Population size impacts host-pathogen coevolution.

Andrei Papkou,Hinrich Schulenburg,Mike-Christoph Barg,Rebecca Schalkowski,Svenja Koepper

doi:10.1098/rspb.2021.2269

Abstract

Ongoing host–pathogen interactions are characterized by rapid coevolutionary changes forcing species to continuously adapt to each other. The interacting species are often defined by finite population sizes. In theory, finite population size limits genetic diversity and compromises the efficiency of selection owing to genetic drift, in turn constraining any rapid coevolutionary responses. To date, however, experimental evidence for such constraints is scarce. The aim of our study was to assess to what extent population size influences the dynamics of host–pathogen coevolution. We used Caenorhabditus elegans and its pathogen Bacillus thuringiensis as a model for experimental coevolution in small and large host populations, as well as in host populations which were periodically forced through a bottleneck. By carefully controlling host population size for 23 host generations, we found that host adaptation was constrained in small populations and to a lesser extent in the bottlenecked populations. As a result, coevolution in large and small populations gave rise to different selection dynamics and produced different patterns of host–pathogen genotype-by-genotype interactions. Our results demonstrate a major influence of host population size on the ability of the antagonists to co-adapt to each other, thereby shaping the dynamics of antagonistic coevolution.

Highlights

The evolutionary success of species depends on their ability to adapt to a changing environment
We combined host–pathogen experimental evolution with time-shift experiments to assess the influence of host population size on host–pathogen coevolution
We found that population size generally increased fitness of the hosts from coevolution and host-adaptation conditions, when these were exposed to the ancestral pathogen

Summary

Introduction

The evolutionary success of species depends on their ability to adapt to a changing environment. The aim of our study was to assess the effect of host population size and periodic bottlenecks on coevolution using the nematode Caenorhabditis elegans and the Gram-positive bacterium Bacillus thuringiensis as an established, laboratorybased host–pathogen model [30,31,32,33]. Host and pathogen populations from transfers 1, 10 and 23 from the same replicate of the coevolution treatment (large, small and bottlenecked populations) were combined in all possible time-point combinations (figure 1b), resulting in 432 infection experiments (9 time-shifts × 3 population size treatments × 16 evolved replicate populations). Experimental treatments (control, adaptation, coevolution), population types (large, small or bottlenecked) and infection doses (whenever more than one dose was used) were modelled as fixed predictors, generation time as continuous predictor and population identities of biological replicates as random intercepts. We used the 16 biological replicates to determine frequencies of each of the four possible coevolutionary patterns (1 > 10 > 23, 1 < 10 < 23, 1 < 10 > 23, 1 > 10 < 23) and tested whether their distribution deviates from null expectation (0.25, 0.25, 0.25, 0.25), using exact multinomial tests (including Holm–Bonferroni adjustment of p-values)

Results

Findings

Discussion