Disentangling time-scales to explore persistence of leptospirosis in California sea lions

Background/Question/Methods

Understanding the mechanisms underlying pathogen persistence is a fundamental component of disease control and surveillance. However, signatures of long-term persistence mechanisms may be obscured by shorter-term seasonal drivers. Thus, determining how seasonally varying host contact patterns and birth pulses interact with persistence mechanisms, such as chronic shedding of infectious pathogens by individual hosts, will depend crucially on the time-scale over which these processes are considered. Here, we explore seasonality and persistence of the causative agent of leptospirosis, Leptospira interrogans serovar Pomona, in California sea lions (Zalophus californianus). Using a unique long-term dataset of strongly seasonal leptospirosis incidence in sea lions and empirically derived host birth pulse functions, we fit seasonal and constant models of Leptospira transmission both with and without chronic shedding. Models are fit to time series of leptospirosis incidence of varying length ranging from 0.5 to 3 years. We then simulate our best fit models out over 30 years to assess persistence.

Results/Conclusions

Our best fit models show that chronic shedding of leptospires by sea lions is not necessary for long-term persistence (i.e., all models predicted persistence). However, over time-scales greater than 2 years, models with chronic shedding were much better at explaining the observed patterns in the time series. Additionally, models with seasonal transmission are always better fits to the data than their non-seasonal counterparts, and models with a distinct birth pulse are needed to capture dynamics in data that cover 1-2 years. Consequently, our data shows time-scales that are fundamental to observing mechanisms driving Leptospira dynamics – short for seasonal transmission, intermediate for host birth pulses, and long for chronic shedding. This result highlights the need to understand how the time-scale of study impacts inference about the relative importance of processes involved in infectious disease dynamics.