Abstract
Aim of this study was to reconstruct the phylogeography of variola virus (VARV) in the XX century, using 47 VARV whole genome sequences available in public databases, through two different methods for ancestral character reconstruction: a frequently used Bayesian framework and a fast maximum-likelihood (ML) based method. The substitution rate of the whole VARV genome was estimated to be between 6.7×10(-6) and 1.1×10(-5) substitutions/site/year. Both ML and Bayesian methods gave similar trees topology, showing two distinct monophyletic groups: one (known as P1) including the great part of variola major and the second (P2) including West African and American (variola minor) isolates and close evolutionary rate estimations, between 6.73×10(-6) and 1.1×10(-5) for the whole genome. The phylogeographical reconstruction of P1 suggested that the common ancestor of the variola major circulating in the Old World between the 1940s and the 1970s most probably originated in the Far East in the first decades of the XX century, and then spread to Indian subcontinent in the 1920s. India represented a center of further spread of VARV to eastern Africa in the 1940s and to the Middle East in the 1960s. The phylogeographic scenario obtained by the maximum-likelihood based method was congruent with that obtained by Bayesian framework, but the analysis was faster indicating the usefulness of this method in the analyses of large viral genomes. Our results may help to explain the controversial reconstructions of the history of VARV obtained using long or short timescale for calibration.