Metagenomic strain tracking reveals patterns of bacterial spread and the impact of water chlorination

Bacterial infections are a major cause of morbidity and mortality among children under five in low- and middle-income countries (LMICs). Children in LMICs are exposed to and colonized by a range of pathogenic bacteria, yet patterns of bacterial exchange between humans are not well known, in part because culturing and sequencing single bacterial isolates is labor-intensive. Here, we apply a machine learning strain tracking approach to metagenomic data from 511 stool samples from children and mothers across urban and rural Kenyan communities to characterize bacterial dissemination and assess if community-wide water chlorination disrupts transmission. We identified distinct strain-sharing dynamics across species; potentially pathogenic taxa (e.g., Escherichia, Enterococcus, Campylobacter) exhibited distance-dependent dissemination driven by young children, while commensal taxa (e.g., Bifidobacterium, Bacteroides) showed patterns consistent with dietary exposure. Drinking water chlorination reduced community-level strain-sharing in rural communities. Our study provides the first strain-level insights into multi-species bacterial transmission dynamics in LMIC communities, identifying distinct dissemination pathways for facultative versus mostly anaerobic bacteria. Moreover, our findings highlight the utility of metagenomic strain tracking to uncover how community spread can be disrupted.