Evaluation of temperature models for predicting bud burst in Norway spruce

Spring bud development was assessed on cuttings of 17 Norway spruce (Picea abies (L.) Karst.) clones for up to 7 years at two sites in southern and central Sweden. The ability of various temperature models to predict bud-burst timing was analysed. All temperature-based models resulted in significantly better predictions than the null model, day number. Temperature sum (TS) based on a linear response to temperature above a defined threshold, gave a more precise prediction than forcing units based on a logistic response to temperature. The most precise model could predict bud burst to within 2 days, and it included TS with a threshold of 5 degrees C and a start day arbitrarily chosen between January and March. The models were not improved by incorporating chilling, in terms of chilling days (number of days with mean temperature <5 degrees C) or chilling units. It is suggested that chilling requirement is fulfilled already in December under normal winter conditions in southern and central Sweden. Ranking of clones in bud-burst timing was stable over years and sites. For routine measurements of phenology in applied breeding programmes, it is proposed that standard reference material is thoroughly tested for temperature reactions and that this reference material be used each time new material is tested. With this procedure, an accurate estimate of required temperature sum for bud burst can be obtained from a 1-year assessment.