Host-pathogen coevolution is assumed to play a key role in eco-evolutionary processes, including epidemiological dynamics and the evolution of sexual reproduction [1-4]. Despite this, direct evidence for host-pathogen coevolution is exceptional [5-7], particularly in vertebrate hosts. Indeed, although vertebrate hosts have been shown to evolve in response to pathogens or vice versa [8-12], there is little evidence for the necessary reciprocal changes in the success of both antagonists over time [13]. Here, we generate a time-shift experiment to demonstrate adaptive, reciprocal changes in North American house finches (Haemorhous mexicanus) and their emerging bacterial pathogen, Mycoplasma gallisepticum [14-16]. Our experimental design is made possible by the existence of disease-exposed and unexposed finch populations, which were known to exhibit equivalent responses to experimental inoculation until the recent spread of genetic resistance in the former [14, 17]. Whereas inoculations with pathogen isolates from epidemic outbreak caused comparable sub-lethal eye swelling in hosts from exposed (hereafter adapted) and unexposed (hereafter ancestral) populations, inoculations with isolates sampled after the spread of resistance were threefold more likely to cause lethal symptoms in hosts from ancestral populations. Similarly, the probability that pathogens successfully established an infection in the primary host and, before inducing death, transmitted to an uninfected sentinel was highest when recent isolates were inoculated in hosts from ancestral populations and lowest when early isolates were inoculated in hosts from adapted populations. Our results demonstrate antagonistic host-pathogen coevolution, with hosts and pathogens displaying increased resistance and virulence in response to each other over time.

Host resistance through immune clearance is predicted to favor pathogens that are able to transmit faster and are hence more virulent. Increasing pathogen virulence is, in turn, typically assumed to be mediated by increasing replication rates. However, experiments designed to test how pathogen virulence and replication rates evolve in response to increasing host resistance, as well as the relationship between the two, are rare and lacking for naturally evolving host–pathogen interactions. We inoculated 55 isolates of Mycoplasma gallisepticum, collected over 20 y from outbreak, into house finches (Haemorhous mexicanus) from disease-unexposed populations, which have not evolved protective immunity to M. gallisepticum. We show using 3 different metrics of virulence (body mass loss, symptom severity, and putative mortality rate) that virulence has increased linearly over >150,000 bacterial generations since outbreak (1994 to 2015). By contrast, while replication rates increased from outbreak to the initial spread of resistance (1994 to 2004), no further increases have occurred subsequently (2007 to 2015). Finally, as a consequence, we found that any potential mediating effect of replication rate on virulence evolution was restricted to the period when host resistance was initially increasing in the population. Taken together, our results show that pathogen virulence and replication rates can evolve independently, particularly after the initial spread of host resistance. We hypothesize that the evolution of pathogen virulence can be driven primarily by processes such as immune manipulation after resistance spreads in host populations.

The virulence-transmission trade-off hypothesis has provided a dominant theoretical basis for predicting pathogen virulence evolution, but empirical tests are rare, particularly at pathogen emergence. The central prediction of this hypothesis is that pathogen fitness is maximized at intermediate virulence due to a trade-off between infection duration and transmission rate. However, obtaining sufficient numbers of pathogen isolates of contrasting virulence to test the shape of relationships between key pathogen traits, and doing so without the confounds of evolved host protective immunity (as expected at emergence), is challenging. Here, we inoculated 55 isolates of the bacterial pathogen, Mycoplasma gallisepticum, into non-resistant house finches (Haemorhous mexicanus) from populations that have never been exposed to the disease. Isolates were collected over a 20-year period from outbreak in disease-exposed populations of house finches and vary markedly in virulence. We found a positive linear relationship between pathogen virulence and transmission rate to an uninfected sentinel, supporting the core assumption of the trade-off hypothesis. Further, in support of the key prediction, there was no evidence for directional selection on a quantitative proxy of pathogen virulence and, instead, isolates of intermediate virulence were fittest. Surprisingly, however, the positive relationship

Understanding how hosts minimize the cost of emerging infections has fundamental implications for epidemiological dynamics and the evolution of pathogen virulence. Despite this, few experimental studies in natural populations have tested whether, in response to disease emergence, hosts evolve resistance, which reduces pathogen load through immune activation, or tolerance, which limits somatic damages without decreasing pathogen load. Further, none has done so accounting for significant natural variation in pathogen virulence, despite known effects on host responses to infection. Here, we investigate whether eastern North American house finches (Haemorhous mexicanus) have evolved resistance and/or tolerance to their emerging bacterial pathogen, Mycoplasma gallisepticum. To do so, we inoculated finches from disease‐exposed and disease‐unexposed populations with 55 distinct isolates of varying virulence. First, although peak pathogen loads, which occurred approximately eight days postinoculation, did not differ between experimentally inoculated finches from disease‐exposed versus unexposed population, pathogen loads subsequently decreased faster and to a greater extent in finches from exposed populations. These results suggest that finches from exposed populations are able to clear the infection through adaptive immune processes. Second, however, finches from exposed populations also displayed lower symptom severity for a given pathogen load, suggesting that a damage‐limitation mechanism, or tolerance, has accompanied the evolution of immune clearance. Our results highlight that resistance and tolerance should be seen as complementary, not alternative, defense strategies: the evolution of resistance benefits from the concomitant evolution of tolerance mechanisms that protect against the damage of immune activation, whereas the evolution of tolerance without resistance will risk runaway selection on pathogen virulence.

Quantifying variation in the ability to fight infection among free-living hosts is challenging and often constrained to one or a few measures of immune activity. While such measures are typically taken to reflect host resistance, they can also be shaped by pathogen effects, for example, if more virulent strains trigger more robust immune responses. Here, we test the extent to which pathogen-specific antibody levels, a commonly used measure of immunocompetence, reflect variation in host resistance versus pathogen virulence, and whether these antibodies effectively clear infection. House finches (Haemorhous mexicanus) from resistant and susceptible populations were inoculated with > 50 isolates of their novel Mycoplasma gallisepticum pathogen collected over a 20-year period during which virulence increased. Serum antibody levels were higher in finches from resistant populations and increased with year of pathogen sampling. Higher antibody levels, however, did not subsequently give rise to greater reductions in pathogen load. Our results show that antibody responses can be shaped by levels of host resistance and pathogen virulence, and do not necessarily signal immune clearance ability. While the generality of this novel finding remains unclear, particularly outside of mycoplasmas, it cautions against using antibody levels as implicit proxies for immunocompetence and/or host resistance.