Forecasting Sclerotinia stem rot in spring sown oilseed rape

A new forecasting model for predicting the need for fungicide applications for Sclerotinia stem rot on oilseed rape is presented. Logistic regression was used for calibration and receiver operating characteristic (ROC) curves were used to compare different models. Data from about 800 fields collected during a 10-year period were used in the analysis. The new forecasting model is a risk point table based on field specific data and precipitation. Six factors that affect Sclerotinia infection (number of years since last oilseed rape crop, disease incidence in last Sclerotinia host crop, crop density, rain in the last 2 weeks before flowering, weather forecast and regional risk for apothecium development) are given points with regard to the risk of heavy infestation. The new model was tested in 48 fields during 1996 and 30 fields during 1997 and predicted the need for fungicide applications with high accuracy. At a given risk point threshold, spraying recommendations were given in 75% of the fields that needed fungicide application and in 16% of the fields that did not need to be sprayed. If a lower threshold was used, spraying was recommended in more fields in need of spraying but also in more fields that did not need to be sprayed. (C) 1998 Elsevier Science Ltd. All rights reserved.