The spread of rabies in Ethiopian wolves: from field data to transmission parameters

With fewer than 500 individuals remaining, the Ethiopian wolf is an endangered species threatened by regular outbreaks of rabies which have occurred in 1992-1993, 2003-2004, and 2008-2009. An important step towards effective control planning is to quantify the spread of infection. However, epidemic processes are often only partially observed, making parameter inference hard. Here we explore the use of Approximate Bayesian Computation to assess rabies transmission in Ethiopian wolves during the 2008-2009 outbreak. We developed a spatially explicit stochastic SEIR model for infection spread within a metapopulation ofseven packs. Two parameters were inferred: the within (βw) and between pack (βb) transmission rates. The data comprised pre and post-outbreak pack compositions, and date and place of vaccination events and carcass recoveries. We used the model to simulate datasets using values for βw and βb sampled from priordistributions. These simulated data were reduced to summary statistics (number of carcasses per group of packs, total time of outbreak, and time between the first carcass and the first carcass in a neighbouring pack).Those parameters leading to the smallest distances between simulated and observed summary statistics were accepted, using an algorithm based on local linear regression. The basic reproduction number (R0) was then computed. The median value of βw was approximately eight times higher than βb, consistentwith behavioral studies. The mean R0 was 2.6, in agreement with previous estimates for the 2003 rabies outbreak. Besides biological insights following the estimation of transmission parameters, these outputs will be used to calibrate a simulation model to assess theeffectiveness of different vaccination strategies in populations of Ethiopian wolves.