there was a low incidence of botrytis bunch rot on bunches of all vines, but this may be attributed to the exceptionally dry year (personal observation). The untreated vines had considerable levels of trash accumulated in the crown, and the damaged shoots appeared to be at risk of disease. Dead material would possibly cause more of a problem in more humid years, and less of a problem in drier climates, when the risk of botrytis bunch rot is lower.

The results indicate that removing frost-damaged material, by a relatively light superficial pruning of damaged material, will have little effect on the current and subsequent crops when this level of damage occurs in vineyards. Conversely however, although heavy post-frost pruning reduced yield in the current season, it may significantly improve the number of bunches and hence yield, in the second season. It is important to note that the results and conclusions generalize across a series of different levels of damage from most severe at the bottom of the slope to little or no apparent damage at the top. From a practical point of view and considering the expense incurred to prune after a frost, the control appeared to give the best results.

Although not presented above, the data showed suggestive but inconclusive evidence (P∼0.09) of an interaction between relative elevation and the heavy pruning treatment. Predictions obtained under our model (1) show that the heavy pruning treatment is associated with a reduction in expected bunch number of around 30% at all slope positions. Inclusion of the interaction term changed the predicted reduction in expected bunch number to range from 42% at the top of the slope to less than 1% at the base. This is consistent with the hypothesis that the relative effect of the pruning treatment is suppressed with increasing severity of frost damage. No similar interaction was observed for the lighter pruning treatments. This apparent difference in response to pruning with severity of damage showed no carry-over effect into the second season.

Overall these results suggest that, if growers do decide to prune after a damaging frost, severity needs to be matched against the amount of damage on each vine rather than a prescribed pruning guideline. With variable damage across a block, there is clearly a risk that prescribed pruning may limit both quantity and quality of an otherwise salvageable crop.

Acknowledgements

The authors wish to thank Geraldine Colombo and the owners of Tolpuddle Vineyard and Adrian Hallam and the owners of Meadowbank Vineyard for their support. Miss Anna Robertson from the University of Tasmania and Mr Paul Schupp from the Tasmanian Institute of Agricultural Research, also provided valuable assistance in the field.