A predictive model of ebolavirus spillover incorporating change in forests and human populations across spatial and temporal scales
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Past research has found associations between ebolavirus spillover and forest loss and fragmentation, although most predictions of the spatial distribution of ebolaviruses have not utilized these data. Spatial and temporal scales of covariate data measurement have also not previously been accounted for in predictive models of ebolavirus spillover, making it difficult to account for variables that influence ebolavirus transmission dynamics, such as movement and interaction of human and animal populations, trade, and animal behavioral responses to human presence and a changing environment. Using annual data on forest cover, forest loss and fragmentation, human population distribution, and meteorological variables, we developed models of ebolavirus spillover events from 2001-2021 to estimate the annual relative odds of ebolavirus spillover in equatorial Africa in 2021 and 2022. Analyses were done separately for all ebolavirus species (All-species analysis) and Zaire ebolavirus alone (Zaire-only analysis). Locations with the highest estimated relative odds of spillover occurred in patches throughout Democratic Republic of the Congo (DRC), Republic of the Congo, Gabon, Cameroon, and coastal west Africa in a spatial trend that resembled that of maps of forest fragmentation and forest loss. Reduced analyses that ignored forest loss and fragmentation data produced estimates that were distinct from the full analyses, highlighting locations where forest loss and fragmentation drove model predictions, which included coastal west Africa, southern Cameroon, and parts of DRC. Estimated spillover odds and increase in spillover odds at 2022 spillover sites ranked among the highest in equatorial Africa, suggesting the potential of predictive analyses to prioritize locations for surveillance and research efforts.
Significance
Using annually updated data on forest change, human population distribution, and meteorological conditions, we developed models to estimate the relative odds of ebolavirus spillover in 2021 and 2022 in equatorial Africa, identifying locations with elevated odds of spillover and temporal shifts in spillover odds between the two years. Locations were identified whose estimates of spillover odds were driven by forest change. During 2022, two ebolavirus spillover events were identified in locations with elevated spillover odds estimates. Annually updated estimates of ebolavirus spillover can guide public health surveillance programs to high-risk locations at specific times, and guide potential ebolavirus reservoir sampling efforts to locations that recently experienced increases in spillover odds.
