Predicting fusarium head blight epidemics with weather-driven pre- and post-anthesis logistic regression models

Area of Science:

• Plant Pathology
• Agricultural Meteorology
• Crop Science

Background:

• Fusarium head blight (FHB) is a significant disease impacting wheat production globally.
• Accurate prediction of FHB epidemics is essential for effective disease management and yield protection.
• Existing prediction systems may not fully capture the influence of weather variables on FHB development.

Purpose of the Study:

• To identify weather-based predictors for forecasting major Fusarium head blight (FHB) epidemics in the United States.
• To evaluate the efficacy of pre- and post-anthesis weather conditions in predicting FHB severity.
• To improve the accuracy of national FHB prediction systems.

Main Methods:

• Utilized logistic regression and bootstrap methods with a dataset of 527 unique observations.
• Analyzed 380 predictor variables including temperature, relative humidity, and rainfall in various time windows around anthesis.
• Incorporated variables for cultivar resistance, wheat type, and corn residue presence into predictive models.

Main Results:

• Models incorporating weather variables, especially relative humidity and temperature, significantly improved FHB prediction.
• Relative humidity emerged as a more effective predictor of FHB than other weather variables.
• The developed models achieved a 19% lower average test misclassification rate compared to current national systems.

Conclusions:

• Specific weather-based variables, particularly relative humidity during pre- and post-anthesis periods, are critical for predicting major FHB epidemics.
• Integrating these weather predictors enhances the accuracy of FHB forecasting systems.
• The findings provide a basis for more precise and timely FHB risk assessment in wheat-growing regions.