Elena, S. F.,Lalic, J.

Forecasting plant virus emergence depends on identifying the factors that determine the distribution of genetic variants within the primary host as well as across potential new hosts. It is crucial to determine: (i) the distribution of mutational fitness effects (DMFE) on the primary host, (ii) how it changes on different hosts, (iii) the way in which multiple mutations interact in determining viral fitness in the primary host, and (iv) whether this interaction is host-dependent. To illustrate points (i) and (ii), this paper reviews recent reports showing that the DMFE for a potyvirus markedly differs between natural and non-natural hosts. Changes in genetic variance for fitness are the main cause of the observed pattern among related hosts, whereas sign pleiotropy mainly explains differences observed among unrelated hosts. To illustrate point (iii), comments are made on experiments showing significant epistasis among random pairs of mutations in the potyvirus genome. A large fraction of the interactions correspond to the reciprocal sign epistasis, meaning that the sign of the effects of mutations at two loci are mutually dependent. Finally, to illustrate point (iv), evidence is presented that epistatic interactions for an RNA virus varied among hosts, with magnitude epistasis being stronger in the primary host but becoming weaker as host's taxonomic relatedness decreased. The existence of all these interactions jeopardizes the prediction of the fitness and evolutionary fate of a given mutation, as it will depend on the genetic background but also on the host wherein the virus replicates.