Fig 1.
Schistosome life cycle, study design, and study location.
A) Schistosome life cycle: eggs are deposited into water by infected humans; miracidia hatch from eggs and infect freshwater snails; larval schistosomes develop and free-swimming cercariae emerge and are infective to humans. B) Visual description of our hypothesized relationship between nearshore human schistosome risk and offshore intermediate host snail populations: we hypothesized that offshore snails represent a population source of susceptible intermediate host snails that, over relatively short periods of time (days), can disperse to nearshore areas and contribute to schistosome transmission cycles. C) Location of 16 study villages (encompassing 32 distinct water access sites) in the Senegal River Basin; 6 villages are located on river settings, and 10 villages are located adjacent to Lac de Guiers. Map was created using ArcGIS Pro v2.8.1 (www.esri.com). D) Overhead drone imagery (left) and classified non-emergent vegetation (red), emergent vegetation (green), water (blue), and land (tan) superimposed by sampling bands used to compare model fit across sampling radii, for a village on a river setting.
Fig 2.
Results from offshore, deep water snail sampling in non-emergent vegetation for three villages located on river settings, and one village located on a lake setting.
A) Snails were sampled in three spatial polygons with increasing distances from water access site shorelines: within a discrete water access site (access areas surrounded by emergent vegetation walls, from shorelines to edge of open river channel or lake water), 0 to 75 m outside a water access site area, and 75 m or more outside a water access site area. "Satellite image courtesy of the DigitalGlobe Foundation (now Maxar Technologies, www.maxar.com)." B) Bulinus spp. snails were found at their highest density offshore in non-emergent vegetation outside water access sites, and at their lowest density nearshore inside water access sites. C) Biomphalaria pfeifferi snails were found in equal densities across the three spatial polygons sampled. For each snail species, estimated marginal means are shown with 95% confidence intervals, and p-values for pairwise comparisons were adjusted for multiple comparisons. *p < 0.05, **p < 0.01, ***p < 0.001.
Table 1.
Regression table for snail-habitat models (ZINB GLMM) of snail density (snail counts per sampling quadrat) in water access sites as a function of freshwater habitat type, water depth, and village location (lake vs. river), for both Bulinus spp. snails and Biomphalaria pfeifferi.
Snails of both species were most abundant in non-emergent vegetation. In all habitat types, snails were less abundant in village water access sites on lake settings than those on river settings. The number of quadrats sampled differs due to missing data for Bulinus spp. snail counts in two quadrats. Coefficients for differences in snail density according to habitat type are shown for non-emergent vegetation and open water with emergent vegetation as the reference category. Model coefficients are shows with standard errors in parentheses. Bolded coefficients represent statistically significant factors. *p < 0.05, **p < 0.01, ***p < 0.001.
Fig 3.
(A) Estimated odds ratios (left) and incident rate ratios (right) reflecting the association between individual infection and egg burden, respectively, for S. haematobium (black circles) and S. mansoni (grey triangles) and the total area of non-emergent vegetation measured within different sampling radii; each point estimate and 95% confidence interval were derived from an independent model considering non-emergent vegetation measured at the specified sampling radius (x-axis). Asterisks indicate sampling radii at which a statistical difference (*p < 0.05) between the two species was detected, as determined by an additional model that included an interaction term for species X non-emergent vegetation, independently for each sampling radius. (B) Visual interpretation of results: non-emergent vegetation within a sampling radius of 45 to 90 m was positively associated with S. haematobium egg burden as compared to S. mansoni, which was not positively associated with non-emergent vegetation at any sampling radius. (C) Predicted relationship between non-emergent vegetation and S. haematobium egg burden at a 90 m vegetation sampling radius (95% CI shaded in grey) (left); S. mansoni (right) was not associated with non-emergent vegetation.
Fig 4.
(A) Visual description of size and shape characteristics measured at each water access site; shown is a fully circumscribed water access site, where the man-made site is fully enclosed by emergent vegetation. (B) Prediction interval (solid line) and partial model residuals of a logistic GLMM testing for differences in infection probability for S. haematobium (left) versus S. mansoni (right), given the width of water access sites, using an interaction term for effect X schistosome species. (C) Prediction interval and partial model residuals for the effect of circumscription at villages on river settings, assessed using a three-way interaction term for circumscription X schistosome species X location (river versus lake). (D) Prediction interval and partial model residuals for the effect of circumscription at villages on lake settings. For all (B)-(D), pairwise contrasts were derived from a model fit using non-emergent vegetation data measured within a sampling radius of 90 m. All predictor variables were centered and scaled. Full model results are shown in S6 Table.