Using Predictions from a Fusarium Head Blight Risk Assessment Tool as Predictors of the risk of Deoxynivalenol Contamination of Wheat Grain
Wanderson Bucker Moraes: The Ohio State University
<div>Fusarium head blight (FHB), caused by <em>Fusarium graminearum</em>, is associated with wheat grain contamination with mycotoxins such as deoxynivalenol (DON). A pre-anthesis weather-based tool was previously developed to help guide fungicide applications for FHB management based on predicted risk of an FHB epidemic. However, producers and grain buyers also use FHB predictions as indicators of the risk of DON contamination when making grain marketing decisions. But the accuracy of the FHB risk tool for DON prediction has not been formally evaluated. Here we used the predicted risk of an FHB epidemic (FHB-RISK) as an independent variable to predict DON contamination above a grain marketing threshold of > 2 ppm. A total of 583 unique cases was used to build a logistic regression model to estimate the probability of DON exceeding 2 ppm when the predicted risk of an FHB epidemic was low, moderate, or high (FHB index > 10%). The area under the ROC curve, a measure of model discrimination between cases with and without DON > 2 ppm, was 0.66, and the overall correct classification rate was 70.7%. The sensitivity (proportion of cases with DON > 2 ppm correctly classified) and specificity (proportion of cases with DON ≤ 2 ppm classified correctly) were 39.3 and 92.4%, respectively. These results suggest that one should not rely on FHB risk predictions alone to assess the risk of DON contamination as such predictions performed poorly classifying cases with DON levels > 2 ppm.</div>
|
