Bacterial communities are intricate collections of bacterial species that each provide functions which contribute to the stability of the community. In a recent issue of the Proceedings of the National Academy of Sciences USA, Thomas and colleagues show that in the bacterial communities that form on the surface of the alga Ulva australis, the gene content of the community is more constant than the phylogenetic content, a finding that has important implications for our understanding of how bacterial communities are formed.

Theories to explain how bacterial communities form have been derived mostly from ecological studies of plants and animals. One prominent theory, the 'niche' theory, states that a community is selectively formed by specific species which are adapted to the environment that they occupy. An alternative theory states that a community is formed randomly, with limited apparent selection. In support of the second theory, the species composition of the bacterial communities associated with U. australis varied markedly between individual algae. However, the authors then determined that the functional composition of the bacterial communities present on six algae (identified through Cluster of Orthologous Groups (COG) and SEED annotations) shared Bray–Curtis similarities (a measure that quantifies the similarity between two samples) of at least 70%, whereas the similarity in species composition was below 15%. Functions that were conserved between the bacterial communities include chemotaxis and motility, attachment, biofilm formation and response to changes in the algal environment.

These results support a hybrid model of community formation, the 'lottery hypothesis' model, which was originally formulated for coral reef fishes. For these alga-dwelling bacteria, the presence of genes (and, hence, the corresponding protein functions) is selected by the niche, but the organisms in which these genes reside are much less important. Most ecological models of community assembly are based on species-level diversity, but this correlation does not seem to hold for these microbial communities, and future studies may need to consider adopting models that are specifically based on functional characteristics of microorganisms, rather than their phylogeny.