Geographic Range Overlap Rather than Phylogenetic Distance Explains Rabies Virus Transmission among Closely Related Bat Species

Abstract

The cross-species transmission (CST) of pathogens can have dramatic consequences, as highlighted by recent disease emergence events affecting human, animal and plant health. Understanding the ecological and evolutionary factors that increase the likelihood of disease agents infecting and establishing in a novel host is therefore an important research area. Previous work across different pathogens, including rabies virus (RABV), found that increased evolutionary distance between hosts reduces the frequency of cross-species transmission and of permanent host shifts. However, whether this effect of host relatedness still holds for transmission among recently diverged hosts is not well understood. We aimed to ask if high host relatedness can still increase the probability of a host shift between more recently diverged hosts, and the importance of this effect relative to ecological predictors. We first addressed this question by quantifying the CST frequency of RABV between North American bat species within the genus Myotis, using a multi-decade data set containing 128 nucleoprotein (N) RABV sequences from ten host species. We compared RABV CST frequency within Myotis to the rates of CST between nine genera of North American bat species. We then examined whether host relatedness or host range overlap better explains the frequency of CST seen between Myotis species. We found that at the within genus scale, host range overlap, rather than host relatedness best explains the frequency of CST events. Moreover, we found evidence of CST occurring among a higher proportion of species, and CST more frequently resulting in sustained transmission in the novel host in the Myotis dataset compared to the multi-genus dataset. Our results suggest that among recently diverged species, the ability to infect a novel host is no longer restricted by physiological barriers but instead is limited by physical contact. Our results improve predictions of where future CST events for RABV might occur and clarify the relationship between host divergence and pathogen emergence.

Keywords: Myotis bat; North America; cross-species transmission; genetic divergence; host shifts; host-pathogen interaction; niche overlap; rabies; range overlap.

Figure 1

Location and host species origins of samples used in this study (A) Map of Myotis species sample locations and host ranges, colours indicate host species range and points show the location of collection of a Myotis bat infected with rabies virus (RABV). (B) Maximum Clade Credibility tree of Myotis bat RABV nucleoprotein gene sequences. Branches are coloured by associated host species as specified in the map legend. Node circles indicate nodes with posterior support >0.9.

Figure 2

Frequency and characteristics of rabies virus cross-species transmission (CST) for ten North American bat species within the genus Myotis, compared to a dataset containing multiple genera of North American bats. Posterior distributions of (A) the mean rates (+1E⁻¹⁰) of inferred CST events per year per capita for all pairs of species on a log scaled x axis; (B) the proportion of CST events on internal branches (indicative of sustained transmission and host shifts); (C) the proportion of species pairs with non-zero rates.

Figure 3

Support and contribution of potential predictors in the phylogenetic Generalized Linear Model (GLM) of rabies CST events within the Myotis genus only. For each potential predictor, support is represented by an inclusion probability and a relative contribution indicated for log scale GLM coefficients conditional on the predictor being included in the model (posterior mean and 95% Bayesian confidence interval). Range overlap was the only supported predictor (BF = 27.9).

Figure 4

Frequency of inferred cross-species transmission events per year per capita among bat species as function of the genetic distances between species. Intervals of genetic distances values were delimited according to bin widths of 0.5 and means of mean number of expected CST events per year for each bin are shown. Point sizes are log-proportional to sample size (number of pairs of species) within each bin.