FLOWERING DATE PREDICTION IN ‘ROCHA’ PEAR UNDER MILD WINTER CONDITIONS

The main region for ‘Rocha’ pear cultivation in Portugal is the central coastal region of the country, where insufficient annual winter chilling frequently occurs. The use of existing models to predict the date of flowering (e.g., the Utah model and the “chill hours” model) is often unsuitable under mediterranean conditions. A model (Model 1) that is a generalisation of the Utah type of model was calibrated on half of a ten-year data set and validated using the remaining data. Its performance was compared to the Utah model (Model 2) and the “chill hours” model (Model 3), and suitable statistics are presented. Model 1 was a better predictor of flowering date. For example, the root mean square error of the predictions was 0.52, 4.91 and 7.74 days for models 1, 2 and 3, respectively. Model 1 may be useful to predict the time of flowering in other locations and environments, and to yield realistic predictions under global warming scenarios.