Asymmetrical over-infection as a process of plant virus emergence.

Disentangling the role of epidemiological factors in plant pathogen emergences is a prerequisite to identify the most likely future invaders. An example of emergence was recently observed in France: in 10 years, "classic" (CL) strains of Watermelon mosaic virus (WMV) were displaced at a regional scale by newly introduced "emerging" (EM) strains. Here we analyse a 3 years dataset describing the co-dynamics of CL and EM strains at field scale using state-space models estimating jointly: (i) probabilities of primary and secondary infection and (ii) probabilities of over-infecting with a CL [EM] strain a plant already infected with an EM [CL] strain. Results especially indicate that it is more than 3 times less probable for a CL strain to over-infect an EM infected plant than for an EM strain to over-infect a CL infected plant. To investigate if these asymmetric interactions can explain the CL/EM shift observed at regional scale, an exploratory model describing WMV epidemiology over several years in a landscape composed of a reservoir and a cultivated compartment is introduced. In most simulations a shift is observed and both strains do coexist in the landscape, reaching an equilibrium that depends on the probabilities of over-infection.