In their recent Opinion article (Prisoners of war — host adaptation and its constraints on virus evolution. Nat. Rev. Microbiol. https://doi.org/10.1038/s41579-018-0120-2 (2018))1, Simmonds et al. use studies of hepatitis B virus (HBV)2 and B19 parvovirus3 from Bronze Age remains, along with endogenous viruses4, to propose that evolutionary rates in viruses decline massively as they adapt to hosts. Although these data have changed our perspective on evolutionary timescales, we disagree that they predict long-term evolutionary stasis.

Simmonds et al. claim that there is “increasing evidence for extreme genetic conservation of viruses over longer periods of evolution”. However, most families and genera of RNA viruses exhibit limited sequence similarity5, and as expected with rapid evolution over long time periods, even the amino acid sequences of the most conserved proteins can be difficult to align5. We believe that the regression analysis of rates of viral evolution performed by Simmonds et al. is misleading as it combines single host associations with multi-host comparisons. Importantly, the deeper the timescale of analysis the more virus–host associations are compared, such that the lowest rate estimates coincide with the highest frequency of host-jumping, counter to their adaptive model.

Isolates of HBV dating back to the sixteenth century lack temporal structure6. Accurately estimating an evolutionary rate therefore requires a longer sampling period6, as confirmed by the Bronze Age samples2. Hence, higher rate estimates for HBV are erroneous and likely reflect counts of transient mutations, as expected under time-dependent virus evolution7,8. Although there is temporal structure among recent B19 isolates9,10, this may again reflect time dependence or the impact of a small number of ancient samples in the regression. Simmonds et al. also claim that our earlier paper10 “predicted a time of origin of current genotype 1 strains to the 1960s or 1970s”. However, no divergence times were presented in this paper, but were previously by Simmonds and colleagues9. Moreover, no data are presented for the lowest rate of spumavirus evolution covering ~750 million years. The relatively slow evolution of spumaviruses may reflect low rates of replication11 and the occurrence of nonsynonymous substitutions argues against extreme purifying selection11.

We believe that there is no biological reason why the evolutionary rate in RNA viruses, which encode their own RNA polymerase, should decline to that of hosts that use entirely different replication enzymes. The likely changing nature of the complex environments faced by viruses, combined with the size of sequence space (~410,000 for a typical RNA virus), make evolutionary stasis unrealistic. For example, although influenza viruses have probably been associated with wild bird species for millennia, their evolutionary rates are of the same magnitude as the mammals in which they periodically emerge12,13. Finally, the high mutation rates in RNA viruses ensure that evolutionary stasis would result in massive purifying selection and an enormous mutational load.

Viruses are ancient entities and their remarkable sequence diversity reflects a long evolutionary history characterized by high rates of genetic change. Along with host adaptation, much of the apparent discrepancy between short-term and long-term evolution in viruses may be an illusion caused by the inappropriate use of molecular clock dating without temporal structure, incorrect calibration points, differences in replication rates, site saturation that is even apparent at shallow genetic distances14 and the inherent time dependence of evolutionary rates8.

There is a reply to this letter by Simmonds, P., Aiewsakun, P. & Katzourakis, A. Nat. Rev. Microbiol. https://doi.org/10.1038/s41579-019-0169-6 (2019).