Predictors of virus prevalence and diversity across a wild bumblebee community

AbstractViruses are key regulators of natural populations. Despite this, we have limited knowledge of the diversity and ecology of viruses that lack obvious fitness effects on their host. This is even the case in wild host populations that provide ecosystem services, where small fitness effects may have major ecological and financial impacts in aggregate. One such group of hosts are the bumblebees, which have a major role in the pollination of food crops and have suffered population declines and range contractions in recent decades. In this study, we used a multivariate generalised linear mixed model to investigate the ecological factors that determine the prevalence of four recently discovered bumblebee viruses (Mayfield virus 1, Mayfield virus 2, River Liunaeg virus and Loch Morlich virus), and two previously known viruses that infect both wild bumblebees and managed honeybees (Acute bee paralysis virus and Slow bee paralysis virus). We show that the recently discovered bumblebee viruses were more genetically diverse than the viruses shared with honeybees, potentially due to spillover dynamics of shared viruses. We found evidence for ecological drivers of prevalence in our samples, with relatively weak evidence for a positive effect of precipitation on the prevalence of River Luinaeg virus. Coinfection is potentially important in shaping prevalence: we found a strong positive association between River Liunaeg virus and Loch Morlich virus presence after controlling for host species, location and other relevant ecological variables. This study represents a first step in the description of predictors of bumblebee infection in the wild not driven by spillover from honeybees.