


VIEW ARTICLE
Disease Control and Pest Management
A Predictive System for Timing Chemical Applications to Control Pseudomonas syringae pv. tomato, Causal Agent of Bacterial Speck. D. J. Jardine, Graduate assistant, Department of Botany and Plant Pathology, Michigan State University, East Lansing 488241312; C. T. Stephens, Associate professor, Department of Botany and Plant Pathology, Michigan State University, East Lansing 488241312. Phytopathology 77:823827. Accepted for publication 27 October 1986. Copyright 1987 The American Phytopathological Society. DOI: 10.1094/Phyto77823.
Stagewise multiple linear regression techniques were used to identify those meteorological and biological variables useful in predicting bacterial speck symptom development caused by Pseudomonas syringae pv. tomato. An initial regression model relating temperature, rainfall, and previous population level was developed from 2 yr of data. The model was BP = –2.99 –0.14(T) + 1.34(R) + 0.81(P), where BP = the predicted bacterial population, T = the average temperature (C) on the previous day, R = the square root of (the sum of the daily rainfall + 0.5 for each of the previous 7 days), and P = the population at the previous sampling time. The model accounted for 85% of the observed variation in the population for the years used in model development. In 1984, the equation correctly predicted values above or below the preselected threshold for spray application 12 of 12 times and resulted in three fewer sprays than used in a calendar spray schedule. There were no significant differences in amount of fruit infection between spray schedules. The data base for the regression model was expanded by combining 1984 data with that from the previous 2 yr. The model was BP = 0.98 + 0.72(R) –0.11(T) + 0.01(H) + 0.51(P) where BP= the predicted bacterial population, T = the average temperature on the previous day, R = the square root of (the sum of the daily rainfall + 0.5 for the previous 6 days), P = the population level at the previous sampling time, and H = the arcsinsquare root of the average relative humidity for the previous day. This equation accounted for 46% of the variation in the population for the years used in model development. The model derived from 3yr data was tested for its applicability to data from 1980 and 1981. Values above or below the threshold were correctly predicted 83 and 86% of the time, respectively. The potential for use of this model in commercial tomato production is discussed.
Additional keywords: epidemiology, forecasting, Lycopersicon esculentum.
