Every gene is everywhere but the environment selects: Global geo-localization of gene sharing in environmental samples through network analysis

The spatial distribution of microbes on our planet is famously formulated in the Baas-Becking hypothesis as ‘everything is everywhere but the environment selects’. While this hypothesis does not strictly rule out patterns caused by geographical effects on ecology and historical founder effects, it does propose that the remarkable dispersal potential of microbes leads to distributions generally shaped by environmental factors rather than geographical distance. By constructing sequence similarity networks from uncultured environmental samples, we show that microbial gene pool distributions are not influenced nearly as much by geography as ecology, thus extending the Bass-Becking hypothesis from whole organisms to microbial genes. We find that gene pools are shaped by their broad ecological niche (such as sea water, fresh water, host, airborne). We find that freshwater habitats act as a gene exchange bridge between otherwise disconnected habitats. Finally, certain antibiotic resistance genes deviate from the general trend of habitat-specificity by exhibiting a high degree of cross-habitat mobility. The strong cross-habitat mobility of antibiotic resistance genes is cause for concern and provides a paradigmatic example of the rate by which genes colonize new habitats when new selective forces emerge.