Fig 1.
A) Locations of dog study sites on the Chari River, Chad. B) Adult female Guinea worm Dracunculus medinensis being extracted from a domestic dog. C) Typical dog fitted with GPS tracking collar. The settlements were: Magrao (10°59'44.31"N, 15°29'29.27"E) which encompasses the associated permanent village of Sawata, Largana (10°45'15.26"N, 16° 1'39.86"E) and the associated permanent village of Gomba (10°44'46.22"N, 16° 2'46.70"E), and Kakale-Mberi (10°53'0.79"N, 15°38'8.45"E) and the associated seasonal settlement of Awine (10°48'6.34"N, 15°37'56.61"E). Guelendeng is the major town in the sous-préfecture. Photographs Jared Wilson-Aggarwal. The satellite image was generated using the Esri world imagery basemap (sources: Esri, DigitalGlobe, GeoEye, i-cubed, USDA FSA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP, swisstopo, and the GIS User Community).
Table 1.
Summary of dog characteristics in three settlements in Chad.
Data are from all collared dogs (n = 150), but some dogs have missing data for individual characteristics. Water provision is the proportion of households which reported providing water for animals.
Table 2.
Summary of dog ranging behaviour in three settlements in Chad.
Core range is the 60% kernel density estimate and total range is the 100% minimum convex polygon. The mean and standard error in parentheses are presented for each variable.
Table 3.
Summaries of analyses of variation in dog range sizes, time spent by dogs in the vicinity of ponds and history of Guinea worm infection in dogs in rural Chad.
Core range is the 60% kernel density estimate and total range is the 100% minimum convex polygon. History of Guinea worm infection is from our field survey records and records held by the Chad Guinea Worm Eradication Program. Model outcomes are shown as the top model sets (ΔAICc < 2 from the top model) for all analyses.
Fig 2.
Distances from human households of the ponds most exposed to visits by dogs in two settlements in Chad.
A is Kakale and B is Magrao. Dog visits were determined by locating dogs within 100m of a pond by GPS tracking. Ponds were located from satellite imagery within a search area enclosing all dog location points for each village and are a sub-sample of all available ponds. All visits by all dogs were summed for each pond and the cumulative total of all dog visits is plotted against the distance of the pond to the nearest household with tracked dogs, with the red line from a fitted generalised additive model. The dotted lines indicate the distance from a household with dogs at which 80%, 90% and 95% of all dog visits are captured.
Fig 3.
Stable isotope analysis of the diets of dogs in two settlements in Chad.
Settlements are A) Kakale and B) Magrao. i) The δ15N and δ13C values for sampled dogs (black dots) and the mean ± standard deviation of δ15N and δ13C for putative food groups. Trophic discrimination factors, derived from package SIDER, have been applied to adjust dog isotope ratios downwards for both δ15N and δ13C. ii) Outputs of Bayesian mixing models, from package SIMMR, of the proportional contributions of the main food groups to dog diets.
Fig 4.
Effects of water provision and of fish consumption on Guinea worm infection history in dogs in Chad.
The odds, with 95% confidence intervals, of having had Guinea worm are shown for dogs A) living in households in which animals are, and are not, provided with water, and B) with different modelled proportions of fish in their diets. The effects of fish consumption are presented separately for dogs living in households, in which animals are (dark grey) and are not (light grey) provided with water. Odds are from exponentiated model coefficients describing the relationship between the history of Guinea worm infection from our field survey, averaged across the top model set and for the different levels of values of predictive factors. Results presented are from dogs in Kakale but dogs in Magrao show the same relationships.