Virulence evolution in a host-parasite system in the absence of viral evolution

Abstract

Question: How does virulence evolve in the Drosophila melanogaster/sigma virus (DMelSV) system?

Organisms: Drosophila melanogaster (host) and DMelSV (parasite).

Empirical methods: Artificial selection on whole-carcass viral titre of infected flies, including two selection regimes (maternal and biparental transmission) and three treatments within each regime (increased titre, decreased titre, and control). The maternal transmission selection regime lasted for six generations, while the biparental transmission selection regime lasted for twelve generations. We further quantified virulence by estimating the fecundity, viability, and development time of infected flies. Finally, we sequenced virus strains at the end of selection.

Predictions and conclusions: Titre is defined here as the number of viral genomes inside a single fly, while virulence is defined as harm to host. We predicted that titre would respond to both increased and decreased selection, that virulence would evolve as a positively correlated response, and that sequence evolution in the viruses would be responsible for these changes. Titre did respond to selection in the biparental regime, although both high and control lines both demonstrated increased titre, while the titre of the low lines did not change. One component of virulence, development time, was positively correlated with titre in the biparental transmission lines (maternal transmission lines were not scored for virulence). However, we detected few (and in some cases, no) genomic changes in the virus, making viral evolution unlikely to be responsible for the response to selection and the association between development time and titre.

Keywords: DMelSV; Drosophila melanogaster; host–parasite co-evolution; rhabdovirus; sigma virus; virulence evolution.

Fig. 1

Selection regimes and response of virus titre to selection: (a) biparental selection regime; (b) maternal selection regime. Virus titre is plotted on a square-root scale. Light lines show response of individual line-by-treatment combinations; dotted lines connect line-by-treatment combinations before and after generation 8 (i.e. missing data; see text). Heavy lines show fitted slopes of the linear mixed model (ribbons are 95% confidence intervals).

Fig. 2

Median development time, in hours, as a function of titre (×10⁶) for each line-by-treatment combination. Points represent values at generation 12 from individual line × treatment combinations (● = control; ▲ = low; ■ = high); grey lines link points from the same genetic background (line). Some line-by-treatment combinations were lost, so some point types are missing for some lines. Lines show predicted titre–eclosion time relationships for each treatment; these are not significantly different from each other. The heavy line and grey ribbon show the overall titre–eclosion time relationship across all treatments, with 95% confidence intervals.