Infected host competence overshadows heterogeneity in susceptibility in shaping experimental epizootics

Abstract

The accelerated rate of disease emergence in recent decades underscores the need to understand conditions that promote or dampen epidemics. Theoretical models consistently show that epidemics are smaller in populations with higher among-individual heterogeneity in susceptibility. Experimental tests of these predictions are rare but critical for understanding how heterogeneity in susceptibility shapes epidemics in natural systems. We directly link data-parameterized models from previous dose response experiments in the house finch and Mycoplasma gallisepticum system to experimental epidemics in replicated aviary mesocosm flocks. We manipulated flock-level heterogeneity in susceptibility by seeding epidemics in flocks composed of either pathogen-naive or previously exposed birds, which prior work showed have higher heterogeneity in susceptibility relative to pathogen-naive populations. We tracked epidemics for over two months, combining empirical data and stochastic compartmental models to address how heterogeneity in susceptibility changes epidemic severity. Consistent with previous work, estimates of heterogeneity in susceptibility based on coefficients of variation were higher for flocks given prior pathogen exposure relative to pathogen-naive flocks. However, in contrast with prior work on individually-housed birds which showed relatively homogeneous susceptibility for pathogen-naive birds, the pathogen-naive flocks in this study were better described by heterogeneous, rather than homogenous, models of susceptibility. This suggests that flock-level epidemics captured sources of heterogeneity absent in controlled experiments, such as transmission heterogeneity. Finally, although prior exposure conferred protection from disease at the individual level, we did not detect predicted effects of prior exposure and its associated flock-level heterogeneity on prevalence. However, our ability to detect effects of prior exposure on flock-level prevalence was obscured by unexpected variation in the competence of the initially pathogen-naive index birds that seeded each epidemic. This variation in infectiousness among index birds significantly predicted flock-level prevalence, with low index bird infectiousness contributing to the absence of detectable epidemics in two of the three naive flocks. Our stochastic simulations generated a wide range of prevalence outcomes for small epidemics over the timescales examined, further underscoring the challenges of measuring transmission dynamics in naturalistic settings, where unexpected variation in host traits such as competence can obscure other factors of interest. Open research statementData are not yet provided. Data and code will be permanently and publicly archived in the Virginia Tech Data Repository if the paper is accepted for publication