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. We conducted an infection experiment in a North American songbird using 55 distinct isolates of its emerging infectious bacterial pathogen that differed in virulence. We demonstrate that more virulent variants transmitted faster, but had shorter infection durations, leading to variants of intermediate virulence being more evolutionarily successful. Interestingly, we did not find support for the common suggestion that the number of pathogen cells underpins their virulence and transmission rate. 

Wild house finches from populations that are outside the current range of M. gallisepticum, and have thus not evolved genetic resistance, were captured in Arizona during summer 2015 (N = 118, 64 males and 54 females). The birds were divided into 2 groups of 59 birds. Each of 55 M. gallisepticum isolates sampled over the course of the epidemic was inoculated into one bird selected at random from the 59 birds of the first group (30 males and 29 females), with 4 isolates inoculated in two birds each. Birds from the second group served as uninfected sentinels. Protocols were approved by Institutional Animal Care and Use Committees (IACUC) of Arizona State University (permit #15-1438R), as well as by Institutional Biological Use Authorizations to Auburn University (# BUA 500), and the University of Exeter’s ethics committee.

Putative host mortality was determined in situ on days 3, 6, 8, 14, 21, 25, 28 and 34 post-infection, with infection considered lethal when birds lost the avoidance response to hand capture. A quantitative proxy for virulence was also obtained from the extent of conjunctival swelling measured from photographs taken at 6, 13, 25 and 34 days post-inoculation (dpi) from a standardized distance. Transmission rate to the sentinel was determined by amplification of M. gallisepticum DNA from conjunctival and tracheal swabs obtained from the sentinel at 2, 3, 4, 5, 6, 7, 8, 11, 14, 17, 20, 23, 26, 29, 32 and 35 days post-inoculation. Fitness was calculated as the product of infection duration (in days to putative mortality) and transmission rate (in days^-1). Bacterial load was measured by quantitative amplification of M. gallisepticum DNA from pooled conjunctival and tracheal swabs obtained at 8, 14, 21 and 28 dpi. 

To test the associations between transmission rate and mortality, transmisssion rate and conjunctival swelling, and fitness and conjunctival swelling, we considered only the isolates that had successfully established an infection (infection =1). To test the association between pathogen load/replication rate and transmission, we considered only the isolates that successfully replicated during the experiment (included =1).