Risk Prediction Models of Frost Damage During Flower Bud Development for Four Deciduous Fruit Species

Abstract

For efficient frost protection of fruit species, we constructed risk prediction models of frost damage for apples, Japanese pears, peaches, and cherries, using data obtained from the freezing treatment of twigs bearing flower buds. These are logistic regression models designed to estimate the occurrence of frost damage during flower bud development from the bud break stage to the post bloom stage. These models were obtained from logistic regression analysis using the freeze treatment temperatures as explanatory variables. On the basis of a likelihood ratio test, the fitness of all models was good, and judging by the identification rate, the prediction accuracy was high, except for the bud break stage model of the ‘Fuji’ apple and the calyx red stage model of the ‘Akatsuki’ peach. Nevertheless, with the exception of some of the development stage models for cherries, it became clear that the logistic curve of each model shifted to the high temperature side from the bud break stage to the post bloom stage. This phenomenon suggests that the importance of frost protection increases as the flower bud develops. Risk prediction based on the estimation of economic losses caused by frost damage is needed for effective frost protection. However, in our experiment, we did not examine damage after the post bloom stage. Therefore, in the future, we will evaluate the utility of models by examining past and future cases of frost damage caused under field conditions.