Changes in the proportions of hosts that are infectious with different strains within a 2 variants per epitope, 5 epitope system in an “evolutionary free” framework: for all of the possible 32 strains the existence of at least one carrier was assured and no antigenic mutations were introduced.

<p>The superimposed time series were smoothed and ordered back to front by peak prevalence, maintaining the least prevalent strain in the front. The 3th highest peaking strain was outlined as an example. Single strain dominance was calculated based on the quantity ε from <a href="http://www.plospathogens.org/article/info:doi/10.1371/journal.ppat.1003104#ppat.1003104-Recker1" target="_blank">[22]</a>. Major peaks of incidence are generally associated with one or two dominant antigenic-phenotypes with ε = 0.36±0.07 (mean ± standard-deviation across 5 simulations) and a myriad of lower prevalence ones. Antigenic-phenotypes reemerge with alternating frequency. This simulation includes a single homogeneously mixed host population of 40M hosts, contact rate β = 0.6 and a 4 day recovery rate. Each epitope unencountered by the host contributes to a 17.5% increase in the risk of infection with a different strain (see <a href="http://www.plospathogens.org/article/info:doi/10.1371/journal.ppat.1003104#s4" target="_blank">methods</a> for full description of epidemiological parameters).