Persistence of disease in territorial animals: insights from spatial models of Tb

Early models of directly transmitted wildlife disease focused on rabies transmission as a travelling wave, usually in a homogeneous density of wildlife. Such models of epi-enzootic diseases paid little attention to local-scale disease prevalence. Historical data on bovine tuberculosis (Tb) in cattle indicates that very localised areas can suffer from frequent repeat breakdowns, indicating that some environmental factors might be the cause. There are a number of different ways to simulate such local disease ‘hotspots’ in wildlife, and these resultant hotspots may mean that, overall, wildlife disease prevalence is very low. However, spatial and temporal persistence of this hotspot is more difficult to model. This heterogeneity in disease prevalence is difficult to produce in non-spatial models, and is one of the reasons why such models gave poor predictions of disease dynamics in the field. For example, Nigel Barlow struggled with finding a way to produce this spatial heterogeneity in mathematical models, culminating in his 2000 paper in Journal of Animal Ecology. This gave a phenomenological treatment, but not a causative solution. I take a look at the various causative methods of producing disease heterogeneity in simulation models of Tb, a chronic wildlife disease. These include (1) chance, (2) model artefacts, (3) population (e.g. demographic, genetic) heterogeneity and (4) environmental heterogeneity. I further argue that only (4) can be predicted over a medium timescale, and propose methods to assess the contribution of (1) and (2) in a model. I also discuss how spatial heterogeneity may affect Tb management.