Climatic models to predict occurrence of Fusarium toxins in wheat and maize

Abstract

Although forecasting Fusarium infections have useful implications, it may be argued that forecasting Fusarium toxins is more useful to help reduce their entry into the food chain. Several disease incidence models have been commercialized for wheat, but only one toxin prediction model from Ontario, Canada, “DONcast”, has been validated extensively and commercialized to date for wheat, and another has been proposed for maize. In the development of these predictive tools, the variation in toxin levels associated with year and agronomic effects was estimated from simple linear models using wheat and maize samples taken from farm fields. In wheat, environment effects accounted for 48% of the variation in deoxynivalenol (DON) across all fields, followed by variety (27%), and previous crop (14 to 28%). In maize, hybrid accounted for 25% of the variation of either DON or fumonisin, followed by environment (12%), and when combined 42% of the variability was accounted for. The robust site-specific, DON forecast model accounted for up to 80% of the variation in DON, and has been used commercially for 5 years in Canada. Forecasting DON and fumonisins in maize is more difficult, because of its greater exposure to infection, the role of wounding in infection, the more important role of hybrid susceptibility, and the vast array of uncharacterized hybrids available in the marketplace. Nevertheless, using data collected from controlled experiments conducted in Argentina and the Philippines, a model was developed to predict fumonisin concentration using insect damage and weather variables, accounting for 82% of the variability of fumonisins. Using mycotoxins as a measure of disease outcome, as opposed to disease symptoms, offers a more robust prediction of mycotoxin risk, and it accounts for mycotoxin accumulation that occurs frequently in the absence of any change in Fusarium symptoms.