Not all surveillance data are created equal—A multi‐method dynamic occupancy approach to determine rabies elimination from wildlife

Abstract

Abstract A necessary component of elimination programmes for wildlife disease is effective surveillance. The ability to distinguish between disease freedom and non‐detection can mean the difference between a successful elimination campaign and new epizootics. Understanding the contribution of different surveillance methods helps to optimize and better allocate effort and develop more effective surveillance programmes. We evaluated the probability of rabies virus elimination (disease freedom) in an enzootic area with active management using dynamic occupancy modelling of 10 years of raccoon rabies virus (RABV) surveillance data (2006–2015) collected from three states in the eastern United States. We estimated detection probability of RABV cases for each surveillance method (e.g. strange acting reports, roadkill, surveillance‐trapped animals, nuisance animals and public health samples) used by the USDA National Rabies Management Program. Strange acting, found dead and public health animals were the most likely to detect RABV when it was present, and generally detectability was higher in fall–winter compared to spring–summer. Found dead animals in fall–winter had the highest detection at 0.33 (95% CI: 0.20, 0.48). Nuisance animals had the lowest detection probabilities (~0.02). Areas with oral rabies vaccination (ORV) management had reduced occurrence probability compared to enzootic areas without ORV management. RABV occurrence was positively associated with deciduous and mixed forests and medium to high developed areas, which are also areas with higher raccoon ( Procyon lotor ) densities. By combining occupancy and detection estimates we can create a probability of elimination surface that can be updated seasonally to provide guidance on areas managed for wildlife disease. Synthesis and applications . Wildlife disease surveillance is often comprised of a combination of targeted and convenience‐based methods. Using a multi‐method analytical approach allows us to compare the relative strengths of these methods, providing guidance on resource allocation for surveillance actions. Applying this multi‐method approach in conjunction with dynamic occupancy analyses better informs management decisions by understanding ecological drivers of disease occurrence.