Model-based characterization of interactions between plum pox virus strains from a field survey

Abstract

Multiple pathogen strains often co-circulate within the same host population and may lead to multiply infected individuals. However, the characterization of the interactions between strains and of their epidemiological impact is still a burgeoning research field. Here we show how modeling multi-strain epidemiological dynamics provides a more reliable assessment of strain interactions in the field than statistical tests of independence. The constructed model, which accounts for orchard age and all possible strain combinations, was applied to test for interactions between three virus strains under the assumption of endemicity. The Serbian survey data used for the inference showed that sharka, caused by the plum pox virus (PPV), is widespread in the 91 sampled plum orchards and that three strains (M, D, and Rec) are present even as double and triple infections in the field. The observed co-infection frequencies were compared with their predicted frequency under the neutral/null hypothesis that there is no direct or indirect interaction between strains. We showed that D+Rec coinfections were less frequent than expected in the neutral case but also, in contrast with model predictions, that D+Rec coinfection frequency decreased with the orchard age. Our results raise biological and epidemiological questions on the Rec strain of PPV and emphasizes the importance of coupling models and data to understand and predict dynamics in complex epidemiological systems.