Improved modelling of fusarium to aid mycotoxin prediction in UK wheat


Cereals & Oilseeds
Project code:
01 April 2009 - 15 July 2014
AHDB Cereals & Oilseeds.
AHDB sector cost:
Total project value:
Project leader:
Edwards SG1, Jennings P2 and Pietravalle S2 1Harper Adams University, Newport, Shropshire, TF10 8NB, UK 2Fera Science Ltd, Sand Hutton, York, YO41 1LZ


pr566-final-project-summary pr566-final-project-report

About this project


European legal limits for fusarium mycotoxins deoxynivalenol (DON) and zearalenone (ZON) for unprocessed cereals intended for human consumption and cereal products were introduced in 2006. The aim of this study was to monitor the fusarium mycotoxins in UK wheat from 2009 to 2013. Collated data was then used to develop models to predict fusarium pathogen incidence and mycotoxin risk at different timings (early and late season) and at different levels (national, regional and field). Results showed large seasonal differences in fusarium pathogen incidence and mycotoxin risk with high levels in 2012 (10% and 15% exceeding legal limit for DON and ZON, respectively) and low levels in 2011 (no samples exceeding legal limits for DON and ZON). Prediction of mycotoxin content early season was not successful with poor prediction in the validation years of 2012 and 2013. This is likely due to a number of factors. This includes the large number of zero values in the model, the poor ability to predict the key growth stages for fusarium infection in winter wheat and the paucity of local meteorological data for some fields. This study has highlighted the need for greater recording and/or prediction of growth stages to improve mycotoxin prediction alongside greater availability of in-field weather data. Due to the multi-step infection process of fusarium head blight (spore production, spore release, infection and mycotoxin production) then mixed or mechanistic models maybe more appropriate. Analysis of agronomic factors identified the same factors as previously determined, namely previous crop, cultivation and variety. A new factor determined was harvest date with a one month delay in harvest resulting in a large increase in risk for both DON and ZON, as was experienced with the delayed harvest of 2008. Prediction of mycotoxin late season, based on a combination of fusarium head blight pathogen incidence (as recorded at growth stage 73) and agronomy was a better predictor of risk and proved reasonable at the national and field scale. At field scale, the DON model was a better predictor of risk than the ZON model and could be used to predict ZON as well as DON risk. If a consignment of wheat was deemed negative by the prediction model, it would not require a mycotoxin test and would enter the food chain. If it was subsequently found to be above the legal limit, then the consignment was a false negative and considerable costs may be incurred as well as a potential risk to human health. As false negatives have a greater consequence for the industry (consignments of wheat exceeding legal limits entering the food chain), then the probability of exceeding the legal limit can be calculated and this can be used to determine a lower risk threshold for predicting false negatives. If a threshold of 5% is set, then a sample is deemed to be below the legal limit if the probability that the actual concentration will exceed 1250 ppb DON is below 5%. With a 5% probability that a sample exceeded the DON limit, no samples above the legal limit were predicted not to exceed the DON limit (zero false negatives) and only 0.4% of ZON samples exceeded the ZON limit of 100 ppb but were predicted not to (0.4% false negatives). This study shows the benefit of the continued determination of fusarium head blight incidence data as generated by Fera from the Defra funded winter wheat disease survey. Late season national risk 5 prediction could be used by the cereal chain to identify the availability of home-grown food grade wheat in the UK and the extent of due diligence testing required at harvest each season. Late season field scale risk prediction could be used by the cereal chain to determine which consignments of wheat require testing for fusarium mycotoxins prior to delivery into the human food chain.