Abstract
Fusarium Head Blight (FHB) is caused by a species complex of Fusarium and Microdochium. This disease, common in wheat, can induce losses of yield but also degrade safety quality of grains. Indeed, the most common species of Fusarium in France is Fusarium graminearum (teleomorph=Gibberella zeae) and it can produce toxins, in particular deoxynivalenol (DON) regulated by the European Commission for cereals intended for human consumption. As this standard has been in effect since July 1st 2006, it is critical for grain producers and processors to have a better knowledge of DON content as well as the factors influencing its level and the agronomic practices that reduce risk. Therefore, ARVALIS - Institut du vegetal has developed decision-making tools to help them manage this risk of grain contamination before its commercialisation. Three kinds of approaches have been performed: a decision-grid to help farmers spray, an epidemiologic model to assess the risk of infection by F. graminearum spores and a statistical model to predict DON content in grain before harvest.
Decision grid and statistical model have been developed by exploiting data from a field survey started in 2001 in collaboration with grain store partners. Over 2700 field samples were harvested and analysed for their DON content. For each field, agronomic parameters and weather conditions during flowering were recorded. A variance analysis has led to the identification of critical factors involved in DON content, namely: the previous crop, the tillage practice, the susceptibility of wheat varieties to DON and the sum of rain around flowering. These factors have been integrated in a DON risk assessment grid which classifies risk categories according to agronomy and weather. This grid is now used by farmers to help them spray or not at flowering. Although this grid is useful for farmers, a model has been set up for grain collectors in order to manage DON risk in silo. The model is based on a linear regression with the backward selection method applied to agronomic and weather parameters. The reliability of these models have been evaluated through a cross validation and show an efficiency modelling (EF) of about 40%. This criterion is low but in more than 90% of cases, the models accurately predicted a DON content below or above the legal limit.
As some years it’s difficult to explain results from statistical model, a mechanistic approach has been performed in collaboration with the Universita Cattolica del Sacro Cuore in Piacenza to predict ascospore risk. In France, the presence of DON in wheat is clearly linked to crop debris remaining on soil at flowering stage. Crop debris are usually colonized by F. graminearum which can produce sexual fruit bodies named perithecia. Those perithecia release ascospores under specific environmental conditions which are assumed to be the major part of inoculum for wheat infection. In the model, the sexual stage of G. zeae was divided in 5 stages: perithecia formation and maturation, ascospore maturation, discharge and germination. For each stage, a specific equation was developed using weather variables (rain, relative humidity, temperature) from literature as independent variables. The final model combines each stage and provides a daily relative risk for ascospores. Specific experiments are on-going to further validate the model and develop a decision-making support system to help farmers to spray or not against FHB at wheat flowering stage and so, reduce the DON content at harvest.
Original language | English |
---|---|
Title of host publication | Proceedings of 7th World Mycotoxin Forum and XIIIth International IUPAC Symposium on Mycotoxins & Phycotoxins |
Pages | 1 |
Number of pages | 1 |
Publication status | Published - 2012 |
Event | 7th World Mycotoxin Forum and XIIIth International IUPAC Symposium on Mycotoxins & Phycotoxins - Rotterdam Duration: 5 Nov 2012 → 9 Nov 2012 |
Conference
Conference | 7th World Mycotoxin Forum and XIIIth International IUPAC Symposium on Mycotoxins & Phycotoxins |
---|---|
City | Rotterdam |
Period | 5/11/12 → 9/11/12 |
Keywords
- modelling
- mycotoxins