Spring Phenology Models for Temperate Apple Cultivars

The annual dormancy cycle of apple trees is highly temperature dependent, with photoperiod deemed irrelevant for dormancy induction or breaking. In fall, cold days induce endodormancy. Endodormancy, in turn, is overcome by further accumulation of chill, when a cultivar-specific chill requirement is met. To further overcome ecodormancy, a cultivar-specific heat requirement must be met, allowing for bud break and subsequent phenology phases to occur. Thus, compared to species where photoperiod is relevant for the dormancy cycle, apple tree dormancy and spring phenology phases are especially susceptible to climate change.

Chill and heat requirements reported in the literature vary between cultivars, within cultivars with location and even for the same cultivar and location depending on the methodological approach. This is, inter alia, related to an imprecision regarding terminology. The minimum chill and heat to respectively overcome endodormancy and ecodormancy are referred to as chill requirement and heat requirement. Yet, studies often report a chill or heat accumulation that – if a phenological phase occurred – at least met the respective requirement. Thus, the existing literature can provide approximations of the actual chill and heat requirements. However, a large database of phenology observations might include instances at the limit for the occurrence of specific phenological phases. Thus, such a database may most accurately approach a cultivar’s actual chill requirement.

In this work we capitalize on a phenology database by the German Weather Service (DWD) that is openly available. This database entails over 50,000 observations for each bud break, bloom start and full bloom spanning the years 1996 to 2024. While there are data on over 60 apple cultivars, we deemed the data for 23 cultivars sufficient for model development. As source for temperature data, we used a temperature grid with 1x1 km spatial resolution developed by the DWD. As underlying modelling approach, we used the chill overlap model. This model is not merely fitting data statistically, but provides a biological meaningful framework.

This biological footing is crucial to extrapolate findings to different climatic conditions. Therefore, the spring phenology models developed in this work will allow to predict onset of spring phenology phases in a warmer future. Consequently, in the future the risk of late frost events or the risk of approaching climatic conditions that will hinder bud break can be investigated for each cultivar. Thus, a cultivar-specific risk assessment regarding likely future conditions can inform planting decisions.