Abstract
In this review, we briefly discuss factors that increase spring frost risk in viticulture and provide updates on vine susceptibility to frost events and damage assessment. The core of the review describes a physiologically based tool to prevent frost damage by delayed winter pruning (done at or beyond the “wool” bud stage) to postpone budbreak. The exploited principle is related to the inherent acrotony of the grapevine, which would “sacrifice” the already-developing apical shoots to frost, while basal nodes are still dormant and thus preserved. A survey of 21 published papers confirms that final pruning, performed not later than when two to three unfolded leaves are borne on apical shoots, would delay budbreak by ∼15 to 20 days, while yield is only mildly affected. At times, such a delay can carry on until harvest, postponing fruit maturity into a cooler time of year. Most recommended late-winter pruning protocols use a two-step intervention. In spurred cordons, a mechanical pre-cut that shortens canes to seven to eight nodes while also shredding wood can be made anytime during the dormant season. Thereafter, a final hand spur-shortening is made at a suitable developmental stage of the apical shoots. In a cane-pruned vine, previous year fruiting cane(s) can be removed any time in winter, while selecting at least two canes to keep vertical and longer than the required spacing-dictated length. Shortening of the two canes along with horizontal positioning should take place no later than when there are two to three unfolded leaves borne on the apical end of last season’s shoots.
Earlier phenology and the compression of the growing cycle in grapevines because of global warming have been reported widely (Jones et al. 2005, Van Leeuwen et al. 2019, Santos et al. 2020, Venios et al. 2020). Additionally, predictions are for longer growing seasons (Webb et al. 2012, Santos et al. 2020). Taken together, these scenarios justify the increasing need for cultural techniques that, especially in warm/hot regions, can delay key phenological stages and fruit ripening into a cooler period. A key characteristic of climate change is that the frequency and severity of weather extremes are increasing (Aghakouchak et al. 2020). Weather is “extreme” when unusually violent or abnormal in its frequency or length. Weather and climate-related extreme events encompass meteorological events (storms), hydrological events (floods, mass movements) and climatological events (heatwaves, cold waves, droughts, and forest or orchard fires; EEA 2022). In Europe, the increase of extreme events causing relevant production loss has been estimated at ∼60% over the past three decades. The European countries with the most crop failures related to extreme events between 1980 and 2019 is France, with 23,491 events causing 50% of insured losses, followed by Italy with 20,735 events causing 5% of insured losses, and Spain with 14,679 events causing 26% of insured losses (COM 2021, EEA 2022). Within this context, a puzzling interaction exists between global warming and grapevine susceptibility to spring frost damage (Muffler et al. 2016, Zohner et al. 2020). Despite a warming trend related to climate change and a steady decrease in the number of freezing days in Italy (ISPRA 2020), late spring frost is impacting crop production heavily, causing significant economic losses (Cicogna and Tonello 2017, COM 2021). The warming trend has indeed been beneficial, causing longer and warmer growing seasons, but has not reduced the risk of frost. In fact, frost events are not decreasing during overall warming in several viticulture regions (Shultze et al. 2016, De Rosa et al. 2021, Dinu et al. 2021). Budbreak is occurring earlier on average in several varieties and locations, but frost still occurs at about the same rate up until the frost-free date (Schultze et al. 2016). The severity and extent of damage caused by a late frost depends on several factors: altitude, flat versus sloped sites, distance of the vine from the ground, presence of windbreaks or physical barriers, proximity to and size of bodies of water, type of floor management, and so on. However, the increased frost risk is primarily due to earlier budbreak, which widens the time frame within which the probability of incurring freezing temperatures is greater (Kartschall et al. 2015), and to the fact that the frost event may occur when the vine is at a more advanced phenology stage, increasing the risk of severe damage. Typically, in warm/hot growing areas with an average growing season temperature (April to October in the northeast) from 17 to 24°C (Jones et al. 2005), any frost event occurring in mid-April would find vegetative growth at five to six expanded leaves, a phenological stage at which the vine is very susceptible to frost damage. Every degree (°C) of warming would advance budbreak by about seven to 10 days, greatly increasing the chance of frost injury (Webb et al. 2012). Dramatic frost events occurred in Italy, France, Germany, and Austria in 2017, 2020, and 2021 (COM 2021). Similar disruptive frost events happened in the eastern United States: 2014 (polar vortex), 2007 (Easter freeze), 2010 (Mother’s Day freeze), and 2012 (killer frost). These events cost crop insurance brokers an average $250 million per year (nearly half of total market value) in payments to grape and fruit growers for crop loss (www.rma.usda.gov).
These weather challenges have generated an interest toward innovative solutions for damage prevention or minimization (e.g., selective extraction of the coldest air [Arias et al. 2010], usage of hydrophobic particle film and acrylic polymers [Fuller et al. 2003], electrical heating cables [Lamb 2009], and vegetal oil application [Centinari et al. 2018, Herrera et al. 2018, Wang and Dami 2020]). Such relatively recent approaches are often coupled with other methods of frost protection in vineyards involving wind machines, air heaters, and sprinklers (Snyder and Paulo de Melo-Abreu 2005, Davenport et al. 2008). Coverage of these methods is beyond the scope of this review, which focuses on the technique of very late winter pruning, which delays budbreak and subsequent vine phenological stages until fruit ripening, generating a physiological approach to frost protection.
Incidence of spring frost damage in a climate change scenario. Incidence of spring frost damage in a climate change scenario is a complex relationship between several phenomena, including thermal length of the growing season (Trnka et al. 2011), number of frost-free days (FFD, defined as annual number of days with a minimum daily temperature above 0°C), and the advancement of the growing season triggered by the warming effect (Lavalle et al. 2009, Leolini et al. 2018). Several studies have reported a lengthening of the period between the occurrences of the last spring frost and the first autumn frost (Schwartz et al. 2006, Schultze and Sabbatini 2019). This has occurred in recent decades in several areas in Europe and the U.S. and more generally in the northern hemisphere. However, across all of Europe, the delay in the end of the season (8.2 days in the period 1992 to 2008) was more significant than the advanced start of the season (3.2 days over the same time period; Jeong et al. 2011). A somewhat related effect is a general and clear increasing length of the frost-free period in Europe between 1985 and 2014 (Figure 1). The trend is not uniformly spread over Europe, however, and the fastest rates of change (an extension of the frost-free period by more than 0.8 days per year) were in eastern and northern Europe (EASAC 2013). An interesting regression analysis between time (1974 to 2008) and number of FFD in four different regions/countries in Europe shows a regional pattern (Lavalle et al. 2009). In Denmark and the highlands of the United Kingdom, a significant positive linear regression was observed, with an FFD increase of ∼65 and 90 days over a 35-year span, respectively, while in Extremadura (Spain) and Thessalia (Greece), a negative linear regression was found with an FFD reduction of ∼50 and 24 days, respectively. In areas with a significant decrease in the length of the frost-free period, such as in southern Europe, plants are more at risk from frost damage due to delay in the last winter-spring frost (Lavalle et al. 2009). In the eastern U.S., similar trends are reported. Interannual variability of early-season temperatures has led to large increases in frost occurrence from year to year (Schultze et al. 2016). Subfreezing temperatures occurred frequently in March, April, and May. For example, the spring frosts of 2010 reduced juice grape production by 60% (Schultze et al. 2014). An abnormally warm early spring, followed by a return to normal climate, devastated the entire fruit crop in 2012. The spring was 3.7°C warmer than the previous 30-year average throughout the area, featuring days as much as 20°C warmer than their climatological average (Schultze et al. 2016). Many grapegrowing regions reported accelerated phenological development, only to experience devastating frosts in early April across the country. According to the United States Department of Agriculture (USDA), crop losses that year were as high as 95% for tart cherries, and 75 and 40% percent for juice and winegrapes, respectively.
Annual rate of change of frost-free days representing the trend coefficient for long-term changes in the annual number of days with a minimum daily temperature above 0°C. For example, a value of 1 indicates that the number of frost-free days has increased on average by one day per year over the last 30 years (1985 to 2014). The analysis is based on the JRC-MARS gridded meteorological data at 25 km resolution. Data source: http://agri4cast.jrc.ec.europa.eu/DataPortal/Index.aspx?o=d in https://www.eea.europa.eu/data-and-maps/figures/rate-of-change-of-frost-1.
The pivotal factor is how global warming is advancing the growing season in fruit trees and grapevines. A bloom advancement of 2.3 days/10 years was reported in several fruit trees: 2.0 days/10 years in cherry and 2.2 days/10 years in apple (Chmielewski et al. 2004). In the grapevine, the advancement of phenology is likely one of the most consistent trends supported by historic data and projection analysis. Budbreak advancement over several decades in grape varieties grown in different environments can vary between seven and 14 days. The advancement is driven by the increased growing season mean temperature, as +1°C advances budbreak by about seven to 10 days, greatly increasing chances of frost injury (Fraga et al. 2012, Webb et al. 2012, Van Leeuwen et al. 2019, Santos et al. 2020, Bernáth et al. 2021, Droulia and Charalampopoulos 2021).
The general expectation is that advanced budbreak may expose vines to spring frost more frequently, because the last frost event will occur at a more advanced stage of shoot growth development, when the organ is most susceptible to freezing (Fuller and Telli 1999). However, several climate analyses carried out in different European locations to estimate spring frost risk under a warming scenario reported contradictory results. A comprehensive chapter on impacts, adaptation, and vulnerability related to climate change in Europe, when addressing projected changes in climate extremes, states a general high confidence concerning several phenomena: a) changes in temperature extremes (toward increased number of warm days, warm nights, and heat waves); b) increases in extreme precipitation in Northern Europe (all seasons) and Continental Europe (all seasons except summer); and c) in extreme sea level events (Kovats et al. 2014). Any other change is defined as medium or low confidence and some, like cold waves and spring frost risk projection, are not even listed. Uncertainty related to projections for spring frost risks in Vitis vinifera L. seems to be confirmed in several other, more specific, studies. Phenological changes and forecast frost risk for the cultivar Riesling grown in Germany were calculated using historic data (1901 to 2019) and a projected time span (2011 to 2100) under two climate change scenarios (RCP8.5 and RCP2.6; Kartschall et al. 2015). An acceleration of all main phenology phases was found from the late 1980s, while projection for 2031 to 2060 modeled an acceleration of 11 ± 3 days under the RCP8.5 scenario. Within the same scenario, frost risk is expected to increase slightly over the next decades. Assessment of late frost damage risk in the French regions of Alsace, Champagne, and Burgundy throughout the 21st century was made using three different phenological models, predicting either statistical occurrence of the last frost day or the characteristic budbreak date (Sgubin et al. 2018). Outputs showed that the probability of late frost is expected to increase significantly in two out of three models, while the third model gave contradictory results. For southwest England, the risk of late spring frosts increases under many future climate projections extending until 2099 due to advancement in the timing of budbreak (Mosedale et al. 2015). However, estimates of frost risk were highly sensitive to the choice of phenological model. A case study of the Swiss Rhone Valley (Sion and Aigle locations) employing 12 phenological models projected over 2021 to 2050 concluded that frost risk might increase or decrease, depending upon location and climate change projections (Meier et al. 2018). To add even more variability, a study conducted in Luxembourg reported that the frequency of spring frost damage in the Luxembourg winegrowing region will decrease, but not disappear in the near (2021 to 2050) or far (2069 to 2098) future projections (Molitor et al. 2014). The most comprehensive survey on changes in weather extremes for Italy, for the period 1961 to 2012, reported significant trends in the mean decrease of number of freezing days (-11 on average), especially after 1990 (ISPRA 2013). Despite this apparent decrease, several grapevine districts in Italy were hit in 2017, 2020, and 2021 by severe late spring frost events. The damage in 2017 and 2021 was quite variable. Moving north to south through the country, vineyards reported 20 to 100% damage and loss of the potential crop was estimated at ∼20 to 25% in Lombardy, 20 to 30% in Emilia Romagna, 20 to 25% in Tuscany, and 25 to 35% in Puglia (Atzeni 2017). Indeed, heterogeneity in viticultural traits of Italy play a pivotal role in damage variability. These include a) a wide range of cultivars grown, with large variation in budbreak dates; b) different growing meso-climates, which might enhance or limit susceptibility to late frost; and c) high diversity in training systems with different susceptibility to damage, mostly due to varying distance of the fruiting area from the ground. Summarizing the examples provided above, late spring frosts are a significant risk to grape production in frost-prone viticultural regions. The increase in air temperature due to climate change is likely to advance grape budbreak and exacerbate last frost events and consequent damage in the spring.
An update about grapevine spring frost susceptibility and assessment. Physiological mechanisms and methods. The physiology of grapevine cold hardiness across several budbreak stages (Table 1) has been extensively reported in the literature and results are used to assess and interpret vineyard damages (Howell and Wolpert 1978, Johnson and Howell 1981, Wolpert and Howell 1986, Moncur et al. 1989, Fuller and Telli 1999). On Madeleine Angevine and Siegrebbe cultivars grown in the U.K., budbreak is accompanied by a linear increase in water content in the bud during initial developmental stages up until wool buds (BBCH 05; Fuller and Telli 1999). Freezing tests, enclosing vines in chambers with progressive cooling to a minimum temperature of -4.5°C, found large changes in bud water content from dormant (BBCH 00, water content ∼40%) to green pointing buds (BBCH 07, water content ∼83%). Bud freezing correlated linearly with bud water content and at the mean exotherm temperature of -3.5°C, causing damage of 10 to 20% until stage BBCH 03, defined as end of bud swelling; buds swollen, but not green (Lorenz et al. 1995). Damage was scored at ∼80% at the BBCH 05 stage (Fuller and Telli 1999), and the authors concluded that the temperature at which buds froze was not influenced by cultivar or acclimation treatment. This is consistent with research on Pinot noir, where bud development stage did not influence ice nucleation temperature (Luisetti et al. 1991).
Critical temperatures of Pinot noir and Concord at different phenological stages. CT is the temperature at which plant tissues are damaged and is expressed as CT50, corresponding to 50% damage.
Endogenous freezing, assessed using infrared thermography, found that freezing in two grapevine varieties initiated in the cane, then travelled into the buds at a speed of 0.47 cm/sec (Hamed et al. 2000). A model for dormant bud cold hardiness and budbreak prediction in 23 Vitis genotypes was calibrated extensively (Ferguson et al. 2014). Most notably, budbreak occurred earlier in hardier genotypes, consistent with more rapid de-acclimation of genotypes originating from colder climates. As a paradox, these genotypes were more vulnerable to spring frost in warmer environments.
As young, developing leaves are more susceptible to frost due to their greater water content (Fuller and Telli 1999), it was hypothesized that they can function as a better proxy for cultivar phenotyping against frost tolerance. The degree of frost resistance in young leaves of 15 grape cultivars was assessed for this purpose (Sun et al. 2019). The super cooling point was not suitable for comparing frost resistance; instead, the most effective parameters for discrimination were T2 (freezing point) and t2 (time when temperature raises from T1, the super-cooling point, to T2). Based on such analysis, the most frost-resistant varieties were Muscat Hamburg and Frontenac; Summer Black had an intermediate resistance and the remaining varieties were categorized as low-resistant.
Another challenge in determining the frost resistance of plant tissues is choosing the method of air temperature measurement that best represents the actual temperature of the plant organ. Plant and organ temperature is a complex interaction of absorbed radiation, wind speed, wetness, and organ shape that provide shelter or stimulate significant turbulence (Grace 2006). However, buds are dry structures and incapable of effective transpirational cooling; thus at night, when the radiant energy balance is negative (long wave energy streaming from the bud to the sky), the buds are colder than the atmosphere and the magnitude of such a difference is a strong function of wind speed (Michaletz and Johnson 2006). The issue of representativeness of air temperature measures for true organ temperature has been addressed in budding leaves of the grapevine, apricot flowers, and unripe pear fruits (Litschmann and Středa 2019). For young grapevine leaves, the greatest deviations were found between the surface temperature measured with an infrared thermometer sensor and a traditional sheltered thermometer (Stevenson screen type); the latter invariably showed higher nighttime temperatures by 0.4 to 0.8°C on average, with a peak of 1.5°C. Conversely, both a wet bulb thermometer (unsheltered thermometer covered with a wet cloth) and a simple unsheltered thermometer were closer to the actual organ temperature with one important distinction. While the unsheltered thermometer provided the best relationship with the temperature of the plant tissue under any condition, the wet bulb readings were affected by the relative humidity (RH) of the air. When RH was below 65 to 70%, temperature values at night were several degrees lower than bud surface temperature. Thus, when a wet bulb was used to activate a frost protection device (e.g., sprinkler irrigation), the warning could occur prematurely.
Cultural factors. The role that soil temperature can play in regulating the response of grapevine leaves to frost was investigated (Sun et al. 2018). In a pot experiment where Merlot seedling roots were kept either in warm (∼20°C) or cold (∼0°C) soil and then subjected to frost treatment, severe damage to the young leaves occurred with cold soil, while warm soil led to reduced frost injury. A non-targeted metabolomic analysis showed that in the warm treatment, pathways related to the citrate cycle, glycine, serine, and threonine were enhanced. This outcome raises the question of whether root distribution in soil volumes at different temperatures could significantly impact vine frost susceptibility. While soil/root temperature is indeed affected by factors such as lithological features, soil texture, water holding capacity, color, and organic matter content, root depth is quite responsive to under-trellis floor management (Centinari et al. 2016, Klodd et al. 2016). Any practice that favors a shallower root system due to an undisturbed soil surface (e.g., mulching, herbicides) may help mitigate the consequences of a frost event (Guerra and Steenwerth 2012). Moreover, clean cultivated soil absorbs and then re-radiates more heat, providing frost risk mitigation. Any condition that favors soil warming in spring, and therefore root metabolism, is also expected to advance budbreak, increasing the risk of frost damage and potentially counteracting beneficial effects discussed above. This hypothesis is still uncertain: work done on potted Shiraz grapevines exposed to two different soil temperatures (13°C and 23°C) showed no effects on the time of budbreak, anthesis, or the number of flowers per inflorescence (Field et al. 2020).
New digital technologies. Recent efforts in precision viticulture (Matese and Di Gennaro 2015, Ozdemir et al. 2017, Giovos et al. 2021) have provided interesting approaches to monitor and detect spring frost damage in vineyards by remote sensing. Some vegetation indices (VI) such as Red-Edge 7, NIR, EVI, MTVI1, and CARI, all calculated from medium resolution Sentinel-2 acquired data, proved effective in estimating lower light reflectance in pergola-trained vineyards after severe frost damage when compared to undamaged vineyards (Cogato et al. 2020). Moreover, the same VI bands provided evidence of recovery to full canopy size ∼40 days after the frost event. It is indeed encouraging that, likely due to the frequent revisit time of the Sentinel-2 constellation, generating robust time-series for spatial and temporal analyses can be used to assess the impact of late frost in vineyards. The pergola trellis type, which forms an almost horizontal continuous green cover, likely facilitates the performance of indices calculated from low or medium spatial resolutions. However, the robustness of this method will require association with yield or grape quality data and extension to vertically shoot-positioned (VSP) trellises showing a typical discontinuous green cover.
An even broader application is a practical remote-sensing monitoring framework for late frost damage in winegrapes based on in-situ measurements and multi-source satellite data (Li et al. 2021). This framework estimates the daily minimum air temperature (Tmin) with a spatial resolution of 100 m and was used to map the severe late frost damage that occurred in northwest China in April 2020. About 41% of the vineyards suffered severe frost damage, and the total affected area was ∼16,381 ha. The predictions of late frost damage obtained by estimating Tmin agreed with the statistics of the Agricultural Meteorological Disaster Department.
Using high spatial resolution analyses of minimum night temperatures to explore the impact of current and future frost risk is another active research field. Multivariate Adaptive Regression Splines (MARS) were used to model high-resolution (30-m grid) minimum temperatures in the Yarra Valley wine region in southeastern Australia (Gobbett et al. 2020). The accuracy for predicting minimum night temperature records was good (R2 = 0.68), while all future climate scenarios project down-elevation movement of the frost line between 10 and 30 m, depending on scenarios. A similar approach in a study on viticulture suitability in Tasmania identified land areas prone to damaging spring and late-season frost (Webb et al. 2018). The main outcomes were that risk classifications for the -1°C threshold were appropriate for this study and that the model’s classifications of suitable, moderately suitable, and unsuitable areas, defined as <1 frost every 10 years (<10%), >1/10 to one frost every two years (10 to 50%), and >1 frost every two years (>50%) for temperature values ≤-1°C, respectively, correlated well with viticulture suitability and grower expectations.
Technical Protocols for Application of Delayed Winter Pruning
Using late winter pruning close to or after the time of normal budbreak to postpone the onset of vegetative growth is neither a brand new nor revolutionary concept in viticulture. Quite old work had already determined the main traits of temperature-driven dormancy-breaking mechanisms in the grapevine bud (Pouget 1966), and that different mid-winter pruning dates, spanning from November until March (northern hemisphere), were ineffective at altering budbreak response (Wample 1994). According to the literature search provided in the present review paper, pioneer studies hypothesized that a purposely delayed winter pruning could achieve a consistent budbreak delay to be used as a prevention tool in areas at high risk of spring frost (Friend and Trought 2007, Friend et al. 2011). Other research covered still-unknown or poorly-studied characteristics of the technique; among them: i) seasonal variation in canopy physiology and efficiency; ii) impact on yield components and return crop next year; iii) chances that the initial growing cycle postponement can carry until ripening; and iv) consequences on grape composition and wine styles. Before addressing these topics, we will describe current best practice for delayed winter pruning on either spur-pruned or cane-pruned vines.
Spur pruning. Once other conditions are similar (e.g., the distance of the productive cane or cordon from the soil, bud load per meter of canopy length, etc.), susceptibility to spring frost damage of spur-pruned cordon vines is usually greater than that observed in a cane-pruned system (Figure 2A and 2B). The reason is related to the different pruning cuts, where a short spur with the typical two-count node has reduced acrotony compared to a long cane with eight to 10 buds. This leads to very uniform sprouting, which is more conducive to severe frost damage (Lavee and May 1997, Intrieri and Poni 2000, Ezzili and Bejaoui 2001, Daskalakis and Biniari 2019). In a situation such as in Figure 2A, any residual crop to be harvested in the current season derives from shoots from latent, base, and secondary buds, which are much less fruitful than primary buds.
Close view of a spurred cordon and a cane-pruned Sauvignon blanc vine after a very severe frost event in central Italy (spring 2017). Notably, shoots produced on the spurs suffered 100% mortality (A), while in the cane-pruned system, only apical shoots were killed while the still-dormant first four to five basal nodes avoided the freezing injury (B).
A common working protocol is used in most studies of the efficacy of highly delayed winter pruning strategies in spur-pruned cordons (Figure 3). The operation can be performed in two steps or in a single passage. The two-step procedure uses mechanical prepruning, executed anytime during the dormant season. The over-row rotating disk machine used, unlike a cutter bar machine, easily avoids posts and rigid obstacles along the row, even when the cutting distance from the cordon is regulated to leave more than seven to eight nodes per cane (Figure 3A). Moreover, pruning machines employing rotating disks can perform on-the-go cane chopping and shredding, greatly facilitating any subsequent hand operation. Final shortening to the required spur length (Figure 3B) is performed quickly by hand when, on average, two to three unfolded leaves are formed on the apical buds of the canes. Depending upon vine vigor, lateral canes may be present on the apical portion of last season’s shoots. While their vigor is expected not to be very high as the mechanical prepruning removes most apical nodes and, with them, vigorous laterals developed after shoot trimming the previous season, their presence exerts some control on the development of the subtending nodes (Pellegrino et al. 2020). Therefore, they should be the main target to observe when visual scouting is performed after budbreak to assess whether the two to three unfolded leaf stage has been reached.
Diagram of the two pruning steps recommended for delayed winter pruning in a spur-pruned cordon. (A) Mechanical prepruning is performed anytime during the dormant season to leave seven to eight node canes and concurrently perform wood shredding. (B) At the optimal stage of two to three unfolded leaves developed in the apical part of the prepruned canes, manual shortening to the desired spur length is executed.
Based on working time needed to manage a fully or partially mechanized VSP trellis (Intrieri and Poni 1995), combining winter mechanical prepruning with a late hand-finishing should require ∼50 to 60 working hrs/ha (10 to 15 hrs for mechanical hedging and 40 to 45 hrs for the subsequent follow-up, depending on cane number per vine and degree of residual cane detachment), which is less than the estimated working time needed to manage one-step hand pruning in VSPs of comparable vigor (70 to 90 hrs/ha).
Cane pruning. Due to factors related to apical dominance exerted by apical buds on a horizontally positioned cane (Intrieri and Poni 2000), susceptibility to late frost damage of cane-pruned vines is usually milder than that on spur-pruned cordon training systems. The staggered budbreak occurring along horizontal canes (Figure 2B) allows basal and median nodes to be at a delayed growth stage when the freezing event takes place, limiting damage. However, applying a delayed winter pruning on a long cane training system (e.g., Guyot, Pendelbogen) is slightly more cumbersome and risky than on a spur-pruned training system with a permanent cordon. If the successful two-stage pruning described previously for a spur cordon is to be replicated, then a cane-pruned system should receive a preliminary pruning adjustment in the winter (i.e., past year fruiting cane separated from the one to two canes selected for renewal) and the final pruning will then pertain to the selected cane(s). However, this approach was tested on cane-pruned Pinot noir grapevines and had only mild effects on vine phenology and overall vine performance (Gatti et al. 2018). Conversely, when all pruning operations were postponed at a stage of about two to three unfolded leaves on the distal portion of the unpruned canes, a delay in budbreak of 18 days (data pooled over three years) was obtained when compared to the standard winter pruning. However, the one-time pruning method is unlikely to be favored by growers for several reasons: i) postponing and performing in one step the quite complex cane pruning method requires access to skilled hand labor within a narrow time window; ii) an intervention made on the whole canopy when budbreak has already initiated on the apical part of the canes will unavoidably slow down the operational times and potentially damage the swollen buds or initially-developed shoots; and iii) a psychological barrier may exist against the idea of operating so late in a still-untouched canopy. Therefore, a good compromise must be found between the one-step and two-step procedures, giving preference to the latter. In the absence of previous on-site experience, a reasonable starting point would consist of a first passage (A), during which a past-year cane is removed and at least two vertical long canes are retained (Figure 4). Final pruning (B, B’) would shorten the two canes to a length that maintains mostly dormant nodes (B) and simultaneously fills the spacing on the wire (B’). Although no specific research addresses this, the chances of inducing a significant budbreak delay on retained median and basal nodes are maximized if: i) retained canes are as close to vertical as possible, as this position enhances the effects of acrotony; and ii) retained canes are longer than the ones retained for production, dictated by the intrarow vine spacing. When such requirements are overlooked, the chances of delaying budbreak decrease significantly. For example, when short shoot trimming is performed during the growing season, optimal cane length is often unavailable at the time of cane selection during winter pruning, and this may interfere strongly with technique efficacy. Although horizontal cane positioning still obeys the principle of acrotony, the inhibition of basal buds is weaker, which undermines the efficacy of late pruning (Figure 2B). The effects of different cane-pruning dates on Sauvignon blanc was determined, with the last treatment performed just prior to budbreak, therefore without any active vegetation (Trought et al. 2011). Nevertheless, the last pruning dates postponed budbreak by about five days after that of the winter pruning dates, a delay that was recovered by the time of bloom.
Diagram of the pruning steps recommended for delayed winter pruning in a cane-pruned training system (i.e., Guyot type). (A) First run of hand pruning is performed anytime during the dormant season to remove the previous year production cane and to select at least two long canes, maintaining them in nearly vertical positions. (B) At the optimal stage of two to three unfolded leaves at the apical part of the retained canes, manual shortening to the desired cane length is performed, along with positioning and tying on the horizontal support wire. (B’) Red boxes in panel B indicate selected canes with extra length, which should be maintained to: i) increase acrotony control of the subtending nodes, and ii) ensure suitable cane length to fill space on the trellis.
Effects of delayed winter pruning on seasonal phenology and vine performance. Budbreak response. A list of published research articles, arranged according to country, cultivar, pruning type and timing, budbreak time, yield response, and ripening delay as compared to standard winter pruning, is provided to identify works published in peer-reviewed scientific journals and conference proceedings that focused on the effect of late winter pruning on the grapevine (Table 2). The primary criteria for data curation were the inclusion of research that, within the definition of late pruning, used at least one treatment with pruning at the swollen bud stage or later. Out of 21 published papers studying an array of cultivars and wine regions, mostly using spur pruning, late winter pruning was carried out anywhere between the two extremes of swollen buds and seven to eight unfolded leaves.
Synoptic information from research papers on the impact of late winter pruning (LWP) strategies on growth, yield, and fruit quality of grapevine. C, control; TSS, total soluble solids; TA, titratable acidity.
It is hypothesized that the delay in basal bud development associated with delayed pruning is related to correlative inhibition, as suppression of growth in basal nodes occurs when longer canes, with additional buds, are left at pruning time (Howell and Wolpert 1978). This phenomenon is the expression of shoots growing basipetally along a grapevine cane and becoming stronger with increasing cane length and position tending to become vertical (Bangerth 1989, Lavee and May 1997). These principles find robust confirmation in the listed works as, compared to a control pruning performed at full vine dormancy, any late winter pruning has generated a budbreak delay varying from five to 56 days (Table 2). Such delays in vine growth greatly increase chances to avoid or limit frost damage: the wider the time window between control and late-pruned treatments, the greater the probability that a frost event occurring within this period might hit the already-developing apical shoots while the basal nodes are still dormant (Figure 5). Only one study examined a sprawling canopy, and while many of the canes remained horizontal, few basal buds had burst when the delayed pruning occurred (Petrie et al. 2017). The impact of delayed budbreak can be different in warm and cool climates, where it could also significantly delay ripening. Additional studies in such regions should check whether and how the vines use the recovery mechanisms which, in a warm climate, often allow progressive depletion of the initial delay (Gatti et al. 2016).
(A) A spurred cordon in Central Italy (cv. Sangiovese) showing almost 100% main shoots killed by late spring frost in 2021. (B) and (C) Two different details of cane growth two weeks after the frost occurred. Notably, the already-developed apical shoots are dead, while nodes located underneath either show healthy green tissue or are still dormant.
The inherent difficulty of these studies is that data validation, provided by the occurrence of a significant frost event within the trial duration, is truly unpredictable, for obvious reasons. However, such a situation occurred on 29 April 2019 during a two-year trial on Lemberger, when a freezing event occurred when the phenological stage of the control averaged between E-L 3 and 4, woolly buds and green leaf tips visible, respectively (Persico et al. 2021). In 2019, late-pruned (1 May) vines had 61% greater yield than control vines, reflecting differences in shoot freeze damage between the two treatments. Moreover, final grape quality was not affected. Another instance of documentation of effects of late winter pruning in the presence of a significant frost event was a Chardonnay trial in New Zealand (Friend et al. 2011). On 25 to 26 September 2000, a radiation frost occurred with minimum night temperature reaching -1.7°C, causing damage to a portion of the developing buds, quantified at 33% killed primary shoots. Conversely, in the late pruning treatments, killing of the primary shoots was limited to no more than 3%. Then, albeit on a more observational basis in a mature Sangiovese vineyard in Tuscany, Barmpa et al. (2021) verified the effectiveness of late pruning over a severe frost event which occurred between 6 to 8 April 2021 (Figure 6). Cane shortening to required spur length was performed on 22 to 23 April, and the study also included an intermediate treatment where, after mechanical prepruning done the previous December, cane trimming was quickly performed in January 2021 to retain only one cane per spur. With frost damage of ∼90% primary shoots killed, leading to a final cluster number and yield of 7.6 and 1.51 kg/vine, the recommended two-step late pruning maintained at harvest 15.2 clusters and 3.10 kg/vine, while a three-step approach led to 11.5 clusters and 2.34 kg/vine (Barmpa et al. 2021).
(A) A spur-pruned vineyard of Sangiovese in the Chianti Classico area, photographed 22 April 2021, as it appeared after mechanical prepruning performed December 2020 and before spur shortening. Severe frost had occurred on 6 to 8 April. (B) The same vineyard where a variant of stage two was adopted and only one long cane for each previous year spur was maintained and then shortened to the desired spur length.
Yield response. The yield response to the phenological stage at which winter pruning is performed seems quite straightforward: a low-to-moderate yield limitation is found when pruning time does not exceed the two to three unfolded leaves stage, while pruning performed at a much later stage severely impairs yield, reaching >50% reduction below standard pruning (Frioni et al. 2016, Gatti et al. 2016, Petrie et al. 2017, Silvestroni et al. 2018, Allegro et al. 2019). The hypothesis of increasing yield limitation with later interventions was first described in a study where the latest treatment (more than seven to eight unfolded leaves) caused a 92% yield reduction due to a very low cluster number (Gatti et al. 2016). The mechanism involved is hypothesized as severe source limitation imposed by very late pruning, causing carbon starvation to the developing inflorescences, a large majority of which deviate into a tendril differentiation (Yang and Hori 1979). The same paper also suggests that to avoid a major yield limitation on a 1 m spur-pruned cordon with an eight to 10 node bud load, removed leaf area per vine should not exceed 0.2 to 0.3 m2.
An examination of case studies suggests that cultivar-dependent yield reduction is an important factor to consider (Table 2). A distinct outlier is a study in a Merlot vineyard in New Zealand, which reported a spectacular yield increase with winter pruning delayed until the development of ∼5 cm long shoots (about two unfolded leaves; Friend and Trought 2007). While there is some evidence that a progressive shift of winter pruning closer to budbreak is conducive to a slight yield increase (Coombe 1964), the results reported by Friend and Trought (2007) highlight a special case, as it found in the latest pruning a significant increase in the proportion of large, seeded berries and a drastic reduction in smaller and shot berries. This study was conducted on the east coast of New Zealand, where sudden changes in temperature (10°C change in the space of 20 min) are possible even as late as Nov/Dec (flowering time), favoring setting of a high fraction of seedless, lightweight berries. Conversely, other studies where yield showed a sometimes remarkable increase over standard winter pruning (Friend et al. 2011, Barmpa et al. 2021, Persico et al. 2021) all reported a late frost event occurring before the hand finishing was performed. Not surprisingly, in all cases, such yield preservation occurred without impacting fruit quality at harvest.
Carrying delay until harvest? Chances to extend the delayed growth and development cycle obtained with late winter pruning until harvest is primarily a function of the magnitude of the delay in spring shoot development. This can vary from a few days (Martin and Dunn 2000, Buesa et al. 2021) to >50 days (Zheng et al. 2017, Silvestroni et al. 2018, Allegro et al. 2019). Such a large variation in the postponement of the start of budbreak is related to several factors, including: i) the phenological stage, hence advancement of growth, when the pruning is performed (usually the later the pruning, the longer the budbreak delay); ii) seasonal conditions characterizing the post-pruning phase and differences in crop level, which may hinder late season growth; and iii) the length and position of the unpruned canes, which can affect the number of burst nodes at the time of pruning and the amount of leaf area removed.
Among papers assessing key phenological stages following delayed winter pruning (Gatti et al. 2016, 2018, Frioni et al. 2016, Silvestroni et al. 2018), a general erosion during the season of the maximum delay at budbreak is quite clear. The one exception is likely due to the specific canopy type (sprawl) used in the experiment (Petrie et al. 2017). Physiological bases underlying this behavior are not easy to disentangle, as the canopy in the delayed-pruning treatments developed over a month later than usual, so that leaf formation, growth, senescence, and all stages of berry development took place under different environmental conditions and the dynamic of the source-to-sink balance was deeply altered.
Due to the complexity of the interactions involved, following the seasonal canopy changes to compare a severely delayed winter pruning with standard winter pruning is difficult. The challenge was undertaken using potted vines and a whole-canopy gas exchange system (Poni et al. 2014) and tracked, from budbreak until almost leaf shedding, the net CO2 exchange rate (NCER) of canopies subjected to either late (two to three unfolded leaves; LWP) or very late (seven to eight unfolded leaves) winter pruning and compared them to standard winter pruning (SWP; Gatti et al. 2016). LWP achieved a 17-day delay at budbreak that shortened progressively during the season. The harvest threshold set at a total soluble solids (TSS) of 20 Brix was reached three days before SWP. Three mechanisms contributed to this efficient compensation: i) greater canopy efficiency, measured as shorter time needed to reach maximum NCER/leaf area (22 days versus 34 in SWP); ii) greater maximum NCER/leaf area (+37% more than SWP); and iii) greater NCER/leaf area ratios from veraison to end of season. As a result, seasonal cumulated carbon in LWP was 17% greater than in SWP.
In most studies examining the effect of late pruning, there was a significant delay in grape maturity, measured either as the estimated number of days needed to reach the same maturity as the control vines at harvest or, if there was single harvest date, as reduced physiological ripeness. A frequent trait of this delayed ripening was that technological maturity, assessed as sugar-to-acid ratio, almost invariably confirmed slower sugar accumulation combined with better acid retention and improved phenolic maturity as a consequence of late winter pruning. Thus, late winter pruning of Merlot carried out at three unfolded leaves produced clusters with higher titratable acidity (TA) and anthocyanin-to-sugar ratios at harvest than standard pruning (Allegro et al. 2020). Similarly, pruning Shiraz and Cabernet Sauvignon vines between E-L 2 and E-L 15 produced fruit with an increased anthocyanin: TSS ratio for a sugar concentration >13.5 Baumé (Petrie et al. 2017). Late pruning of Sangiovese (Palliotti et al. 2017, Silvestroni et al. 2018) and cane-pruned Pinot noir (Gatti et al. 2018) arrived at a consistent common scenario of significantly delayed technological maturity at harvest (i.e., lower TSS and higher TA), unaffected total anthocyanins, and improved phenolics. Despite working under largely different conditions and with different cultivars, three studies on delayed pruning found delayed sugar accumulation was associated with an increase of total anthocyanins and phenolics at harvest (Frioni et al. 2016, Moran et al. 2017, Zheng et al. 2017).
The above results show decoupling of anthocyanins and sugar accumulation: a primary challenge under a global warming scenario, especially in warm/hot regions where the rate of sugar accumulation should be reduced, while accumulation of phenolic components should remain stable or even be improved. Late winter pruning joins the array of techniques used for this purpose (Sadras et al. 2012, Sadras and Moran 2012, Palliotti et al. 2014, Poni et al. 2018, 2020, Gutiérrez-Gamboa et al. 2021). However, an explanation for this decoupling potential of a late winter pruning technique must be provided. Based on evidence that anthocyanin synthesis is inhibited and its degradation, enhanced, at berry temperature >35°C (Mori et al. 2007), the most likely hypothesis is that if delayed pruning shifts the onset of veraison into a cooler period, berry pigmentation is enhanced. Recent work on the interaction between timing of winter pruning and elevated temperature in Shiraz concluded that late pruning maintained the anthocyanin-to-sugar ratio, which decreased with heating in two seasons (Moran et al. 2019, 2021). More importantly, wine color density and concentration of anthocyanins and phenolics correlated negatively with daily mean temperature in a short window (two weeks) immediately after veraison. Other effects cannot be ruled out, however. In a trial on Merlot, late winter pruning treatments significantly increased the skin-to-pulp ratio at harvest (Allegro et al. 2020). If cell skin formation takes place within four to five weeks after flowering (Coombe and McCarthy 2000), the process can benefit from more rapid cell division, allowed by postponing the first stage of berry growth into a warmer period. However, the proposed mechanisms have some important deviations: the latest winter pruning applied on Shiraz at E-L 15 and on Cabernet Sauvignon at E-L 11, despite a significant yield reduction, did not achieve the highest anthocyanin:TSS ratio (Petrie et al. 2017). Thus, the decoupling effect of late winter pruning risks being spoiled when a very late timing of intervention (i.e., later than three to four unfolded leaves) leads to prolonged source limitation and carbon deficit (Gatti et al. 2016), which might ultimately impair color accumulation. As reported in Sangiovese and Cabernet Sauvignon vines, a reduced leaf area-to-yield ratio (i.e., 0.33 m2/kg in source-limited vines versus 1.15 m2/kg in control vines) decreased the total anthocyanin concentration to 84.3% of the non-source-limited control, while it decreased the sugar concentration by only 27.1% (Bobeica et al. 2015, Zhu et al. 2019).
The sugar/color decoupling issue assumes quite different traits in a cool climate viticulture, where the most desirable outcome is the longest budbreak delay that doesn’t severely curtail yield and the ability to properly ripen. However, some light can also be shed onto such scenarios. While previous work conducted in cool climate regions (Dami and Beam 2004, Loseke et al. 2015, Wang and Dami 2020) already demonstrated that budbreak-delaying products did not affect grape composition at harvest or wine chemistry, provided that cycle postponement did not exceed 10 to 14 days, when a similar delay was induced through late winter pruning (Persico et al. 2021), wine chemistry, carbohydrate storage, and bud-free tolerance in the following dormant season were not affected.
Conclusions and Future Directions
More advanced grapevine phenology has consistently aggravated damage and related economic losses due to spring frost events in viticulture worldwide. Such worsening is less related to the increased frequency of the freezing events, than to more-advanced bud and shoot growth stages, which significantly increase the probability of severe damage. These events are also now occuring in warm areas which, traditionally, have been almost exempt from such fatalities (i.e., several wine districts in central and southern Italy).
This has increased concern about late frost damage and stimulated new approaches to prevent and mitigate damage. In this review, we described a physiological solution consisting of moderately delayed winter pruning. The principle is both simple and effective. Extra apical buds are retained on canes before final shortening to the pruning length required by the training system. To avoid a significant yield decrease, the late pruning should be done when apical shoot development does not exceed two to three unfolded leaves, or ∼5 cm long. This intervention can postpone budbreak by 15 to 20 days in the basal or subtending nodes, greatly increasing the likelihood of avoiding or limiting frost consequences. Using the technique implies adjustments in the winter pruning calendar and training of the work crew. In medium to large vineyards, where postponing winter pruning may not be realistic, an alternative strategy would be to identify, based on experience and historic weather data, areas at greater risk of spring frost, and concentrate efforts on those. In spur pruning cordons, mechanical pre-pruning with an over-row disk machine in winter, followed by a fast manual shortening to the required spur length, is recommended.
Late pruning must be part of a broader strategy to improve risk assessment and prevent damage due to spring frost in vineyards. Additional work in cool climate areas is desirable to balance the existing knowledge gap between cool and warm/hot regions. On a more operational note, the best compromise between limitations imposed by the short time window for hand finishing and the need to carry out the technique on an entire vineyard still needs fine tuning. While spur pruning seems more suited to this approach than cane pruning, the ideal solution would be a single late, mechanical intervention to perform a short pruning in a previously untouched canopy.
A future ideal strategy would integrate delayed winter pruning into an action plan to be applied, pre- and post-planting, in vineyards now deemed at high risk for frost damage. The strategy must incorporate the following: i) reconsider cordon and cluster distance from the ground at planting and during vine training; ii) adapt floor management to maximize heat absorption during the day and reradiate more heat during the night and early morning hours; and iii) update current anti-frost irrigation systems using a micro-dripper mounted on top of each post that can nebulize water onto a narrow strip along the row, where the organs to be protected are located.
Footnotes
Acknowledgments: The authors wish to thank Tenute Ruffino (Tuscany) and in particular Maurizio Bogoni and Barmpa Despoina Maria for allowing vineyard access and performing data collection.
By downloading and/or receiving this article, you agree to the Disclaimer of Warranties and Liability. The full statement of the Disclaimers is available at http://www.ajevonline.org/content/proprietary-rights-notice-ajev-online. If you do not agree to the Disclaimers, do not download and/or accept this article.
- Received February 2022.
- Revision received April 2022.
- Accepted May 2022.
- Copyright © 2022 by the American Society for Enology and Viticulture. All rights reserved.
Literature Cited
Vol 73 Issue 4
