Temperate woody perennials generally form buds during summer and fall that contain new leaves and flowers for the following growing season. These buds must then overwinter, remaining dormant and attaining cold hardiness. In the spring, loss of dormancy and cold hardiness – modulated by chilling accumulation that occurs during winter, and precedes the visual display of spring phenology.

Using a previously published cold hardiness model (NYUS.1), we predicted cold hardiness of grapevine (Vitis spp.) buds based on temperature. Cold hardiness trajectories were then used to predict timing of budbreak. The model was applied to temperature records without training, and predictions were validated using a spring phenology dataset comprised of 329 datapoints (Year x Location x Cultivar combinations) from three grapevine cultivars (V. vinifera ‘Cabernet Sauvignon’ and ‘Riesling’, and Vitis hybrid ‘Concord’) from eight locations (one in Canada, four in France, one in Germany, and two in the United States).

The RMSE of the predictions was 7.3 days (Bias=-0.83), despite no training of the model for this dataset (or data type). We then identified years with damage, and predictions of damage occurrence were validated based on newspaper and extension records. Prediction of timing of budbreak was then corrected based on expected delays that occur when tissue is damaged by cold, resulting in an improvement of predictions (RMSE=7.2d, Bias=0.58). Using predicted budbreak timing across locations and years, we evaluated the sensitivity of budbreak to mean dormant season temperature (MDST; 1 Nov – 30 Apr). Sensitivity analysis showed a general advancement of phenology at -5.8d/°C. However, in locations and years with MDST>10°C, delays are expected in phenology at +1.9d/°C.

Cold hardiness dynamics link the fall and spring seasons through winter, and can be used to accurately predict phenology in woody perennials.