Abstract
The cold hardiness of 33 different grapevine genotypes, representing six wild North American grapevine species, one wild Asian grapevine species, and six hybrid grapevines, was evaluated by measuring lethal temperatures for dormant buds using low temperature exotherms. Studies were conducted over three different winters to characterize the relative level of cold hardiness and responsiveness to changing weather patterns of each species. Major differences in the winter conditions demonstrated that wild grapevine has a great capacity for responding to both warm and cold temperature events during the dormant season. Results indicated that wild grapevine species with northern distributions tended to exhibit greater cold hardiness and responsiveness to temperature fluctuations than their southern-dwelling counterparts. Statistical modeling of low temperature exotherm results across the three winters indicated that each grapevine species had a distinct and innate capacity for responding to temperature. These results may be important for future grapevine breeding in areas with expected increases in winter temperature variation. Additionally, these results demonstrate the potential for genetic determinants of temperature responsiveness that may be investigated and mapped for future grapevine improvement.
Low temperature is the most important factor governing plant distribution (Parker 1963). Most cultivars of the European grapevine Vitis vinifera are not particularly cold hardy and suffer from freeze damage when temperatures decrease below ~−15°C during the winter, and are therefore not well-suited to grapevine production outside of Mediterranean climates (e.g., eastern United States). However, the grapevine industry continues to grow and expand rapidly in regions of North America where winter conditions limit productivity and sustainability (MKF Research 2007, Tuck and Gartner 2014). Wild North American grapevine species, particularly species native to the north and northeastern parts of the US and Canada, carry adaptations that allow survival in winter temperatures as low as −35 to −40°C. Growth of the grapevine industry outside of California and the Pacific Northwest relies in large part on the breeding and development of cold hardy hybrid varieties made from crosses between V. vinifera and these wild grapevine species. In fact, the economic value of cold hardy hybrid grapes reached $401 million in 2011 (Tuck and Gartner 2014) and the value of hybrid grapes is likely greater today because grape production and winery establishment in the eastern US has grown since 2011. Cold hardy hybrid grape varieties are a result of the breeding efforts of many different public and private grape breeders, but all share the use of wild or wild-related germplasms as genetic sources for traits of cold hardiness.
Protection from winter freezing damage of the bud is essential for grapevine production because inflorescence structures are patterned within the dormant bud (Goffinet 2004, Zabadal et al. 2007), and death or damage of the primary bud causes a loss of clusters in the subsequent growing season. To protect this critical organ, grapevines initiate a transition into dormancy at the end of the growing season and begin a cold hardiness process called acclimation to prepare for freezing temperatures. Photoresponse and decreasing temperatures are important aspects of this acclimation process (Schnabel and Wample 1987, Fennell and Hoover 1991, Grant et al. 2013). Differences in the degree of photoresponse have been observed between cultivars and grapevine species, and northern species of wild grapevines have been found to be more responsive than V. vinifera (Fennell and Hoover 1991, Garris et al. 2009). Photoresponse is sufficient to initiate the transition from growth to endodormancy, which is the suspension of growth due to physiological factors within the bud (Lang 1987), but studies have also shown that cooling temperatures enhance the acclimation process (Schnabel and Wample 1987, Fennell and Hoover 1991, Wake and Fennell 2000, Grant et al. 2013). Interestingly, acclimation continues long after the leaves have senesced and dropped from the vine, suggesting that the characteristics of temperature tracking and response also lie in the dormant bud. The transition from endodormancy to ecodormancy after fulfilling the chilling requirement also suggests that temperature tracking occurs in the bud (Londo and Johnson 2014). Ecodormancy is the suspension of growth due to environmental factors (e.g., cold temperatures) (Lang 1987).
Previous studies of dormant grapevine buds and grapevine cold hardiness demonstrate a fairly standard pattern in response to winter temperature (Pool et al. 1990, Mills et al. 2006, Ferguson et al. 2011, 2014, Dami et al. 2016), described as a U-shaped curve with gradual acclimation in early winter as temperatures decrease, maintenance of cold hardiness in midwinter, and rapid deacclimation as temperatures rise in late winter/early spring (Zabadal et al. 2007). Throughout these stages of winter, the grapevine bud is hypothesized to utilize a process called deep supercooling (Sakai and Larcher 1987). Physiologically, deep supercooling is the ability of the dormant bud to survive sub-freezing temperatures. Mechanistically, deep supercooling in grape is not well understood, but studies in other systems have indicated that a number of complex mechanisms may contribute, including the presence of ice nucleation barriers in different tissues (Wisniewski et al. 2003), bud dehydration to remove free water (Endoh et al. 2009), concentration of cellular osmolytes (Grant and Dami 2015), cryoprotection through protein stabilization to control ice damage (Fernandez-Caballero et al. 2009), and inhibition of ice crystal growth through ice-binding antifreeze proteins (Kuiper et al. 2015). Despite its complexity, the underlying hypothesis of supercooling in the grapevine bud is that the bud retains a minimal amount of intracellular free water to prevent irreversible dehydration, while also allowing for the safe freezing of extracellular water. The V. vinifera cultivars with the greatest cold hardiness typically supercool and survive winter temperatures as low as −25°C, while some wild grapevine species can supercool and survive in temperatures as low as −40°C when completely acclimated. The process of supercooling is advantageous under a normal, cold winter climate if temperatures do not decrease below the genotype-specific freeze point. However, when temperatures drop below the supercooling point, intracellular water is hypothesized to freeze dynamically, leading to lethal cell injury and bud death.
Historically, evaluating death and damage caused by freezing injury was conducted through manual dissection of buds and evaluation of live tissue following freezing events (Wolf and Cook 1994). However, differential thermal analysis (DTA) of grapevine bud tissue (Pierquet et al. 1977, Pierquet and Stushnoff 1980, Wample et al. 1990) has become the standard method for evaluating cold hardiness of the bud and allows for a sequential evaluation of freeze resistance capacity throughout the winter. DTA involves using thermoelectric modules to record the release of heat due to phase change when water rapidly transitions into ice, which corresponds to when the grapevine bud supercooling point is exceeded and is referred to as low temperature exotherm (LTE), which signals the death of the dormant bud tissues (Pierquet and Stushnoff 1980, Wample et al. 1990). Studies have shown that LTE measurements correlate with field-based assessments of damage and death in grapevine buds (Wisniewski 1995). Extensive studies of the cold acclimation patterns, bud hardiness, and deacclimation processes in cultivated and hybrid grape cultivars (Pool et al. 1990, Mills et al. 2006) have resulted in development of a predictive model (Ferguson et al. 2011, 2014) of bud cold hardiness that works well for 23 different cultivars of grapevine in the Pacific Northwest. The model predicts daily shifts in LTE based on the temperatures experienced by separating the process between endo- and ecodormant buds, assigning different temperature thresholds from which either chilling or heating degree-days are accumulated, and calculating different acclimation and deacclimation rates. However, this model does not appear to capture all variables that affect the LTE response of cultivated grapevines in the northeast (Londo, unpublished data, 2017). Additionally, no comprehensive time series-based data exist for wild grapevine species, and thus the model cannot be used currently with wild grapevine germplasm or new hybrid varieties.
All wild and hybrid grapevine genotypes housed at the United States Department of Agriculture (USDA) cold hardy grapevine germplasm exhibit greater cold hardiness than the majority of documented V. vinifera varieties (Pool et al. 1990). Thus, the USDA cold hardy germplasm is a major source of parental germplasm for many grapevine breeding programs that aim to develop cold hardy cultivars. Many grape breeders focus on breeding schemes incorporating the northern distributed species V. riparia and V. labrusca as cold hardy parents based on observations of excellent cold hardiness in these species, and in part on the assumption that cold hardiness is an adaptive trait. Pool et al. (1990) examined a suite of hybrid and wild genotypes held in the germplasm using DTA analysis. Their study characterized wild and hybrid genotypes at a single time point during winter, but did not comprehensively examine how LTE values change over time. As winter kill events can occur at any time during the winter, understanding how wild and hybrid grapes respond to changing temperatures throughout the winter is important. The threat of climate change is another concern facing grapevine production and sustainability in the eastern US. Climate change is expected to increase winter temperature variability (Vasseur et al. 2014; and references within, Williams et al. 2015), and occurrences of unseasonable cold or warm events may dramatically alter the pattern of cold hardiness during the winter, increasing risk of cold damage at all points in the dormant season.
The objective of this study was to characterize the dormant bud cold hardiness ability of different genotypes of many wild grapevine species throughout the winter season. The studies were conducted over several years to improve understanding of the acclimation and deacclimation capabilities of wild grapevines in different winter conditions. This information may be useful for aiding grapevine breeders and growers to determine which wild species may be used for future varietal development.
Materials and Methods
Sampling design
Thirty-three unique grapevine genotypes were sampled throughout the course of this experiment and represent six North American wild grapevine species (V. aestivalis, V. cinerea, V. labrusca, V. riparia, V. rupestris, and V. vulpina), one wild Asian grapevine species (V. amurensis), and examples of interspecific hybrid grapevines held at the USDA germplasm. Wild genotypes were chosen with the widest geographic distribution represented in the germplasm based on their associated passport information (www.ars-grin.gov). A list of germplasm identities and species is shown in Table 1. Weather data was compiled from the nearby (~4.8 km) Network for Environment and Weather Applications station (www.newa.cornell.edu) located on the New York State Agriculture Research Station, Research North experimental farms.
List of surveyed wild and hybrid grapevine genotypes. GRIN ID indicates the United States Department of Agriculture national plant germplasm designation. GEO indicates geographic information obtained from passport information.
Grapevines were sampled biweekly from the USDA cold hardy grapevine germplasm following the first frost of the fall season. In each of the sampled years, the first frost occurred within the first two weeks of November, and assessments of LTE began immediately after collection. Collections were terminated when evidence of budbreak was observed in the germplasm. Sampling dates occurred from 6 Nov 2012 to 31 March 2013 in year 1, 14 Nov 2013 to 17 April 2014 in year 2, and 14 Nov 2014 to 1 April 2015 in year 3.
DTA and LTE analysis
Measurements of LTEs were conducted using DTA of dormant grapevine bud tissues, a standard practice for dormant grapevine buds that supercool during the winter season (Mills et al. 2006). Freezing experiments were performed using sample plates containing nine thermoelectric module (TEM) chambers (Mills et al. 2006). Nine buds were sampled at each time point and grapevine genotype. Buds were excised from cane tissues using a razor blade, leaving the bud cushion attached to the bud. All nine buds were placed, cut surface down, on lightly moistened Kimwipe tissue (Kimberly-Clark) within a single TEM chamber. In our experience, moistened tissue enables a more consistent high temperature exotherm and cleaner peaks for LTE when close to −10°C. For some genotypes, insufficient vine vigor or lack of acclimation resulted in an insufficient number of buds for full sampling of nine buds at all winter time points. In these cases, bud number was assessed at the beginning of sampling, and the time points and genotypes were included if at least three buds could be sampled.
Previous studies examining cultivated varieties typically examine the hardiness of fruitful buds found in positions 2 to 10 on the cane (Pool et al. 1990, Mills et al. 2006, Ferguson et al. 2011, 2014). The USDA grapevine germplasm is maintained as only two clones per genotype and therefore lacks the number of buds at positions 2 to 10 for a season-long sampling design. Therefore, buds beyond position 10 were included in this study. Prior to the experiment, preliminary tests were conducted on long canes of wild grape to examine the relationship between level of bud hardiness and distance from the trunk. Results indicated very little variation in LTE in the first two-thirds of the cane with increasing LTE values in the terminal one-third. Thus, the terminal one-third of each cane was removed in the field and not sampled to prevent spurious measurements associated with improperly acclimated tissues. At each time point, dormant buds were sampled from both clones, and nodes were collected from basal and medial positions. Freezing runs were conducted in a Tenney Model TC2 programmable freezer (Thermal Products Solutions), using a temperature ramp from room temperature to 4°C, held for 1 hr, ramped to −40°C over 13 hr, and followed by a slow ramp back to room temperature. This program results in a cooling rate of ~0.06°C/min or 3.4°C/hr. The LTE peaks were recorded using a Kiethley 2700 (Tektronix, Inc.), and data results were manually curated in Microsoft Excel (2013). Up to nine primary bud peak LTEs were identified for freezing runs. Secondary and tertiary peaks, representing the LTE for secondary and tertiary buds, respectively, were sometimes observed but not recorded. In cases where nine peaks were not detected, only obvious peaks were included. From the measured peaks, the LTE values reported here were determined by taking the mean of the peak values, often referred to as the lethal temperature for 50% of the buds (LT50) (Wolf and Pool 1987), reported with standard errors.
Statistics
A statistical model was developed using multiple linear regression in R (ver. 3.3.0, R Foundation for Statistical Computing) to determine significant differences between species in LTE curves throughout the winter and responsiveness to temperature fluctuations. This regression model was developed using a 1-year subset to avoid bias and further trained on all three years. The explanatory variables used were time, time2, species, and a temperature index (σT). The time was measured in days from 31 Oct since the LTE measurements started in November of each year. The temperature index σT was created to describe shifts in temperature during a time period preceding the biweekly LTE data points. Briefly, the temperature index (σT) was similar to a standard deviation from a base temperature (Tbase) following the formulas:
where TE is the temperature experienced by the plant and T is the hourly temperature. The Tbase used was 0°C, which was found to be the approximate Tbase for several species (Kovaleski and Londo 2016). The ‘n’ used was 168, indicating that the temperature index was a sum of the temperature variations over 168 hours (seven days) prior to the LTE measurements. The square of TE was obtained by multiplying the value of TE by its absolute value in order to keep the sign. A similar artifact is used when calculating the square-root of
to keep the sign. By using the square of TE, great variations from the Tbase resulted in more influential points.
An initial stepwise selection with a Bayesian information criterion (BIC) correction was used to select a regression model using a null model with an intercept only and a full model with all interactions possible. This was performed using data from a single year (2013 to 2014), and therefore a year term was not included. The BIC correction was chosen due to the high number of sampling points. To use the model with the full dataset (three years, 2012 to 2015), a new stepwise selection with BIC correction was performed with the null model as the regression model previously found and the full model as the interaction between the terms in the null model with year. From this new regression model, data points with studentized residual ≥2 or Cook’s distance ≥0.002 were considered outliers and removed from the dataset. After this, the regression model was re-fit using the non-outlier subset to obtain the final coefficients. Dominance analysis was performed to evaluate the relative contribution of each parameter.
Results
Characterization of winter conditions
Each of the three winter seasons that occurred during this study differed in temperature intensity and consistency (Figure 1). As such, discussion of the measurements of cold hardiness must be placed in a yearly context. Year 1 (2012 to 2013) was a generally mild winter with a few sub-freezing events in midwinter and a minimal cold temperature of −18.6°C. Year 2 (2013 to 2014) was a harsher winter with many acute and severe cold events beginning in early December and continuing throughout the year to late March. Two extreme events occurred with temperatures reaching −24.3°C in early January and −25.5°C in late January. Year 3 (2014 to 2015) was also a harsh winter but differed from year 2 in that the early winter was normal/mild in temperature, but the late winter (beginning in early January) was very cold with sustained temperatures below freezing throughout the day and night until early March. Minimal temperatures in this year were −23.7°C in mid-January and −25.9°C in late February.
Daily max and min temperatures from 1 Oct to 30 April. Temperature scale in Celsius is on the left axis. Red line denotes freezing temperature, and blue shading indicates time spent below freezing. Dashed orange line indicates the index (σT) used to describe the effects of temperature exposure on low temperature exotherm. Note secondary axis for this index.
LTE response of wild grapevines
Wild and hybrid accessions evaluated in this experiment exhibited LTE responses that largely protected the vines from damage during winter. The LTE values for all species were lower (greater cold hardiness) in 2013 to 2014 and 2014 to 2015 than in 2012 to 2013, likely due to the greater exposure to cold temperatures during those years. Figure 2 illustrates the average response of each wild grape species in each of the evaluated years. Regarding LTE response, all species shared the standard U-shaped cold hardiness curve of early winter acclimation and late winter/early spring deacclimation profiles. However, differences were noted during midwinter in the movement of LTE values in response to temperature fluctuations. Looking across all three winters, the northern species V. riparia and V. amurensis demonstrated the greatest response to cold temperatures and exhibited the lowest LTE values. However, the timing of greatest hardiness differed between years for all species (specific yearly observations are described below). Graphs of mean LTEs and standard error for all collection points for each genotype are presented in Supplemental Figure 1.
Average low temperature exotherm (LTE) values for seven wild grapevine species and hybrid genotypes. Temperature scale in Celsius. Black trace indicates daily minimum temperatures over the course of sample collection for the three years of this study. Right panels indicate zoomed-in version of LTE values to clarify the level of variation observed within each species during midwinter. Sample points indicate average LTE measure and error bars indicate ±1 SE.
2012 to 2013, mild winter
At the start of sampling in 2012 to 2013, species LTE values and bud hardiness levels were already at −17.3 to −20.4°C. Cool, but not cold, conditions prevailed during early winter (Figure 1) and thus, LTE values gradually decreased for all species until early January (mean decrease in LTE of −7.26°C). The mild beginning of winter can also be seen in the temperature index σT, which remained positive until late December. Maximal hardiness was measured for V. amurensis (−26.5°C) and V. cinerea (−26.8°C) in early January, for V. labrusca (−26.2°C) and V. riparia (−27.7°C) in early February, and for V. aestivalis (−27.7°C), V. rupestris (−27.2°C), and V. vulpina (−26.2°C) in late February. Hybrid cultivars exhibited greatest hardiness in early February (−27°C).
In early January, a warm weather event occurred and rapid deacclimation was observed in the V. amurensis vines in mid-January, resulting in a loss in hardiness of 5.4°C. Therefore, when the coldest event of 2012 to 2013 occurred near the end of January, V. amurensis buds were much less hardy and the lethal temperature was nearly reached. V. labrusca and V. riparia also demonstrated midwinter deacclimation, losing 3 and 2°C of hardiness, respectively. The rapid deacclimation was also observed toward the end of the season as temperatures rose with the beginning of spring (Figure 2).
2013 to 2014, cold punctuated winter
In 2013 to 2014, bud hardiness levels were slightly greater than the previous year (−19.2 to −24.6°C). Early winter had temperatures that were below freezing and several oscillations between subfreezing and warming events (Figure 1). During some of the periods, temperatures oscillated enough to change σT from negative to positive. A trend of a wider spread between species in LTE values was observed during this year. As a result, deacclimation responses were observed in late December with all species except V. labrusca, which remained unchanged. A strong cold event with min temperatures of −24.3°C occurred in early January, and all species responded by quickly acclimating, gaining between −2.5 and −4.9°C of increased hardiness. Because of the cold event, all species achieved max hardiness in early January that were greater than in 2012 to 2013, demonstrating the effect of varied winters on maximal hardiness. Measures from least to greatest hardiness were as follows: V. aestivalis (−27.7°C), V. cinerea (−28.1°C), V. vulpina (−28.1°C), V. labrusca (−29.1°C), V. riparia (−30.4°C), V. rupestris (−30.4°C), and V. amurensis (−31°C). Hybrid genotypes had a mean hardiness of −27.4°C, which was slightly greater than in 2012 to 2013.
Quickly following this cold event, winter temperatures increased and remained briefly above freezing, resulting in a loss of hardiness in all varieties (~1.5 to 5.3°C), with the greatest deacclimation observed again in V. amurensis. Following this midwinter warming, a second acute cold event occurred with temperatures reaching −25.5°C. Even though all species had previously acquired sufficient hardiness to avoid damage during this event, the preceding deacclimation resulted in damage to buds of V. hybrid genotypes and partial damage to all wild species except V. rupestris, V. vulpina, and V. riparia. LTE values varied following these temperature swings but remained low enough to avoid the subsequent minor cold events during this winter (Figure 2). Once temperatures trended upward toward the end of winter, all species deacclimated quickly, with the fastest rates of deacclimation observed in V. amurensis, V. riparia, and V. labrusca.
2014 to 2015, cold sustained winter
The winter of 2014 to 2015 was also very cold for the northeast. However, the pattern of temperatures throughout this winter was different than that of the previous year. The early winter was mild and LTE values were much higher, between −15.0 and −21.4°C, than the preceding year for all species. After late December, temperatures dropped and remained mostly below freezing until early March (Figure 1), which also caused σT to remain negative. The trend observed in 2013 to 2014, of wider variability in LTE values among species, was also observed in 2014 to 2015, and no major deacclimation events were observed. One exception was an apparent 4°C deacclimation in a V. cinerea in late December and a 1°C deacclimation in V. riparia, which seemed to correspond with the last major warm event of the winter. Winter temperatures were cold and sustained, and as a result, LTE values remained low for all species, protecting them from damage due to a cold event (−23.7°C) in early January. A second cold event occurred in mid-February, which led to decreased LTE levels in all species and resulted in the greatest level of hardiness observed in this study. All species converged on greatest hardiness in this year on 20 Feb 2015 (Figure 2). Measures of least to greatest hardiness were as follows: V. aestivalis (−29.8°C), V. rupestris (−31.1°C), V. cinerea (−31.3°C), V. vulpina (−31.5°C), V. labrusca (−31.8°C), V. amurensis (−32.6°C), and V. riparia (−34°C). V. hybrid accessions reached a mean cold hardiness of −30.9°C, 3°C lower than the previous year.
Determining species level differences in LTE response
Variation in maximal hardiness differed between years, within years, and among species. To better characterize the capacity for wild grapevine species to react and respond to changing winter temperatures, hourly temperature data were used to create a regression model for responsiveness of species’ LTE. The multiple linear regression obtained had a p < 0.0001 and adjusted-R2 = 75.6%. Within the eight interaction terms, only three were interactions with species (time2, σT, and intercept). The final model resulted in equations with five parameters and intercepts for each species × year (Supplemental Table 1). Because the temperature index σT was not the only significant parameter, σT did not explain all the variation, resulting in an effect of time. More importantly, σT had an interaction with species but not with year, demonstrating that the response to temperature was species-dependent and an actual measurement of the biological response to temperatures. V. amurensis was significantly more responsive to temperatures (greatest estimate for σT) than the other species, while V. aestivalis, V. rupestris, and V. vulpina were the least responsive (Table 2), and V. riparia, V. labrusca, and V. cinerea had an intermediate response.
Estimates for the temperature index (σT) parameter of different Vitis species from a temperature responsiveness model. Estimates followed by different letters are significantly different using a t-test with α = 0.05.
Discussion
The results of this study demonstrated the dynamics and complexity of the grapevine “cold hardiness” phenotype. LTE values changed in response to temperature changes and differed greatly depending on the time of the sample and year in which they were evaluated. As such, assessment of LTE in different winters with diverse temperature patterns is essential for grapevine breeders to choose the most suitable germplasm material for their regions. Additionally, these results demonstrated the necessity of evaluating cold hardiness as a continuous time series-based assessment.
The LTE values changed throughout the sampled winters, generally increasing and decreasing in response to fluctuating temperatures, and also reached maximal hardiness at different times of each year. The statistical model created to evaluate the effects of temperature exposure on LTE changes resulted in a good description of the overall behavior of the species in the three years studied (adjusted-R2 = 75.6%), although additional variables remain to be validated. When performing a dominance analysis, the temperature index σT had a relative importance of 59.0%, followed by time2 and time (13.8% and 12.7%, respectively), demonstrating that σT is greatly responsible for the LTE phenotype in the model. Determining the aspects of daily temperature that contribute the most to changing LTE values is important to predict how a given winter is impacting cold hardiness. For example, using mean daily temperatures may misrepresent the actual temperature exposure that a vine receives because it overlooks the effect of wide swings in daily temperature. This contribution of hourly temperature data may be overlooked when a high density of data points is collected, such as in the daily LTE observations of Ferguson et al. (2011). The Ferguson et al. (2011) model uses the LTE of the previous day as a parameter in their estimation, and that value carries the “historical” weather data from the beginning of the season. Ebel et al. (2005) used a different approach to predict cold hardiness of Citrus unshiu ‘Satsuma’, in which they used 500 hours (~3 wks) of temperatures prior to estimate the critical temperatures for injury. Because daily observations were not possible due to scarcity of plant material in the present study, the temperature index σT was calculated using hourly temperature data for 3, 7, and 14 days prior to the date of buds collected for LTE, and σT for 7 days was shown to have the best correlation with the data.
Unlike the predictive models for LTE established by Ferguson et al. (2011, 2014), the objective of our study and the regression model used was to describe the responsiveness of the grapevine species to temperature within the limits of our data. However, the measure of responsiveness that we used may be related to the acclimation and deacclimation rates (ka and kd, respectively) in Ferguson’s model, in which greater values result in higher acclimation and deacclimation. Our results demonstrated that this responsiveness to temperature varies significantly among species (Table 2). The significant interaction with time and time2, but not with species, shows that responsiveness increases with time and is likely a result of increased responsiveness to temperatures once the transition from endo- to ecodormancy occurs, although the regression did not include a term for dormancy state specifically. Similarly, the model by Ferguson et al. (2011, 2014) has different parameters, ka and kd, for endo- and ecodormancy. More importantly, the deacclimation rate kd is generally higher in the ecodormant state than in the endodormant state. Additionally, the effects of time, time2, and interactions are likely due to the different timing of winter onset, which can also be seen in the different intercepts for each species and each year. These terms may also describe the chill accumulation during dormancy and heat accumulation after the transition from endo- to ecodormancy, when buds are responsive to higher temperatures and may resume growth. Studies are ongoing to assess the relative contribution of chill accumulation, heat accumulation, solar radiation, humidity, and other environmental factors on LTE.
To investigate how the rates of acclimation and deacclimation shifted due to the temperature, we used the directional derivatives of the partial slopes of σT and their interactions with time and time2 in the model over the three winters studied (Figure 3). Using these rates, we identified the variations during the winter of acclimation (negative rates: increasing cold hardiness) and deacclimation (positive rates: decreasing cold hardiness). In all three years, the decreasing temperatures enhanced the early acclimation in late fall, although this was affected by other factors (i.e., significant term for time and time2) such as decreasing day length (Schnabel and Wample 1987, Fennell and Hoover 1991). Spring temperatures led to deacclimation, which was also consistent across the years. In 2015, however, a cold event in late March resulted in a move towards acclimation based on temperature. In the midwinter, the rate of LTE change appeared to approach 0°C/day, only varying toward negative or positive values during cold and warm events, respectively.
Plot of directional derivatives used to compare low temperature exotherm (LTE) responsiveness of Vitis amurensis (most responsive), V. riparia (moderate responsive), and V. aestivalis (least responsive). Lines depict the rate of LTE change in response to temperature exposure. A, rapid acclimation in 2012 in response to early frost; B, D, and H, increased rate of acclimation in response to cold shock in 2012, 2013, and 2014; C, E, and G, increased deacclimation in response to midwinter warming events; F and I, rapid deacclimation in 2014 and 2015 in response to spring warming. In each case, V. amurensis responded at a much greater rate and subsequently, was at higher risk for cold damage following warming events.
The three species shown in Figure 3 are from the three different classes of temperature responsiveness (Table 2). V. aestivalis, after achieving a certain degree of hardiness, showed little variation in midwinter due to temperature, while V. amurensis continued to be very responsive to temperature swings over the duration of winter. Figure 3 is also a graphic representation of how V. amurensis has a significantly faster acclimation rate in the fall and deacclimation rate in the spring, while V. riparia has an intermediate rate and V. aestivalis has the slowest rate.
Our results demonstrate the fluid nature of bud cold hardiness and the challenges associated with assessing cold hardiness in grapevine. In general, these results validated previous studies suggesting that northerly-distributed wild grapevine species have greater cold hardiness. Among the species studied, V. riparia, V. labrusca, and V. amurensis had the greatest levels of hardiness in all three winters. Additionally, our data showed that all three northern species were also more responsive to midwinter temperature fluctuations, with V. amurensis being the most responsive species of those tested, losing and gaining bud hardiness significantly more than all other species. These responses seem to correlate to our previous assessments of the chilling requirement and endo-/ecodormancy transition in wild grapevine species (Londo and Johnson 2014). Northern species tend to have lower chilling requirements to satisfy dormancy and resume growth in the spring. Adaptively, responsiveness to temperature may be associated with the shorter growing season experienced by plants in higher latitudes. Following a cold winter, rising temperatures should signal the approach of spring and the time for northern species to quickly resume growth in order to grow and ripen seeds before the next dormant season.
It should be noted that the southern species were only slightly less hardy than northern species and easily survived even the coldest winter in our study. This was perhaps not too surprising, as these species have been maintained in the USDA germplasm for many years. However, determining their relative level of hardiness was important in case they were shown to be more advantageous for breeding purposes. Southern species were significantly less responsive to temperature swings during winter than were the northern species, which may be due to an adaptive and evolutionary history associated with southern species. Moving south, winter conditions generally become milder but the stability may also decrease. Southern species may have evolved to resist these swings in temperature by retaining deeper dormancy and resisting deacclimation during midwinter. In contrast, northern species may have evolved under winter conditions that were sufficiently harsh and consistently cold such that changes in historical weather patterns are insufficient to trigger deacclimation. Better geographic sampling of northern and southern species is needed to address this adaptive trait hypothesis in grapevines. Li et al. (2002) found similar results in Betula pendula, in which northern ecotypes acclimated faster under low temperature and short day exposure than did southern ecotypes, although the final hardiness of northern ecotypes was greater. Schwarz and Reaney (1989) also found that northern ecotypes of C4 grasses acclimated to cold faster than did southern ecotypes, but both reached the same levels of freezing tolerance. Kalberer et al. (2007) showed that some northern species of deciduous azaleas (Rhododendron spp.) exhibited increased cold hardiness along with high deacclimation rates. The authors speculated that the northern species would have a lower chilling requirement and therefore faster deacclimation once temperatures increased in the spring. Myking and Heide (1995) observed that northern ecotypes of B. pendula and B. pubescens had faster budbreak when compared to southern ecotypes. Although cold hardiness was not evaluated, the faster budbreak suggests a faster loss of hardiness. Our observations also suggest that DTA evaluations of species material and hybrids should be conducted in grapevine growing regions of the US with more extreme winter conditions to evaluate whether freeze resistance levels are maintained under harsher winter conditions. For example, winter conditions in the northern and midwestern regions of the US are harsher than in New York, and it is possible that under greater levels of stress, species responses could diverge from those presented here.
The responsiveness trait is undesirable in an age of increased winter temperature fluctuation, as it exposes vines to rapid deacclimation during midwinter warming events when subsequent acute cold is highly likely. Evidence of these variable winters can be seen within our own data set by contrasting the 2013 to 2014 winter with the 2014 to 2015 winter (Figure 2). The most acute temperature drops during each year corresponded to arctic weather patterns (i.e., polar vortex) moving through the region. In 2013 to 2014, the rapid shifts between warm and cold weather patterns led to damage in the wild vines used in this study and corresponded to widespread damage of local V. vinifera and V. hybrid production vineyards (T. Martinson, personal communication, 2014). In contrast, vines had been consistently exposed to cold temperatures when equally cold events occurred in the 2014 to 2015 season, which resulted in no observed damage in our study and reduced damage was recorded in locally cultivated varieties. Although the hypothesis that climate warming could lead to greater cold damage events may seem paradoxical, our study demonstrated that winters with sustained colder temperatures pose a lower risk than those with warmer but fluctuating temperatures.
Our study demonstrated that breeding with northern wild grape species may increase the deepest possible level of bud hardiness (lowest LTEs) in hybrid crosses, but may also introduce enhanced midwinter responsiveness as a negative phenotype. Material from southern species may be essential in future breeding efforts to buffer hybrid varieties against temperature variation during winter. Maximal hardiness traits found in V. amurensis, V. riparia, and V. labrusca could be combined with traits for reduced temperature response from V. rupestris, V. vulpina, or V. aestivalis. However, whether genetic and phenotypic variation within a given species can be exploited for breeding efforts remains to be determined. For example, V. riparia, V. labrusca, and V. aestivalis have wide latitudinal distributions, as they have both northern and southern populations. Comprehensive evaluation of within-species variations in LTE and responsiveness are needed to determine if there are specific genotypes in which these traits are not linked.
Another important result is that we also observed that maximal cold hardiness was not automatically reached each year, as vines did not acclimate to their maximal potential in every year. The greatest hardiness was only observed in the third year, which had a cold sustained winter, in all species, indicating that biological, energetic, or physiological cost may occur with achieving deep midwinter hardiness. Although this hypothesis remains to be tested, it illustrates the need for data points collected over many different winters and temperature patterns to determine how a genotype responds during winter. The timing of maximal hardiness also varied based on the type of winter. In the warm winter, different species achieved maximal hardiness at various time points, demonstrating the effects of varied winters on making predictive assessments of cold hardiness complex.
Conclusion
This study represents the most comprehensive evaluation to date for wild grapevine species germplasm and dormant bud cold hardiness, and demonstrates the dynamic nature of this trait. The temperature-based statistical model presented here used the complex pattern of changing LTE values to identify three classes of responsiveness in the wild grapes. Northern grapevine species were validated as possessing the greatest ability to survive deep freezing temperatures (lowest maximal LTE), but may be at a higher risk of rapid deacclimation and freeze damage in variable winter conditions. The hardiness capacity was nearly equal in southern and northern species, although the southern species had lower responsiveness to temperature changes. The evaluations here demonstrate the challenges associated with breeding for cold hardiness and determining which hybrid varieties will be successful in a wide array of climates.
Acknowledgments
All research was supported by United States Department of Agriculture (USDA) appropriated funds for project number 1910-21220-006-00D. The present study was also partially supported by CAPES, Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, Brazil. The authors would like to acknowledge and thank the following people for contributing to the field collection and processing of dormant bud tissues used in this study: Kathleen Deys and Greg Noden of USDA-ARS-GGRU and Bill Srmack, John Keeton, and Bob Martens from Cornell University. The authors also thank Dr. Tim Martinson, Cornell University, for use of freezer facilities and for access to cultivated and hybrid grapevine samples, and Kevin Packard from the Cornell Statistical Consulting Unit for assistance with statistics.
Footnotes
Supplemental data is freely available with the online version of this article at www.ajevonline.org.
- Received September 2016.
- Revision received December 2016.
- Accepted December 2016.
- Published online December 1969
- ©2017 by the American Society for Enology and Viticulture
Literature Cited
Vol 68 Issue 2
Jump to section
Related Articles
Cited By...
More from this TOC section
Similar Articles
