In Uruguay, a severe Fusarium epidemic in wheat in 2001 resulted in concentrations of deoxynivalenol (DON) in baked goods from 1-5 mg/kg. DONcast is a prediction model used in Canada mainly to aid fungicide spraying, but was adapted to Uruguay to help target pre-harvest regulatory and marketing actions for reducing DON contamination in wheat destined for food markets. Daily temperature (T), and relative humidity (RH) were collected from four automated weather stations, and daily precipitation data were collected from 50 manual stations in the wheat growing regions in Uruguay. Temperature and RH data were interpolated for each of the manual sites. The level of DON was predicted for each day across a range of heading dates, for each weather station. Before harvest, regions were identified across Uruguay with low (< 1.0 mg/kg), moderate (1.0-5.0 mg/kg), and high (> 5 mg/kg) frequencies of predictions, depending on the range of heading dates within each region. DONcast was modified for unique environmental conditions in Uruguay, by accounting for RH between 3 and 10 days after heading (increases DON), precipitation and RH between 20 and 36 days after heading (increases DON), and T-max. > 32 degrees C between 10 and IS days after heading (decreases DON). DON was predicted using one of two equations, depending on whether rain was recorded during heading (R-2 = 0.80) or where no rain was recorded during heading (R-2 = 0.65). Validation analysis showed that the revised model forecasted a mean of 6.1 mg/kg of DON, compared to 7.5 mg/ kg measured in 20 fields, and that trends of elevated DON concentrations were identified, depending on the heading date. The results from this study suggest that DONcast is a practical tool to forecast the severity of regional DON contamination before harvesting wheat in Uruguay and other similar regions.
