Accounting for population structure reveals ambiguity in the Zaire Ebolavirus reservoir dynamics

Abstract

Ebolaviruses pose a substantial threat to wildlife populations and to public health in Africa. Evolutionary analyses of virus genome sequences can contribute significantly to elucidate the origin of new outbreaks, which can help guide surveillance efforts. The reconstructed between-outbreak evolutionary history of Zaire ebolavirus so far has been highly consistent. By removing the confounding impact of population growth bursts during local outbreaks on the free mixing assumption that underlies coalescent-based demographic reconstructions, we find-contrary to what previous results indicated-that the circulation dynamics of Ebola virus in its animal reservoir are highly uncertain. Our findings also accentuate the need for a more fine-grained picture of the Ebola virus diversity in its reservoir to reliably infer the reservoir origin of outbreak lineages. In addition, the recent appearance of slower-evolving variants is in line with latency as a survival mechanism and with bats as the natural reservoir host.

Fig 1. Evolutionary history of EBOV estimated from the available full genomes.

The tree is rooted with the Yambuku outbreak sequences as outgroup [13, 14, 49]. Numbers above branches indicate support values obtained with the approximate likelihood ratio test. The red arrow highlights the position of the newly obtained full genome from the 2017 outbreak in Likati, COD. Representative sequences from each outbreak that were selected for further phylogenetic analyses are indicated in orange. Grey rectangles indicate lineages from outbreaks in the COD as of 2014 that appear to have been evolving at a slower pace than expected. The scale bar is in units of substitutions per site.

Fig 2. The between-outbreak epidemic history based on the EBOV2018 data set.

Values next to branches represent their posterior probability. Bars show the 95% HPD interval for the internal node heights. Branch root posterior probabilities were obtained with RootAnnotator [49]. Branch root positions with >1% posterior probability are indicated in the phylogeny by colored circles and their support is given in the legend.

Fig 3. Time-scaled between-outbreak EBOV evolutionary history inferred without coalescent prior, represented by the maximum clade credibility summary phylogeny.

Branch root posterior probabilities were obtained with RootAnnotator [49]. The 5 best supported branch root positions are indicated in the phylogeny by colored circles and their support is given in the legend. Numbers next to branches indicate their posterior support.