Livestock as sentinels for an infectious disease in a sympatric or adjacent-living wildlife reservoir host

• Published in: Preventive veterinary medicine 2017-12, Vol.148, p.106-114
• Main Authors: Anderson, D.P., Gormley, A.M., Bosson, M., Livingstone, P.G., Nugent, G.
• Format: Article
• Language: English
• Subjects: Animals; Bayesian logic; Bovine tuberculosis; Cattle; Disease Reservoirs - veterinary; Eradication; Introduced Species; Mycobacterium bovis; Mycobacterium bovis - isolation & purification; New Zealand - epidemiology; Prevalence; Sentinel Surveillance - veterinary; Surveillance; Trichosurus - microbiology; Tuberculosis - epidemiology; Tuberculosis - microbiology; Tuberculosis - prevention & control; Tuberculosis - veterinary; Tuberculosis, Bovine - epidemiology; Tuberculosis, Bovine - microbiology; Tuberculosis, Bovine - prevention & control
• ISSN: 0167-5877; 1873-1716
• DOI: 10.1016/j.prevetmed.2017.10.015

A central question to address in managing wildlife diseases is how much effort and resources are required to reduce infection prevalence to below a requisite threshold? This requires surveillance for infection in at least one species involved in the infection cycle, a process that is often expensive and time-consuming but one which could be enhanced using additional sources of readily-obtainable surveillance data. We demonstrate how surveillance data from ruminant livestock monitored for bovine tuberculosis (bTB) in New Zealand can be employed in spatially-explicit modelling to help predict the probability of freedom from Mycobacterium bovis infection in a sympatric wildlife reservoir species, the brushtail possum (Trichosurus vulpecula). We apply the model to a case study and compare resulting probabilities of freedom when utilizing (1) livestock data only, (2) wildlife data only, and (3) combined livestock-plus-wildlife surveillance data. Results indicated that the greatest probability of M. bovis eradication was achieved using wildlife monitoring data supplemented with livestock surveillance data. This combined approach lessened the time required for a confident (95% probability) declaration of regional eradication. However, the combined model was sensitive to the precision of the input parameters, and we describe ways to account for this. In a broad sense, this modelling approach is flexible in that any spatial arrangement of wildlife habitat and farms can be analysed, provided infection is readily detectable in both the wild and domestic animal(s) of interest. It is applicable to monitoring any communicable wildlife disease that affects regularly-tested livestock. The potential benefits to wildlife disease management include reduced surveillance costs and more rapid achievement of targeted reductions in disease prevalence.