OOS 33-9
Identifying triggers of Ebola spillover events using spatio-temporal envelop models

Tuesday, August 11, 2015: 4:20 PM
344, Baltimore Convention Center
John Paul Schmidt, Odum School of Ecology, University of Georgia, Athens, GA, USA
Andrew W. Park, , Odum School of Ecology, University of Georgia, Athens, GA, USA
John M. Drake, Odum School of Ecology, University of Georgia, Athens, GA, USA
Laura Alexander, Ecology, University of Georgia, Athens, GA, USA
Background/Question/Methods

The spatial and temporal conditions which have lead to Ebola outbreaks occurring since the 1970s are poorly understood. Resolving the mystery is now more critical given the recent and catastrophic expansion of Ebola in West Africa. To this end, we constructed a spatio-temporal envelope model to identify possible combinations of spatial and temporal factors that form conditions sufficient to generate Ebola spillover events. From the existing literature, we narrowed documented Ebola occurrences that are likely discrete spillover events to 64 locations/dates, 34 of which resulted in human infections. As predictors of spillover events, we used a gridded time series of monthly rainfall data modeled from satellite imagery (available from USGS) to capture climate cycles, decadal estimates of human population density 1960-2010 (available from NASA’s SEDAC) to capture demographic change, and, as biotic/environmental variables, we relied on static geographic data in the form of gridded climate averages (WorldClim), recent land use classifications of the region, and mammal, fruit bat, all other bat, rodent, ungulate, and primate diversity. Analyses were performed using species distribution modeling algorithms.

Results/Conclusions

The most important spatial effects were fruit bat species richness and rainfall variables within the tropical forest climate zone in Africa. Human population density was an important spatio-temporal effect, and seasonal peaks in rainfall appear to be a temporal driver of spillover events. A small set, 4-5, spatio-temporal variables show high accuracy in predicting where and when Ebola spillover events occur.