Mitigation of aflatoxin contamination in maize with the support of predictive model AFLA-MAIZE.

Aflatoxin B1 is the most toxic natural compound known confirmed as carcinogenic for humans and animals by the International Agency for Research on Cancer. It is produced by fungi belonging to Aspergillus section Flavi, A. flavus in particular, and it is a main concern in several crops like maize, peanuts, pistachio nuts and cotton in areas with warm and dry weather during the crop growing season. Recently, probably due to extreme weather conditions, possibly to be ascribed to climate change, aflatoxin contamination in maize has been signalled as a problem also in southern Europe, principally in Italy. In tropical areas aflatoxin detection in maize is expected all years, with high levels up to thousands µg/kg of kernels. Italian surveys show a different scenario, with maximum tens to hundreds µg of aflatoxin per kg of kernels, depending on the year, detected in some places and not in the whole growing area. Actions aimed to minimise fungal growth and toxin production are crucial for products destined to food or feed, for consumers heath and to avoid carry over in dairy animals and consequently in milk, a potential further source. Costs related to aflatoxin management are considerable; proper preventive actions commonly imply costs increase and the analytical efforts to guarantee the respect of legal limits, imposed in almost all countries throughout the world, are significant. In this context, the development of a predictive model is crucial. A mechanistic, weather-driven model, based on the infection cycle of Aspergillus flavus on maize, was developed and validated. Briefly, the work was done in three steps: i) development of the model prototype; ii) development of a probability index, to estimate the probability to overcome the legal limit of 5 μg of aflatoxin B1 per kg of unprocessed maize destined to human consumption; iii) validation model predictions. Quantitative data used for model development were available in literature or produced with proper experiments. Data input included hourly/daily meteorological data. Field data were collected in Italy from 2005 to 2011 and used for the probability index development and for model validation. Model output was compared with field data and the validation confirmed around 70% of maize contamination at harvest correctly predicted. AFLA-maize model predicts the risk of AFB1 contamination in field on a daily base from silk emergence to harvest; therefore, the dynamic of fungal growth and contamination risk can be checked throughout the season and it can be considered, also in relation to the crop growth stage, to optimise crop management, harvest and post-harvest logistic. Moreover, AFLA-maize can be applied to predict different situations, based on real or simulated meteorological data, applicable to draw risk maps in poorly studied areas or in climate change scenarios