Modelling mycotoxins contamination in maize to optimize agricultural practices and improve the value chain.

Maize is a worldwide distributed crop, crucial for food or and feed in all the growing areas. Many efforts have been devoted to improve the genetic material and the cropping system to increase maize value in all the geographic areas, very different for environmental, social, cultural and economic aspects. A still unsolved problem regards kernels contamination by mycotoxin producing fungi . Health risks related to mycotoxins, due to chronic and acute toxicity impose the application of good agricultural and management practices and the respect of legal limits fixed in most countries; consequently, the analytical check of maize production is mandatory. This careful crop management involves additional costs, and therefore it reduces the value chain, in some areas or it is not applicable in others. A sustainable approach would be based on the prediction of mycotoxin risk to optimise maize chain management and analytical efforts. A modelling perspective was followed as support to improve the value chain. Fusarium section Liseola and Aspergillus section Flavi are the main fungi growing on maize and aflatoxins and fumonisins the main health risk concern. Two mechanistic models were developed to predict the infection cycle of A. flavus and F. verticillioides on maize. Host crop phenology was included in modelling to find out the crucial growth stages for fungal activity intended as silk emergence and ripening. The mandatory input to run models is are meteorological data (temperature, relative humidity and rain) and the output consists of indexes, AFI and FUMI respectively for aflatoxin and fumonisin; they were related to a probability to overcome a fixed threshold. The index is available on a daily base and it allows to follow the dynamic of risk along the growing season, from silk emergence to harvest. The threshold is fixed at 5 µg/kg for aflatoxin B1 and at 4000 µg/kg for fumonisin B1 + fumonisin B2; the reference is the legal limit in force in Europe for unprocessed maize destined to human consumption. The models were validated with data collected in 6 different countries around the world. Field samples and related meteorological data were collected; observed data were compared to predicted data to check the reliability of predictions. The results were encouraging, with a good percentage of correct predictions, very similar in all the countries considered for the validation. Therefore, these models can be used in supporting decisions to add safety and value to the maize chain production