Bayesian Analysis for Inference of an Emerging Epidemic: Citrus Canker in Urban Landscapes

Neri, Franco M.; Cook, Alex R.; Gibson, Gavin J.; Gottwald, Tim R.; Gilligan, Christopher A.

One of the principal challenges in managing emerging epidemics is to predict the likely future development of disease in order to assess the severity of the invasion prior to instituting control measures. However, prediction is difficult when little is known about how a new pathogen is likely to continue to spread in an alien environment, and frequently the underlying epidemiological parameters that influence the spread of disease are not known. Even when there is prior knowledge of a pathogen, as for example foot and mouth epidemics in the UK in 1967, 1982 and 2001, different pathogen strains, changes in farming practices or environmental conditions can markedly change the extent and speed of disease spread through the landscape [7, 9, 22, 23] . Whereas, for example, the spread of foot and mouth disease in the 1967 epidemic was relatively localised, occurring mainly by aerial dispersal [24] , changes in the frequency and distance of livestock movements over large distances [25] led to a strikingly, topologically different epidemic in 2001 [7, 9] . Numerous other examples have been reported of variability in epidemic outcome upon reintroductions of emerging pathogens. This is problematic, because rapid decisions about the introduction of disease control strategies often have to be made early in the course of an emerging epidemic. Sometimes, options are clear. Immediate control aimed at eradication is initiated as soon as an outbreak is detected for certain statutory diseases. Actions against the H1N1-2009 pandemic influenza worldwide [26] , foot and mouth disease in the UK [8] , and Asian soya bean rust in several US states [27] are good examples amongst others of human, livestock and crop diseases that attract an immediate response.