Combining fruit tree varieties for better spring bloom timing

Jenn Hoskins
24th January, 2026

Combining fruit tree varieties for better spring bloom timing

This study evaluated different spring phenology modeling approaches using bloom records for almond (Prunus dulcis), apricot (Prunus armeniaca), and sweet cherry (Prunus avium) collected from diverse climatic regions in the Mediterranean and Germany.

Image adapted from: Caspersen et al. / CC BY (Source)

Key Findings

  • This study focused on almond, apricot, and sweet cherry trees across Mediterranean regions and Germany to improve spring flowering predictions
  • A new “combined-fitting” method, pooling data from multiple cultivars within a species, didn’t improve prediction accuracy but required less data per cultivar
  • While the combined-fitting approach didn’t outperform traditional methods, it offers a practical solution for modeling flowering when data is limited and allows for easier comparison of cultivars
Predicting when fruit trees will flower is crucial for agriculture, influencing harvest times and yields. However, building accurate models for this prediction is challenging because data on flowering times, known as phenological data, is often limited – it can be fragmented, collected over short periods, and focused on specific locations.[1] Researchers at the University of Bonn and the University of California addressed this issue by developing a new method for calibrating spring phenology models, particularly for almond, apricot, and sweet cherry trees. The core problem lies in the need for detailed data for each individual cultivar (variety) of fruit tree. Each cultivar responds slightly differently to environmental cues, meaning a model trained on one cultivar might not accurately predict flowering for another. Traditionally, models are “fitted” to data from single cultivars, requiring substantial data for each one. This new study proposes a “combined-fitting” approach, pooling data from multiple cultivars of the same species to create a single, more robust model. The researchers utilized the PhenoFlex framework, a tool for building these phenology models. They compared the combined-fitting method to two other approaches: a baseline model and the traditional “cultivar-fit” method, where each cultivar is modeled independently. The analysis focused on flowering data collected from nine almond, six apricot, and six sweet cherry cultivars grown in diverse climates across Mediterranean regions (Spain, Morocco, Tunisia) and Germany. The study’s findings showed that the combined-fitting approach did not demonstrably outperform the traditional cultivar-fit method in terms of prediction accuracy, as measured by root mean square errors. However, it did offer a significant advantage: it required less data per cultivar. This is particularly important when dealing with cultivars where extensive historical flowering data is unavailable. Interestingly, the cultivar-fit models revealed substantial variation in chill and heat accumulation parameters among cultivars of the same species. These parameters represent how the trees respond to cold temperatures (chilling) and warm temperatures (forcing) that trigger bud development. The combined-fitting method, by its design, yielded a single set of parameters for each species, effectively averaging out these cultivar-specific differences. This relates to earlier research highlighting the complex relationship between temperature and plant phenology[2]. It's known that both chilling and forcing temperatures are vital for bud development – low temperatures break dormancy, while warmer temperatures promote growth. The study by[2] also points to the potential for global warming to disrupt this balance, particularly at the warmer edges of a species’ range. The parameters estimated in these models are therefore crucial for understanding and predicting these shifts. Furthermore, the Dynamic model[3] has previously been identified as a highly accurate tool for quantifying chill accumulation. While the current study did not directly focus on the Dynamic model, the importance of accurately characterizing chilling requirements is consistent with its principles. The work by[3] emphasizes that these parameters should be specific to each species and even cultivar, which aligns with the observed variation in the cultivar-fit models used in. The researchers acknowledge that the combined-fitting approach didn’t provide a superior predictive advantage in this specific case. However, they emphasize its practical value. By integrating data from multiple cultivars, it provides a viable solution for modeling spring phenology when datasets are limited. Crucially, it also facilitates comparisons of cultivars within the same species, which could be valuable for breeding programs and agricultural management.[4] highlighted the difficulties in comparing phenological models due to their empirical nature and lack of standardization, and this combined-fitting approach represents a step towards addressing that issue by providing a more unified framework. The study underscores the importance of continued data collection, particularly regarding endodormancy break dates[2], to refine these models further and improve the accuracy of phenological predictions.

FruitsAgriculturePlant Science

References

Main Study

1) Combining temperate fruit tree cultivars to fit spring phenology models

Published 21st January, 2026

https://doi.org/10.1007/s00484-025-03068-2

Related Studies

2) Can phenological models predict tree phenology accurately in the future? The unrevealed hurdle of endodormancy break.

https://doi.org/10.1111/gcb.13383

3) Reducing the uncertainty on chilling requirements for endodormancy breaking of temperate fruits by data-based parameter estimation of the dynamic model: a test case in apricot.

https://doi.org/10.1093/treephys/tpaa054

4) A unified model for budburst of trees.

Journal: Journal of theoretical biology, Issue: Vol 207, Issue 3, Dec 2000


Related Articles

An unhandled error has occurred. Reload 🗙