Dispersal and plant arrangement condition the timing and magnitude of coffee rust infection

Abstract

One central issue in coffee-leaf rust (Hemileia vastatrix) epidemiology is to understand what determines the intensity and the timing of yearly infections in coffee plantations. However, most experimental and theoretical studies report infection as an average at the plot level, obscuring the role of potentially key factors like rust dispersal or the planting pattern. Here, we first review the rust epidemic patterns of different sites, which reveal large variability in the duration and magnitude of the different epidemiologic phases. We then present a spatially explicit and parametrised model, where the host population is subdivided into discrete patches linked through spore dispersal, modeled as simple diffusion. With this model, we study the role of the planting arrangement, the dispersal intensity and plant-level variables on the maximum average tree infection (MATI) and its timing. Our results suggest that the epidemic timeline can be divided into two phases: a time lag and a growth phase per se. The model shows that the combination of the dispersal magnitude and plant aggregation modifies the MATI and the time to MATI, mainly by preventing some plants from reaching their maximum peak during the epidemic. It also affects the epidemic curves, which can have a stepped, or a rather smooth pattern in plots with otherwise similar conditions. The initial rust infection modulates the time lag before the epidemic and the infected leaf-fall rate drastically changes the MATI. These findings highlight the importance of explicitly considering the spatial aspects of coffee agroecosystems when measuring and managing rust infection, and help us to further understand the spatio-temporal dynamics of ecological systems in general.