Multiple peaks are a common feature in epidemic curves for a variety of pathogens. Classical epidemiological models do not generally produce this multimodal pattern. We demonstrate this phenomenon using data on the weekly prevalence of West Nile virus-positive dead birds reported to and mosquito pools collected by the New York City Department of Health and Mental Hygiene, between 2000 and 2008. We hypothesize that multimodality in this system is driven by the presence of multiple passerine hosts with unimodal epidemic curves. We adapt a temperature-driven seasonal deterministic epidemiological model for Usutu virus dynamics by Rubel et al., to study the conditions under which such multimodal epidemic curves are predicted in multi-host pathogens. We show that while this model cannot produce multimodal epidemic curves with a single host species, the inclusion of a second host species can lead to a multimodal pattern under specific conditions. We investigate these necessary conditions through a sensitivity analysis using Latin Hypercube Sampling, k-nearest neighbor and local linear regression of the parameters against the number of peaks in the joint epidemic curve. 
Results/Conclusions

Results show that the most important parameter is the difference between the date of the start of breeding for the host that starts to breed later and the date of the first peak in WNv prevalence, which is highly dependent on the date at which mosquitoes start to emerge. Numerical evaluation of the basic reproductive ratio of the seasonal model shows that annual epizootics start abruptly when mosquitoes start to emerge, peak in the late summer/early fall and stop as mosquitoes hibernate with the shortening of the daylight hours. Temporal separation in the breeding of different hosts of multihost pathogens is therefore a potential mechanism for multimodal prevalence patterns in a host community. The model predicts that in the case of multimodal patterns, different host species contribute dominantly to different peaks. However, such a pattern is not apparent in the species found WNv+ in NYC. Therefore, it’s plausible that a different mechanism, potentially the temporal separation of multiple clutches of the same species, causes multimodal patterns of WNv prevalence in New York City.