Participatory mapping identifies risk areas and environmental predictors of endemic anthrax in rural Africa.

Disease mapping reveals geographical variability in incidence, which can help to prioritise control efforts. However, in areas where this is most needed, resources to generate the required data are often lacking. Participatory mapping, which makes use of indigenous knowledge, is a potential approach to identify risk areas for endemic diseases in low- and middle-income countries. Here we combine this method with Geographical Information System-based analyses of environmental variables as a novel approach to study endemic anthrax, caused by the spore-forming bacterium Bacillus anthracis, in rural Africa. Our aims were to: (1) identify high-risk anthrax areas using community knowledge; (2) enhance our understanding of the environmental characteristics associated with these areas; and (3) make spatial predictions of anthrax risk. Community members from the Ngorongoro Conservation Area (NCA), northern Tanzania, where anthrax is highly prevalent in both animals and humans, were asked to draw areas they perceived to pose anthrax risks to their livestock on geo-referenced maps. After digitisation, random points were generated within and outside the defined areas to represent high- and low-risk areas, respectively. Regression analyses were used to identify environmental variables that may predict anthrax risk. Results were combined to predict how the probability of being a high-risk area for anthrax varies across space. Participatory mapping identified fourteen discrete high-risk areas ranging from 0.2 to 212.9 km in size and occupying 8.4% of the NCA. Areas that pose a high risk of anthrax were positively associated with factors that increase contact with Bacillus anthracis spores rather than those associated with the pathogen's survival: close proximity to inland water bodies, where wildlife and livestock congregate, and low organic carbon content, which may indicate an increased likelihood of animals grazing close to soil surface and ingesting spores. Predicted high-risk areas were located in the centre of the NCA, which is likely to be encountered by most herds during movements in search for resources. We demonstrate that participatory mapping combined with spatial analyses can provide novel insights into the geography of disease risk. This approach can be used to prioritise areas for control in low-resource settings, especially for diseases with environmental transmission.