Summary Cobalamin (vitamin B12) is a cofactor for essential metabolic reactions in multiple eukaryotic taxa, including major primary producers such as algae, and yet only prokaryotes can produce it. Many bacteria can colonize the algal phycosphere, forming stable communities that gain preferential access to photosynthate and in return provide compounds such as B12. Extended coexistence can then drive gene loss, leading to greater algal–bacterial interdependence. In this study, we investigate how a recently evolved B12‐dependent strain of Chlamydomonas reinhardtii, metE7, forms a mutualism with certain bacteria, including the rhizobium Mesorhizobium loti and even a strain of the gut bacterium E. coli engineered to produce cobalamin. Although metE7 was supported by B12 producers, its growth in co‐culture was slower than the B12‐independent wild‐type, suggesting that high bacterial B12 provision may be necessary to favour B12 auxotrophs and their evolution. Moreover, we found that an E. coli strain that releases more B12 makes a better mutualistic partner, and although this trait may be more costly in isolation, greater B12 release provided an advantage in co‐cultures. We hypothesize that, given the right conditions, bacteria that release more B12 may be selected for, particularly if they form close interactions with B12‐dependent algae.

Diatoms are globally abundant algae that form extensive blooms in aquatic ecosystems. Certain bacteria behave antagonistically towards diatoms, killing or inhibiting their growth. Despite their crucial implications to diatom health and bloom control, insight of the prevalence and dynamics of antagonistic bacteria in nature is lacking. We report an ecosystem assessment of the diversity and seasonal patterns of bacterial antagonists of diatoms via regular plaque-assay sampling in the Western English Channel (WEC), where diatoms frequently bloom. Unexpectedly, peaks in antagonist detection did not occur during characteristic spring blooms, but coincided with a winter bloom of Coscinodiscus, suggesting bacterial pathogens likely influence distinct diatom host populations. We isolated multiple antagonists, spanning 4 classes and 10 bacterial orders. Many species had no prior reports of pathogenicity towards diatoms, and we verified diatom growth inhibitory effects of 8 isolates. In all cases tested, pathogenicity was activated by pre-exposure to diatom organic matter. Discovery of widespread cryptic antagonistic activity evident under specific conditions, indicates that bacterial pathogenicity towards diatoms is more prevalent than previously recognised. Finally, mining Tara Oceans data revealed the global biogeography of WEC antagonists and co-occurrence patterns with diatom hosts. Our study indicates that multiple, diverse antagonistic bacteria have potential to impact diatom growth and bloom dynamics in marine waters globally.