Validation of a FHB-wheat model in MycoRed project, a further useful tool to improve cereals safety.

MYCORED is a large collaborative project focused on developing strategic solutions to reduce contamination by mycotoxins of major concern in economically important food and feed chains, through mycotoxin research joint actions. The work package 3 of the project is focused on the development and validation of predictive models for mycotoxin producing fungi. Deoxynivalenol (DON) and zearalenone (ZEA) in cereals are considered as an health problem worldwide. They are secondary metabolites associated to Fusarium head blight (FHB), a ear disease caused by a complex of Fusaria, with F. graminearum considered as the main fungus involved. Predictive models and decision support systems (DSS) are useful tools to rationalise both the cropping system and the survey of contamination in order a) to minimise consumers exposure; b) to describe the risk of contamination at harvest and rationalise the harvest/post-harvest logistic; c) to draw different scenarios based on real and simulated meteorological data. The aim of this work is to validate a predictive model developed by Rossi et al (2003) and the related DSS (Rossi et al., 2007) to forecast risk level associated with FHB in wheat, in different geographic areas. Four countries were involved (Italy, Russia, Egypt, and Mexico), where 133 wheat samples with related cropping system and meteorological data were collected in 2009 according to defined protocols. DON and ZEA contamination at harvest was determined in all samples. Meteorological data were used as input in the model and the predicted risk of DON and ZEA in wheat at harvest was obtained as output. Predicted and observed data were compared to validate the model. The model validation gave good results in almost all the geographic areas monitored; correct predictions varied between 75 and 100% in almost all data sets, except for those collected in central-southern Italy. As a global result, correct predictions were 69%; the absence of underestimates, which could represent a problem in practice, is also a positive result. The crucial role of meteorological data was confirmed; in fact, a limited improvement was obtained adding further information related to the cropping system. The importance of using reliable meteorological data was stressed; in fact, when meteorological data were collected from stations placed more than 20 km from the crop, the reliability of predictions decreases significantly. The model gave reliable predictions worldwide and it could be very helpful in small grain management to minimise the risk for consumers health related to mycotoxin contamination.