EVALUATION OF THE COMPUTER-MODEL MARYBLYT FOR PREDICTING FIRE BLIGHT BLOSSOM INFECTION ON APPLE IN MICHIGAN

MARYBLYT was evaluated for predicting fire blight blossom infections using weather data and observations made over four growing seasons in several apple (Malus domestica) orchards in Michigan. The model was more accurate in predicting the appearance of symptoms than in predicting infection periods. It accurately predicted that the weather was unfavorable for infection of flowers in 1983 and 1990 and highly favorable for infection of flowers in 1991. Four infection periods were predicted in 1984, but no or very little fire blight blossom infection was observed. The model predicted that fire blight in 1990 was associated with injury to blossoms and leaves from high winds (trauma blight) rather than to normal blossom infection. Parts of the model are subjective, and some modifications in the software would improve the model's usefulness to users unfamiliar with fire blight and its prediction. Despite limitations, the model is a valuable tool for focusing the attention of growers, extension personnel, and researchers on factors known to influence the disease and how those factors can interact to result in infection and the development of fire blight symptoms.