Host structuring of parasite populations: Some theoretical and computational studies

Abstract

Because the ecological and the genetic interactions occurring between parasites belonging to different infections are constrained by the physical discreteness of the hosts, the host-parasite association imparts a spatial structure to the populations of parasitic microorganisms. Equating infections with demes or islands, the parasite population can be described by a variant of Wright's island model, in which recovery and infection correspond to extinction and colonization and superinfection corresponds to migration. Here we investigate some of the population genetic consequences of host structure using a combination of theoretical and computational methods. In our first study, we introduce a measure-valued process as a model for the evolution of an age-structured parasite metapopulation and show how to approximate this process using the measure flow generated by a jump-diffusion. We characterize the invariant measures and corresponding jump distributions for the approximation and apply these methods to an example involving a single locus subject to mutation, selection, and genetic drift within hosts and to bottlenecks and bias during transmission. When intrahost selection and transmission bias act discordantly, it is shown that the invariant measure and the jump distribution can differ substantially. We discuss the implications of such discordance for vaccine target selection and review the evidence for biased transmission of HIV-1. In our second study, we use a branching Fisher-Wright process to characterize diversity in an exponentially expanding epidemic. We derive a renewal equation for the persistence probability of the branching diffusion and show that with sufficiently rapid branching a set of k neutral alleles can persist indefinitely with positive probability. In the last study, we exploit the relationship between population recombination rates and superinfection rates to quantify intra-subtype superinfection by HIV-1 in populations from Africa, China, Thailand, Trinidad and Tobago, and the US. Comparison of the population recombination rates estimated for these data sets with those found for data sets simulated using a structured coalescent process representing HIV-1 evolution within an epidemiologically closed population indicates that per-sequence superinfection rates are probably not less than 15% of the corresponding infection rates.