We mapped the double-stranded DNA (dsDNA) virus assemblage in groundwater below sub-Saharan urban poor settlements in Arusha (Tanzania), Dodowa (Ghana), and Kampala (Uganda). Our results indicated that ∼80% of dsDNA virus sequences matched the order of Caudovirales, i.e., indigenous bacteriophages; 1.8% of the dsDNA virus sequences matched those of viral pathogens that infect humans and larger animals, which we defined as so-called above-ground hosts. Within this group, the relative abundances of the genera Chordopoxvirinae, Alphaherpesvirinae, and Betairidovirinae were the highest. Culturable Escherichia coli bacteria were found even in deeper wells, indicating that all water was fecally contaminated. The community assemblages sampled in the cities were statistically significantly different from each other. Dissolved ions, population density, and sanitary status had no significant influence, but pH and latitude did. We concluded that the transport of dsDNA virus in groundwater was location specific but was not determined by input concentrations (i.e., related to population density) or related to groundwater chemistry. We hypothesize that other parameters, like the presence of macropores, cause these variations in these shallow, highly populated, heavily polluted terrestrial groundwater systems. Approximately 34% of Africa has similar hydrogeology, so this may affect many urban areas across the continent.

We combine satellite imagery, urban growth modelling, groundwater modelling and hydrogeological field expeditions to estimate the potential impacts in 2050 of rapid urbanization and climate change on groundwater in Arusha, Tanzania, and by extension similar areas in Sub-Saharan Africa. Our analysis suggests that a reduction of groundwater recharge by 30–44% will cause groundwater levels to drop by up to 75 m, mainly due to increased evapotranspiration and to an expansion in paved surface. If this scenario becomes reality, we predict that wells will run dry, creating health, social and environmental risks.