To survive the harsh conditions of winter, woody perennial species such as grapevine have adapted to use environmental cues to trigger physiological changes to induce dormancy, acquire cold hardiness, and measure the length of winter to properly time spring budbreak. Human induced climate change disrupts these cues by prolonging warm temperatures in fall, reducing the depth and consistency of midwinter, and triggering early budbreak through false spring events. We evaluated variation in dormant bud cold hardiness and chilling hour requirements of 31 different grapevine varieties over 3 years. Differential thermal analysis was used to track changes in cold hardiness and deacclimation resistance was assessed throughout the season to track dormancy progression. Results demonstrate wide variation in maximum deacclimation rate (1.03 – 2.87 °C/day) among varieties under forcing conditions. Absolute maximum rates of deacclimation show signatures of species-level responses to forcing temperatures. When integrated with variation in cold hardiness, these rates revealed a relationship between winter cold hardiness, changes in deacclimation rate and budbreak phenology. Standardizing rates among varieties as deacclimation potential demonstrated a conserved response to chilling exposure among varieties that alters our interpretation of the concept of high and low chill varieties and chilling requirement in grapevine.

Plant Sampling

This study was conducted over three consecutive winter seasons: 20 September 2018- 16 April 2019 (Year 1), 5 September 2019- 2 April 2020 (Year 2), and 2 October 2020- 31 March 2021 (Year 3). Thirty-one cultivars of grapevine were sampled from local commercial and experimental vineyard locations (Table S1) at approximately weekly intervals in each of the three years to evaluate cold hardiness in response to natural field conditions, deacclimation resistance, and budbreak phenology under controlled temperature conditions. Cultivars included both red (R) and white (W) fruited European Vitis vinifera cultivars (hereafter, vinifera); Cabernet Franc (R), Cabernet Sauvignon (R), Chardonnay (W), Gewurztraminer (W), Gruner Veltliner (W), Lemberger (R), Merlot (R), Pinot gris (W), Pinot noir (R), Riesling (W), Sangiovese (R), Saperavi (R), Sauvignon blanc (W), Syrah (W), and Tocai Friulano (W), and hybrid cultivars (hereafter, hybrid) with various wild species genetic backgrounds; Aromella (W), Cayuga White (W), Chambourcin (R), Chancellor (R), Concord (R), Corot noir (R), La Crescent (W), Marechal Foch (R), Marquette (R), Niagara (W), Noiret (R), St. Croix (R), Traminette (W), Valvin Muscat (W), Vidal (W), and Vignoles (W). These cultivars represent a subsample of the diversity of cultivars grown in the Northeastern United States and were chosen in part to also represent diversity in optimum climate niche. For the vinifera cultivars, we included both “cool” and “warm” climate varieties (Jones et al., 2012) and for hybrid cultivars, we used cultivars that represent both French American hybrids and New American hybrids, as well as hybrids derived from wild grapevine species that have Northern (V. labrusca or V. riparia), or Southern (V. rupestris or V. lincecumii) distributions. Distributions of wild species were obtained from county level data found at the Biota of North America Program (https://bonap.net/napa). Pedigree data for hybrid cultivars was retrieved from the Vitis International Variety Catelog (VIVC, https://www.vivc.de/). All cultivar information is included in Table S1.

Dormant cane cuttings with nodes 3-20 (depending on cultivar and pruning method) were collected from the field and transported to the bud physiology lab at Cornell AgriTech for processing. During each collection time point in 2018-2019, 40 nodes of each cultivar were collected from the field by randomly pruning from at least 4 different individual vines at each collection. In 2019-2020 and 2020-2021, sampling was increased to 50 nodes of each cultivar at each collection. Upon returning from field collections, dormant cane/node material were promptly pruned into single node cuttings, randomized, and distributed into subsamples for experiments to determine an initial cold hardiness assessment (n = 5; 2018-2019, n = 10; 2019-2020 and 2020-2021), deacclimation resistance (n =25), and budbreak phenology (n =10).

Assessment of cold hardiness using differential thermal analysis

Cold hardiness response curves for each year and each cultivar were produced using the mean LTE value for the weekly collection (T0) timepoints. For determining dormant bud cold hardiness levels, grapevine supercooling ability was assessed following standard practices for differential thermal analysis (DTA) (Mills et al., 2006; Ferguson et al., 2011; Londo and Kovaleski, 2017; Londo et al., 2023). This method uses thermoelectric modules to record voltage changes that are associated with the release of heat that occurs due to the phase change of water into ice (Mills et al., 2006). The LTE is associated with the lethal temperature for each sampled primary bud and is correlated with manual assessments of bud death (Wolf and Cook, 1994). Dormant buds were excised from cane tissue leaving an intact bud cushion layer of approximately 1 mm in depth. Buds were placed with the cut-side down on a single ply layer of moistened kimwipe (Kimberly-Clark Corp., USA) to induce ice nucleation during freeze tests (Londo et al., 2023). Buds were covered with a square of thermal open-cell foam and sample plates were placed in a programmable freezer (Tenney T2C, Tenney Environmental, New Columbia, PA, USA) connected to a Keithley 2700 or 2701 multimeter datalogger (Keithley Instruments). Voltage and temperature measurements were collected at 15-second intervals using the BudFreezer program (Brock University, Guelph ON, Canada). Freezing runs were conducted as follows: 1°C/minute cooling ramp to 0°C, hold for 1 hour, 4°C/hour cooling ramp to -40°C, hold one hour at -40°C, 4°C/hour warming ramp to room temperature. Low temperature exotherm peaks were manually annotated using BudProcessor/BudLTE software (Brock University, Guelph ON, Canada). In the first year, 2018-2019, 5 replicate buds were assessed for initial LTE measurements. In the second and third years of the study, 10 replicate buds were used for initial LTE measurements. This adjustment was made to increase the confidence in the mean LTE value as representative of field cold hardiness level. Here we report the mean LTE value for any given time point as the LT50.

Assessment of deacclimation resistance and deacclimation potential

To determine deacclimation resistance of the buds collected during weekly collections, twenty-five single node cuttings were placed in plastic cups with basal cut ends in water. Cups were placed in a Conviron growth chamber (Controlled Environments Inc., North Dakota, USA) and held at constant 20°C without light exposure. Cold hardiness was assessed as described above to measure the loss of cold hardiness under forcing conditions at repeated intervals (e.g. T0-field collected, T1 = T0+6 days, T2 = T0+12 days, ect). Resampling intervals varied throughout the experiments with longer intervals in early winter and shorter intervals in late winter, as deacclimation resistance declined. At each deacclimation time point, five replicate buds for each cultivar were assessed to determine the change in LT50 from the previous time points. Deacclimation rates for each cultivar at each collection time point were analyzed independently using linear regression and taking the slope of the regression as the deacclimation rate (e.g., lm(LTE~SampleDay). To compare deacclimation rates between cultivars as a function of chilling hour estimates, raw rate values for each year were normalized within cultivars using the mean of the three highest rate values observed across the study. The normalized percentage of total deacclimation is referred to as deacclimation potential (Ψ_deacc). (Kovaleski et al., 2018; Kovaleski, 2022; North et al., 2022). The nls() function of the drc package (Ritz et al., 2015) was used with default parameters (fct=LL.2(), type="continuous") to model Ψ_deacc between cultivars, years, and in comparison with the various chill model estimates as described in North (2022).

Assessment of budbreak phenology

In tandem with field LTE assessments, an additional 10 single node cuttings were placed in plastic cups with basal ends trimmed and kept in water. Cups were left at 18-20°C temperatures and without specific light exposure (laboratory lighting). Budbreak phenology was manually assessed on alternating days during the week. Days to bud break (DTB) was recorded as the number of days between field collection and first evidence of green tissue, EL stage 3 (Coombe, 1995). When a cutting was marked as EL 3, they were removed from the experiment to prevent double counting. Budbreak data was manually curated to remove sample collection points where less than 5 of the 10 replicates achieved budbreak in under 150 days. For the remaining collections, budbreak was phenotyped as the average (meanDTB) time to budbreak.