Digital modeling of soil-borne fusarium’s nutritional: Investigating microbiological and microecological dynamics in Moroccan agroecosystems

¹ Laboratory of Agro-Industrial and Medical Biotechnology, Faculty of Sciences and Technics, Sultan Moulay Slimane University, B.P. 523, Beni Mellal, Morocco
² LISAD Laboratory, Industrial Engineering Dept.ENSA, Ibn Zohr University, Agadir, Morroco
³ Intelligent System and Security of System, Mohamed V university, Rabat, Morocco
⁴ Equipe des Mathematiques et Interactions ,Faculty of Sciences and Technics, Sultan Moulay Slimane University, Beni Mellal, Morocco
⁵ Oasis System Research Unit, Regional Center of Agricultural Research of Errachidia, National Institute of Agricultural Research, PO. Box 415, Rabat 10090, Morocco

^* Corresponding author: hilalialaoui.79@gmail.com

Abstract

The areas planted with date palm trees within the Moroccan oases cover more than 48,000 hectares and play a key role in both the environment and socioeconomic stability. Unfortunately, these ecosystems are threatened by the Bayoud disease caused by Fusarium oxysporum f. sp. albedinis, a vascular wilt pathogen that has already devastated millions of date palms in Morocco and Algeria since the 19th century. Any similar outbreak over time poses a serious threat to the long-term sustainability of these oases. This study aimed to elucidate the biological mechanisms associated with Bayoud decline in suppressive soils. To achieve this, soil samples were collected from the Ziz and Draa Valleys, where date palms are infected by Bayoud, as well as from the palm grove of Marrakech, which is considered a suppressive zone for this disease. In other words, the samples were taken from two disease-conducive zones and one suppressive zone for Bayoud. A total number of eighteen samples were removed from various depths to compare two conducive soils and one suppressive soil. Ninety Fusarium strains were isolated using this approach and tested for their antagonistic or competitive properties against the Bayoud pathogen. The bacterial and fungal communities were characterized using ITS1 and 16S amplicon sequencing, respectively, with growth tests conducted on Biolog SF-P2 plates. Using soil samples from the three research regions, we investigated three machine learning techniques to determine the feeding patterns of Fusarium communities: Decision tree models, k-nearest neighbors, and Logistic regression. The performance scores of the models were as follows: the k-nearest neighbors model achieved 80%, the logistic regression model scored 77.78%, and the decision tree classifier obtained a score of 68%. These results highlight the potential of machine learning approaches in understanding the nutritional behavior of Fusarium communities. Our research provides a foundation for modeling efforts aimed at generating forecasts to mitigate the damages caused by Bayoud on Morocco’s vital date palm ecosystems.