Deoxynivalenol prediction in oats

Abstract: In recent years there have been problems with unacceptable high levels of the mycotoxin contam-inant Deoxynivalenol (DON) in oats in Sweden and Norway. This is due to infections of the fun-gal pathogens Fusamium graminearum and Fusarium culmorum. The question which this thesis attempts to answer is: “Is it possible to predict these elevated levels of DON?” in order to time and conduct suitable countermeasures. A prediction model was created to answer this question. The model calculates a DON contamination risk index based on the likelihood of present germi-nated spores, plants susceptible for infection, infection event and production of DON. The model needs hour based weather data input for relative humidity and global radiation. Also, the model uses a leaf wetness model and a temperature driven phenology model to predict input data for leaf wetness, leaf surface temperature and plant growth stages. The model indices were compared towards DON measurements in oats in Norway and a regression analysis was conducted. The model did in a few cases show a strong correlation towards the measurements, but in most cases there was no correlation or a negative correlation. Therefore, it is considered that the model is not capable to predict DON contamination. Alternative model applications were conducted to predict DON in oats. Among the alternative applications, the prediction model ENV also known as GIBSIM for Fusarium graminearum infections in Brazil was included. However, only two in-stances with the ENV applications of all the alternative model applications showed strong posi-tive correlation. Once again the models used failed to predict DON contamination. There is a risk that the models generated incorrect predictions due to calculation errors since no sensitivity anal-ysis was conducted. The models might be capable to predict DON in oats if the study better com-pensates for environmental variance and if the models take into account factors like recovery and spore density.