Speaker
Description
Meryem Ghanem¹˒⁴, Saloua Bensiali², Tarik Benabdelouahab³, Hafsa Houmairi⁴, SanaeKrimi Bencheqroun¹
¹ Plant Protection Laboratory, Regional Agricultural Research Center of Settat, National Institute of Agricultural Research (INRA), Avenue Ennasr, P.O. Box 415, Rabat Principale 10090, Rabat, Morocco.
² Department of Applied Statistics and Computer Science, Hassan II Institute of Agronomy and Veterinary Medicine (IAV Hassan II), Rabat 10101, Morocco.
³ National Institute of Agricultural Research (INRA), Avenue Ennasr, P.O. Box 415, Rabat Principale 10090, Rabat, Morocco.
⁴ Agri-Food and Health Laboratory, Faculty of Sciences and Technology, Hassan First University, Settat, Morocco.
Email :meryem.ghanem@gmail.com
Conference Theme:
Option 1 – Theme 3:Agriculture, Food Security and Climate (Climate-Resilient and Climate-Smart Agriculture)
or
Option 2 – Theme 1:Climate Science, Big Data and Artificial Intelligence (Climate Risks and Early Warning Systems)
Keywords:Phytophthora infestans; epidemiological models; climate change; machine learning; decision support system.
Type of Presentation: Oral presentation (on-site).
Abstract
Potato late blight, caused by Phytophthora infestans, remains one of the most destructive diseases affecting potato production worldwide. Without timely and effective fungicide protection, the disease can cause severe yield losses, oftenleading to complete crop failure. To optimize disease management and reduce reliance on chemicaltreatments, several epidemiological forecasting models have been developed over recent decades, with Smith, Wallin, Blitecast, Winstel, Irish Rules, NegFry, Forsund, and Litschmannbeing the most widely used. However, because these reference models were originally developed forhumid, temperate climates, their applicability under semi-arid conditions, where climate change is driving heightened weather variability, has not yet been evaluated, particularly in regions like Morocco.
This study investigates the transferability and predictive performance of these eight reference models under Moroccan agroclimatic conditions to identify forecasting approaches suitable for local potato production systems. A field experiment was conducted during the 2025–2026 growing season in Berrechid (Casablanca–Settat, Morocco) using four potato cultivars exhibiting contrasting levels of susceptibility to late blight. Microclimate data (air temperature, hourly relative humidity, and rainfall) were monitored continuously on-site, anddisease progressionwas assessed according to the EPPO protocol. The selected forecasting models are evaluated against this field data using appropriate statistical performance indicators.
This work represents the foundational step of a broader research program dedicated to developing a hybrid decision support system (DSS) that integrates epidemiological forecasting models with machine learning algorithms. The proposed framework is expected to provide more accurate and robust disease predictions tailored to Moroccan environmental conditions, thereby promoting sustainable crop protection and enhancing agricultural resilience under a changing climate.