Outbreak reconstruction with a slowly evolving multi-host pathogen: A comparative study of three existing methods on...

In a multi-host system, understanding host-species contribution to transmission is key to appropriately targeting control and preventive measures. Outbreak reconstruction methods aiming to identify who-infected-whom by combining epidemiological and genetic data could contribute to achieving this goal. However, the majority of these methods remain untested on realistic simulated multi-host data. Mycobacterium bovis is a slowly evolving multi-host pathogen and previous studies on outbreaks involving both cattle and wildlife have identified observation biases. Indeed, contrary to cattle, sampling wildlife is difficult. The aim of our study was to evaluate and compare the performances of three existing outbreak reconstruction methods (seqTrack, outbreaker2 and TransPhylo) on M. bovis multi-host data simulated with and without biases. Extending an existing transmission model, we simulated 30 bTB outbreaks involving cattle, badgers and wild boars and defined six sampling schemes mimicking observation biases. We estimated general and specific to multi-host systems epidemiological indicators. We tested four alternative transmission scenarios changing the mutation rate or the composition of the epidemiological system. The reconstruction of who-infected-whom was sensitive to the mutation rate and seqTrack reconstructed prolific super-spreaders. TransPhylo and outbreaker2 poorly estimated the contribution of each host-species and could not reconstruct the presence of a dead-end epidemiological host. However, the host-species of cattle (but not badger) index cases was correctly reconstructed by seqTrack and outbreaker2. These two specific indicators improved when considering an observation bias. We found an overall poor performance for the three methods on simulated biased and unbiased bTB data. This seemed partly attributable to the low evolutionary rate characteristic of M. bovis leading to insufficient genetic information, but also to the complexity of the simulated multi-host system. This study highlights the importance of an integrated approach and the need to develop new outbreak reconstruction methods adapted to complex epidemiological systems and tested on realistic multi-host data.