Link to home

Evaluating Predictors of Apple Scab with Receiver Operating Characteristic Curve Analysis

June 2005 , Volume 95 , Number  6
Pages  679 - 691

William W. Turechek and Wayne F. Wilcox

Cornell University, NYSAES, Department of Plant Pathology, Geneva, NY 14456


Go to article:
Accepted for publication 2 February 2005.
ABSTRACT

Apple scab (Venturia inaequalis) is a perennial threat to apple production in temperate climates throughout the world. In the eastern United States, apple scab is managed almost exclusively through the regular application of fungicides. Management of the primary phase of disease is focused on preventing infection by ascospores. Management of secondary cycles of infection is largely dependent on how well primary infections were controlled. In this study, we used receiver operating characteristic curve analysis to evaluate how well mid-season assessments of the incidence of apple scab on cluster leaves, clusters (i.e., the whorl of cluster leaves), or immature fruit can serve as predictors of apple scab on harvested fruit (harvest scab) and whether these mid-season assessments of scab could be used reliably to manage scab under various damage thresholds. Results showed that assessment of scab on immature fruit was superior at predicting harvest scab than were assessments made on clusters or cluster leaves at all damage thresholds evaluated. A management action threshold of 7% scab incidence on immature fruit was identified by Youden's index as the optimal action threshold to prevent harvest scab incidence from exceeding 5%. Action thresholds could be higher or lower than 7% when economic assumptions were factored in to the decision process. The utility of such a predictor is discussed.


Additional keywords: Bayes theorem , disease management , risk management .

© 2005 The American Phytopathological Society