Evolutionary biology

Oceans of bacteria

Cloning microbial genes from natural environments has revealed a surprising amount of diversity. In understanding how microorganisms function in ecosystems, how much of this diversity really matters?

On page 551 of this issue, Acinas and co-workers1 provide fresh fuel for the debate about what constitutes a bacterial species. They have analysed microbial diversity in nature using the polymerase chain reaction to clone and sequence a certain gene, the 16S ribosomal RNA gene, a widely used way of detecting organisms that defy cultivation in the laboratory2,3.

Their subject, sea water, is perhaps the most intensively studied microbial habitat on the planet. What is new in this study is the use of techniques to reduce sequence artefacts, and the emphasis on the fine structure of evolution. When Acinas et al. plotted sequence similarity against the number of sequence comparisons, they observed a sharp discontinuity. In other words, when they reduced the error in their sequences, the resulting evolutionary trees, like bonsai, showed distinct signs of pruning. They suggest that the bushy tips of their trees (‘microdiverse clusters’) may represent populations of cells that share similar ecological niches and adaptations, and therefore could be regarded as natural taxonomic units — species or ‘ecotypes’ of species.

At the root of this issue is the observation that copies of microbial genes, recovered by cloning from nature, are rarely identical, which can lead to high estimates of diversity. Earlier this year, Venter and colleagues4 published more than one billion base pairs of genomic sequence data from microorganisms inhabiting surface waters of the Sargasso Sea — equivalent to approximately 775 complete microbial genomes. The technique they used, shotgun sequencing, involves the assembly of sequence fragments into putative genomes. Frustrated by the difficulty of this procedure, Venter et al. argued that the high genomic DNA diversity they encountered was evidence that their samples were populated by at least 1,800 species. Their argument was based on extrapolation from a common rule-of-thumb, which classifies organisms that are more than 3% different in 16S rRNA sequence as different species. Sequences in the microdiverse clusters produced by Acinas et al. are more than 99% similar, thus falling within a species according to the 3% rule.

To take a particular example, Venter's team found that genes from one of the bacterial groups studied by Acinas et al., SAR11 (Pelagibacter), accounted for 380 of the 1,412 16S rRNA genes they recovered. But the largest SAR11 fragment they could piece together was relatively small (about 21,000 base pairs), and even this sequence was not often repeated in the clone library, suggesting very high genomic DNA diversity within this group. In contrast, Acinas et al. found that most SAR11 rRNA genes from their samples could be placed in four or five microdiverse clusters, implying relatively limited diversity.

It is unclear how these data can be reconciled. Part of the problem may lie with the 3% rule and similar guidelines that do not take into account that most sequence change is caused by the clock-like accumulation of neutral sequence variation — that is, variation that has neither a positive nor a negative selective effect. Because new species arise randomly over time, high neutral sequence diversity within an ancient species could cause it to be construed as many species, at least in theory.

The bacterial-species issue may seem esoteric, but it is now assuming prominence as gene sequences from nature are applied to understand the global role of bacteria in biogeochemical processes. The question is, does each small branch of a gene tree represent an organism that plays a unique role in nature, or do the bushy tips of the trees represent sets of organisms (species or ecotypes of species) that essentially play the same role? The second explanation is obviously attractive because it reduces the complexity of modelling the various processes.

For most eukaryotic organisms, including plants, animals, fungi and protists, a species is defined as an interbreeding population. But bacteria don't have sex. Instead they use ‘parasexual’ processes, which, although much less efficient, accomplish the same thing, and are far more likely to result in the acquisition of DNA from unrelated species. In the 1980s, Sonea and Panisset5 argued that the discoveries of molecular biology render the bacterial-species concept obsolete, and that bacteria are indeed a superorganism with a common gene pool. This idea received further support with the onset of the age of genomics, when it was revealed that most microorganisms are composites of genetic information from many sources6. So much DNA has been traded over the eons that almost every cell is a patchwork quilt.

But a big question remains: is the rate of this horizontal gene exchange in bacteria rapid enough to disrupt the emergence of distinct sets of physical characteristics (phenotypes) with superior competitive advantages7,8? Acinas and colleagues' results are not what Sonea and Panisset's theory predicts: the observed pattern of rRNA gene evolution is not at all random, and strongly suggests that selection acts to create populations of cells that share the same niche, and to all intents and purposes might be regarded as species.

This will not be much of a surprise to microbial systematists, who, in the tradition of Linnaeus, have continued to add to and refine the list of named bacterial species. These microbiologists have not acted out of blind adherence to tradition, but out of respect for a common observation. Regardless of the evolutionary processes at work, microbiologists know that bacterial isolates in culture can be grouped into clusters of strains that are recognizable by their phenotypic traits.

The patterns of evolution observed by Acinas et al. fit an evolutionary model advanced by Cohan9, which is based on simple evolutionary assumptions and population-genetic theory (Fig. 1). Cohan argued that evolutionary trees would reveal the history of microbial species by showing when valuable mutations arose that allowed descendants of that cell to dominate, essentially by growing faster, or by evading predation better, than competing cells. In this view, microbial cells in nature are highly adapted and constrained by selection, with any new and better cell rapidly proliferating and eliminating lesser competitors. It is this process, the emergence of new and better cells, that causes periodic selection, also known as selective sweeps, to prune the inner branches from trees.

Figure 1: Selective sweeps and bonsai trees.

a, The data of Acinas et al.1 support a model of microbial evolution in which selective sweeps — the spread of advantageous new genes — purge neutral genetic variation. b, The result is loss of diversity (red segments), and evolutionary trees that look like bonsai with bushy clusters of branches connected to their nearest neighbours by long internal segments. According to this view, the bushy tips of the trees represent the accumulation of neutral variation within microbial populations that occupy the same niche and have similar biogeochemical activity.

It seems paradoxical that an apparently successful and competitive set of organisms such as the SAR11 group should also exhibit high diversity in their genomic DNA: one one might expect that keen selection would truncate diversity. This is particularly the case for populations with large effective population sizes, which Kimura10 predicted would be less subject to the process known as genetic drift and therefore more able to ‘fix’ genetic variants that confer a small selective advantage. The solution to this paradox may be that very large populations that have not been through recent episodes of purifying selection can maintain large reservoirs of neutral genetic variation. If so, then at least within ecotypes of microbial plankton, one would expect to find a core set of genes conferring a relatively conserved phenotype.

Larger and more accurate 16S rRNA data sets than that used by Acinas et al. are already available from environmental sequencing studies4. Moreover, these data sets include flanking DNA sequences that may tell us a great deal about genome evolution. Such additional information provides extraordinarily detailed snapshots of evolution in action that are likely to reshape our understanding of the forces that control microbial diversity. The debate about the origin of microbial species shows no sign of ending, but it is certainly heating up.


  1. 1

    Acinas, S. G. et al. Nature 430, 551–554 (2004).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Rappe, M. S. & Giovannoni, S. J. Annu. Rev. Microbiol. 57, 369–394 (2003).

    CAS  Article  Google Scholar 

  3. 3

    Giovannoni, S. J., Britschgi, T. B., Moyer, C. L. & Field, K. G. Nature 345, 60–63 (1990).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Venter, J. C. et al. Science 304, 66–74 (2004).

    ADS  CAS  Article  Google Scholar 

  5. 5

    Sonea, S. & Panisset, M. A New Bacteriology (Jones & Bartlett, Boston, MA, 1983).

    Google Scholar 

  6. 6

    Nelson, K. E. et al. Nature 399, 323–329 (1999).

    ADS  CAS  Article  Google Scholar 

  7. 7

    Doolittle, W. F. Trends Cell Biol. 9, M5–M8 (1999).

    CAS  Article  Google Scholar 

  8. 8

    Daubin, V., Moran, N. A. & Ochman, H. Science 301, 829–832 (2003).

    ADS  CAS  Article  Google Scholar 

  9. 9

    Cohan, F. M. Syst. Biol. 50, 513–524 (2001).

    CAS  Article  Google Scholar 

  10. 10

    Kimura, M. Nature 217, 624–626 (1968).

    ADS  CAS  Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Giovannoni, S. Oceans of bacteria. Nature 430, 515–516 (2004). https://doi.org/10.1038/430515a

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.