AFLA-hazelnut: a model for the prediction of Aspergillus flavus infection and aflatoxin contamination in Hazelnuts

Hazelnut (Corylus avellana L.) is a high-value crop for the global confectionery industry, yet its production chain is increasingly challenged by aflatoxin contamination caused by Aspergillus flavus. In Azerbaijan, one of the world's leading hazelnut producers, recurrent exceedances of European aflatoxin limits have resulted in economic and trade constraints. In this context, predictive models represent valuable decision-support tools to anticipate periods of elevated contamination risk and to support timely pre-harvest management strategies. This study presents AFLA-hazelnut, a process-based, weather-driven model developed to assess the risk of A. flavus infection and aflatoxin B₁ (AFB₁) contamination in hazelnut orchards. The approach integrates a phenological sub-model for hazelnut development, based on literature, with a fungal sub-model that simulates infection according to temperature, relative humidity, rainfall, and kernel water activity (aw). A hazelnut-specific aw curve was developed to describe kernel water dynamics during the susceptible reproductive stage and to support hourly simulations. The model produces two outputs: an infection risk index (IH) and an aflatoxin risk index (AFB1-I). Model outputs were evaluated using independent pre-harvest field data. IH showed a coherent association with observed fungal occurrence, indicating that the model captures biologically meaningful gradients of infection pressure. In contrast, the limited number of aflatoxin-positive samples did not allow quantitative validation of the AFB1-I index, which remains supported by indirect evidence and model structure. AFLA-hazelnut represents the first process-based framework developed to describe A. flavus infection dynamics and support aflatoxin risk assessment in hazelnuts.