Modeling the origin of species (in bacteria)

While people disagree about whether species exist (and not just in bacteria) it is plain that the space of all possible genomes is not evenly populated. In short, clusters of related bacteria associated with named species certainly do exist. These persist even in the face of extensive recombination, and using the sequences of housekeeping loci (such as those used in MLST) we can construct trees showing the fringes of such species clusters are ‘fuzzy’ as a result of some isolates having sequence typical of both species.

Trees constructed from concatenated MLST data of multiple Neisseria species are suspect in fine detail due to recombination, but the clusters are evident. And the recombination produces evident fuzzy fringes between N. meningitidis and N. lactamica. In contrast the green gonococcus cluster is quite distinct and discrete - likely reflecting limited opportunities for recombination given the different mucosal surface it typically infects.
Trees constructed from concatenated MLST data of multiple Neisseria species are suspect in fine detail due to recombination, but the clusters are evident. And the recombination produces evident fuzzy fringes between N. meningitidis and N. lactamica. In contrast the green gonococcus cluster is quite distinct and discrete – likely reflecting limited opportunities for recombination given the different mucosal surface it typically infects. The original figure can be found here.

How this arose and is maintained is not clear. In previous work with Christophe Fraser of Imperial College London, we were able to show the impact of recombination on this, and specifically that at high enough rates, recombination can prevent the budding of daughter clusters.

Results of a very simple model of species divergence, under conditions of a) no recombination, b) recombination as seen in some bacteria, c) recombination but with a realistic parameter reducing its efficiency with increasing divergence. Figure drawn from Hanage et al 2006, available at Pubmed Central.
Results of a very simple model of species divergence, under conditions of a) no recombination, b) recombination as seen in some bacteria, c) recombination but with a realistic parameter reducing its efficiency with increasing divergence. The impact of recombination is seen visually as the replacement of multiple black areas showing denser regions of sequence space, with a single one reflecting one diverse cluster. Figure drawn from Hanage et al 2006, available at Pubmed Central.

The recombination rates in question are quite similar to those observed in nature. We are presently extending this work to include the accessory genome – those loci that are not present in all strains. This work was done with Dr Pekka Marttinen (then a postdoc, now faculty at Aalto University in Finland) and Professor Jukka Corander.

We are also interested in the possibility that the spread of genetic information between strains and species reflects ecological opportunity, and may be reduced by a reduction in niche overlap. We have empirical evidence for such a scenario in Campylobacter jejuni (see population genomics). The possibility that this explains the existence of ‘satellite species’ (such as nontypeable pneumococci we recently described as distinct and divergent in core and accessory gene content) is under investigation.