Fork me on GitHub

Do bacterial species exist?

Any of my friends or colleagues who have had the "pleasure" of talking about science with me for more than a few hours know that I am prepared, at the drop of a hat, to rant extensively about several standing debates in biology which I consider merely semantic. For instance:

Q: Are viruses alive?

A: Who CARES!? Viruses do what they do. Cellular organisms do something else. What difference does it make if we decide to allow our middle-schoolers to draw little dotted lines around animals, bacteria, and viruses? And I don't even want to hear the word "prion".

I have a similar level of disdain for people who try to decide on a single definition for "species". Ultimately I am a pluralist: the definition should be tailored to the scientific question. Paleontologists, you have your morphological species concept, because what else do stony fossils allow? Are you studying speciation, you can keep the biological species concept, just be prepared for the inevitable craziness when fairly divergent plants hybridize or genes are transferred horizontally. Macro-evolution folks, don't give up on the phylogenetic species concept.

As an aside, I have no tolerance for arbitrary definitions which do not admit to their arbitrariness. Since I have yet to be convinced otherwise, this has precluded my acceptance of higher taxonomic divisions. If someone can give me a non-arbitrary definition for genus, order, family, phylum, etc. I would love to know what it is. As far as I am concerned they're just a naming system for monophyletic groups which pander to dead systematists.

Most species concepts are not applicable or arbitrary

For us microbiologists, several species concepts are in use. The genetic species concept groups organisms whose full genomes or selected genes have diverged minimally. This definition is applied in operational taxonomic unit (OTU) based analysis, and indirectly in whole genome DNA-DNA hybridization, the traditional determination of bacterial species.

OTU thresholds are certainly arbitrary; The definition fully acknowledges that fact. But the OTU concept is particularly useful for culture independent methods, since genetic distance is perhaps the only measure of familial resemblance available. Just as the biological species concept doesn't do much good for a paleontologist who lacks live, mating specimens, a morphological species concept is helpless when faced with sequence divorced from physiology.

When in possession of pure cultures, however, and therefore the luxury of phenotypic data, a physiological (which I'm using synonymously with "morphological") species concept is viable. Unfortunately, application of the morphogical definition depends critically on the exhaustivity of our search for physiological differences and an arbitrary delineation between groups.

The phylogenetic species concept is equally impotent. I challenge the reader to devise a monophyletic "basic unit" for any agamogenic organism which does not classify every daughter as a new species.

Despite the lack of canonical sex in microbes, an unsurprising (if not justifiable) urge to model ourselves after animal ecologists has driven some microbiologists to consider a biological species concept as well. With many microbes partaking in some form of recombination through transduction, transformation, or conjugation, it seems feasible to group those individuals which are capable of exchanging genetic material into a single species. Unfortunately this definition leaves behind all those clades which are not known to horizontally transfer genes. What's more, it fails the "arbitrariness test": how much recombination is required before two individuals share a species?

The ecological species concept saves the day

So if everything else is ultimately arbitrary, what's left? Originally formulated by Leigh Van Valen (1976), the ecological species concept has been distilled down to (Martinez-Gordillo, 2010):

A species is a set of [related] organisms exploiting (or adapted to) a single niche.

Now that's interesting! The ecological concept flips the competitive exclusion principle upside down. This definition sets the threshold for familial resemblance at the point where two lineages are no longer in the same niche. While there are still some rough edges, namely questions about what constitutes a niche, this definition appears to pass the test of arbitrariness and makes a satisfying connection between taxonomy and ecology.

Despite their inherent differences in criteria, and in my opinion defensibility, all of these species concepts (genetic, morphological, biological, and ecological) are getting at the same core idea: what makes one organism more "like" another? What keeps the diversity of life clumped together into identifiable units? There is a certain harmony to these species concepts. Whether recombination or ecological processes, we expect the end result to be genetic similarity which will manifest itself as physiological clustering.

Are species empirically real?

But wait. This is certainly begging the question. Does the microbial world actually have this assumed genetic and physiological clustering? Finally, a hypothesis to be tested!

Hanage et al. have (2006) summarized the current literature on genomic clustering and, in a second paper (2009), the models which may explain the observations. The data is somewhat ambiguous. Figure 1 from the 2009 paper demonstrates several of the problems facing an empirically described microbial species.

Figure 1:

Two examples exhibiting conflicts with species concepts in microbes.

First, and most importantly, 1A shows that genetic distance between and within clusters of Streptococcus isolates is not consistent. Current species designations are indicated. Clearly, the level of genetic distance within any visually (indicated by terminal node color) or a priori designated species varies by a factor of 5 or more. What's more, 1B demonstrates that while ecological preference (terminal node color) is fairly correlated with phylogeny in Vibrio, some clades lack this association, specifically V. splendidus.

The fractal-like clustering in the Streptococcus phylogeny makes me wonder if we're not seeing the result of numerous intense bottlenecks or selective sweeps which occur stochastically. And, while the ecological-phylogenetic mapping is fairly consistent, a point in favor of the species concept, I am curious if stabilizing selection and/or recombination models explain this clustering better than other, less interesting processes (e.g. gradual niche migration). I'm also concerned about problems with sampling. If there were genotypes which bridged the clusters would we know it? And, like every database dependent analysis in microbiology, a bias for pathogens could be misleading us here as well.


Despite a exhaustive debate over the semantics of asexual species, surprisingly little data has has been collected to show that bacteria even exhibit more than trivial clustering. Several mechanisms have been proposed for maintaining genetic conformity, namely recombination and niche specificity. Some of this research has aimed to find parallels with plant and animal ecology, but I would suggest we lose little by discarding the species concept entirely.

I can understand the need to define morpho-species when diagnosing disease, or the utility of OTU based approaches in molecular surveys. I'm also not suggesting that we should stop categorizing strains into species; the convenience for communicating microbiology is undeniable. But by remembering that most species divisions are arbitrary, and that every microbe is a new lineage, we're one step closer to escaping the chains of the old microbiology.