Epidemiology and disease management of Stewart's disease of corn in Iowa

Disease forecasting has been an important disease management component for Stewart's disease (Pantoea stewartii subsp. stewartii (syn. Erwinia)) of corn (Zea mays). To date, forecasting for Stewart's disease has been based exclusively on the examination of winter air temperatures for Dec., Jan., and Feb. Three forecasting models: Stevens (predictor of the early wilt phase of Stewart's disease), Stevens-Boewe (predictor of the late leaf blight phase of Stewart's disease), and Iowa State Model (predictor of the prevalence of Stewart's disease), were examined for their ability to accurately predict the prevalence of Stewart's disease in Iowa. Prevalence data were obtained from field inspections records for seed corn fields conducted under the auspices of the Iowa Department of Agriculture and Land Stewardship for the years between 1972 and 2003. Parameter coefficients for the three models were obtained using weighted binary logistic regression (weights were the number of fields inspected each season in a county) of seed com inspection and air temperature data between 1972 and 1999 {n = 786 county-years). All three models indicated increased odds for Stewart's disease with warmer winter air temperatures. An internal bootstrap validation of each model indicated that the Stevens model had a maximum error between the predicted and calibrated probabilities of 10%, while the other two models had maximum errors of 1 % or less. External validation using 154 independent county-years of seed corn inspection and