Lyssaviruses are unsegmented RNA viruses causing rabies. Their vectors belo
ng to the Carnivora and Chiroptera orders. We studied 36 carnivoran and 17
chiropteran lyssaviruses representing the main genotypes and variants. We c
ompared their genes encoding the surface glycoprotein, which is responsible
for receptor recognition and membrane fusion. The glycoprotein is the main
protecting antigen and bears virulence determinants. Point mutation is the
main force in lyssavirus evolution, as Sawyer's test and phylogenetic anal
ysis showed no evidence of recombination. Tests of neutrality indicated a n
eutral model of evolution, also supported by globally high ratios of synony
mous substitutions (d(S)) to nonsynonymous substitutions (d(N)) (>7). Relat
ive-rate tests suggested similar rates of evolution for all lyssavirus line
ages. Therefore, the absence of recombination and similar evolutionary rate
s make phylogeny-based conclusions reliable. Phylogenetic reconstruction st
rongly supported the hypothesis that host switching occurred in the history
of lyssaviruses. Indeed, lyssaviruses evolved in chiropters long before th
e emergence of carnivoran rabies, very likely following spillovers from bat
s. Using dated isolates, the average rate of evolution was estimated to be
roughly 4.3 X 10(-4) d(S)/site/year. Consequently, the emergence of carnivo
ran rabies from chiropteran lyssaviruses was determined to have occurred 88
8 to 1,459 years ago. Glycoprotein segments accumulating more d(N) than d(S
) were distinctly detected in carnivoran and chiropteran lyssaviruses. They
may have contributed to the adaptation of the virus to the two distinct ma
mmal orders. In carnivoran lyssaviruses they overlapped the main antigenic
sites, II and III, whereas in chiropteran lyssaviruses they were located in
regions of unknown functions.