Effets du microclimat sur le développement de l’épidémie de rouille orangée du caféier Arabica (Hemileia vastatrix – Coffea arabica) dans une gamme de situations de production
DOCTEUR DE L’UNIVERSITÉ DE MONTPELLIER En Biologie des Interactions
École doctorale GAIA
Since 2012, coffee leaf rust, a disease caused by the fungus Hemileia vastatrix, has been responsible for major epidemics of arabica coffee cultivation in all Central American countries. That year, about 20% of the production in this region was lost and the death of branches or even coffee plants caused significant secondary production losses in the following years. As coffee production provides many employments in Central America, these epidemics have led to a major social crisis. To prevent future epidemics, the PROCAGICA programme (Programa Centroamericano para la Gestión de la Roya del Café), initiated in 2016, with funding from the European Union, aims to implement measures such as the creation of a regional warning network, based on improved national systems, including a prognostic component based on weather variables. In most prediction models already available for this disease, meteorological variables are considered over long periods of time to explain indicators such as incidence reflecting disease progression but also host growth.
In this thesis we hypothesized that it is possible to forecast epidemic growth by modeling different stages of the fungal development, each determined by complex combinations of microclimatic variables acting at different periods (times and durations). To better understand the functioning of the pathosystem and thus define the variables to be predicted, we first described the causal relationships existing between the phenology of arabica coffee, the development of coffee leaf rust and their environment in agroforestry systems. Through the use of structural equation modeling, the controversial overall effect of shading on the disease was explained as the result of antagonistic effects on colonization and sporulation stages, depending on the type of shading. This analysis also highlighted the strong interaction between rust and coffee growth. We therefore decided to avoid working on indicators such as incidence but rather to construct three models of symptoms and signs onset of the disease: emergence of lesions without uredospores, onset of sporulation and growth of the lesion's infectious area. As a result of a trial conducted from May 2017 to July 2018 in Costa Rica, we collected data enabling the identification, without a priori, of combinations of moments and durations of action of microclimatic variables on the onset of symptoms and signs. To promote the use of the three models obtained, in Central America where agroforestry is a common practice, we developed simple microclimate estimation models using data from weather stations located in full sun exposure and characteristics, such as shade tree height and canopy openness, that can be easily assessed in agroforestry systems. The three predictive models developed to predict risk of symptoms and signs onset are simple equations that can be used separately to predict risks that involve different disease control recommendations. These models and microclimate estimation models can also be coupled within a simulator.