Résumé
Black leaf streak disease (BLSD), caused by the fungal pathogen Mycosphaerella fijiensis, is considered asthe most destructive foliar disease of banana. To advance our knowledge of the dynamics of the diseaseat plant scale as well as of the components of varietal resistance, we designed, calibrated and evaluateda mechanistic model to simulate the disease on a banana plant. The model runs in discrete time at plantscale and describes plant growth and pathogen dynamics under optimal epidemiological conditions.The model is divided into two modules: a deterministic plant sub-model that simulates the simplifiedarchitecture and growth of the banana, and a pathogen sub-model that simulates the detailed life cycleof the pathogen including infection, lesion growth, asexual and sexual sporulation, and the dispersalof spores on the plant. The three most influential epidemiological parameters identified by sensitivityanalysis of the model (lesion growth rate, infection efficiency, and incubation period) were estimated ina Bayesian framework using Markov chain Monte Carlo methods and acquired data on the dynamics ofleaf lesions under natural conditions. The posterior densities provided precise knowledge on pathogenlife history traits. The evaluation of the model using an independent data set confirmed the good qualityof the predictions. Simulations allowed us to evaluate the impact of host resistance components, auto-infection at plant scale, and of the leaf emergence rate, which is linked with cropping practices, on theseverity of BLSD. This foliar disease simulation model will help design new methods of controlling BLSD.