Paige B Miller (1), and John M Drake (1)
(1) Odum School of Ecology, University of Georgia
Spread of infectious diseases depends on dispersal of the host organism through space. For many disease systems (e.g, White Nose Syndrome, Ebola, and Sudden Oak Death) this results in a patchy distribution of cases. Classical epidemiological theory, based on mean-field models, ignores this spatial aspect of transmission and assumes all individuals have an equal likelihood of becoming infected. Recent work has shown that in non-spatial models of disease spread temporal early warning signals from incidence data change in characteristic ways prior to a large shift the transmission system (i.e., a tipping point). Similarly, spatial statistics change prior to tipping points in spatial models of ecological systems (e.g., grassland-forest transitions). Therefore we wondered how reliable spatial early warning statistics are for predicting tipping points towards disease elimination compared with temporal early warning statistics. We also wondered if spatial early warning statistics are corrupted when transmission of the pathogen is non-constant (i.e., heterogeneous) through space. To examine the reliability of spatial early warning statistics for disease elimination, we first simulated a conceptual spatially-explicit, host-pathogen system. We then numerically calculated when the tipping point occurs (i.e., when Reff < 1) in our model due to depletion of susceptible individuals with increasing treatment uptake. We then compared the reliability of temporal and spatial early warnings statistics for disease elimination in homogeneous and heterogeneous environments. We find that spatial statistics for disease elimination are informative predictors of progress towards disease elimination.