Identification of branching orders among the main phyla within the domains is a way to perceive the course of evolution in each of the domains. It is also a way to structure the classification of the organisms that comprise the domains. There is no sanctioned taxonomy of bacterial phyla, and the ongoing flood of environmental sequences has overwhelmed the accounting (Fig. ). Woese's early surveys of rRNAs from seemingly diverse bacteria identified 12 main phyla, distinct relatedness groups of organisms by rRNA sequences (71
). The number of recognizable bacterial phyla continues to increase due to culture activities and, particularly, environmental rRNA gene surveys. Currently the public databases collectively identify >70 phyla of bacteria, defined as relatedness groups of sequences that have no reliable associations with other phyla in rRNA phylogenetic analyses (11
). Table contains a list of all the named (with cultured representation) and some of the “candidate” (not documented by culture) rRNA phyla in use in at least two of the public databases and which are documented by >100 SSU rRNA sequences. The accounting of bacterial phyla, even with currently available sequences, clearly is incomplete. Large numbers of sequences in the databases do not fall into the defined groups and, along with new sequences, will be the fodder for future expansion of our understanding of the bacterial tree.
Named and candidate rRNA phyla in use in at least two of the public databasesa
Only about half of the recognizable bacterial phyla have any cultured representation, and for most of the phyla, that representation is sparse, with only a few cultured examples. Most of the history of microbiology is based on representatives of the few phyla that happen to contain human pathogens and which tend to be readily cultured with classic methods (Table ). Even sequence representation in the databases is highly skewed and is sparse for most of the bacterial phyla. For instance, Fig. summarizes the distribution of sequences among the dozen most richly covered bacterial phyla; the other ~60 phylum level relatedness groups have comparatively little sequence representation. This skewing of database sequences to only a few phyla possibly reflects environmental abundance, but it seems more likely to be due to limited sampling of diverse environments. For instance, the phylum Bacteroidetes is highly represented in the sequence databases (Fig. ), but most of those sequences are derived from studies of animal feces and have limited diversity. On the other hand, the phylum Chloroflexi is represented by comparatively few sequences yet is conspicuous in many environmental settings, for instance, photosynthetic microbial mats worldwide.
FIG. 4. Distribution of SSU rRNA sequences among the top 12 bacterial phyla. Shown is the SSU rRNA sequence distribution in the SILVA 98 SSU Parc database (52) among the bacterial phyla (Ribosomal Database Project taxonomy) (10) containing the most rRNA sequences. (more ...)
Each of the bacterial phyla is itself a branching radiation from the base of the domain. Phylogenetic trees representing the different levels of bacterial diversity, with variable degrees of accuracy, can be downloaded from the public databases. The diversity and richness of the branches among and within the phyla are only beginning to be perceived because natural microbial diversity is so extremely undersampled. Representatives of some phyla, such as Proteobacteria (which includes, e.g., Escherichia spp., Rhodobacter spp., Rhizobium spp., Caulobacter spp., and Desulfovibrio spp.) or Firmicutes (which includes Bacillus spp., Clostridium spp., Staphylococcus spp., Lactobacillus spp., Heliobacterium spp., etc.), are cosmopolitan in the environment and the human experience and express a diversity of metabolisms, with phototrophic, heterotrophic, and autotrophic representatives. Other groups, while geographically distributed, seem more specialized. Representatives of the Aquificales, for instance, seem mainly to make a living by hydrogen oxidation, with oxygen or sulfate as the electron acceptor. Although this might seem to be a common physiology, representatives of this phylum so far have been detected only at high temperatures, such as in geothermal springs or hot oil wells. As another example, cyanobacteria seem to be restricted to a phototrophic state, so far as is known. As still other examples, sequences representative of the candidate phyla WS6 and OP11 are widely distributed geographically, but only in anoxic environments. This may indicate rather restricted metabolisms for the kinds of organisms that correspond to the sequences.
The structure of the base of the bacterial rRNA tree—the detail of any branching orders of the bacterial phyla—is not clear at this time. Indeed, the early development of the bacterial phyla may not have been treelike. Rather, it may have been a basal radiation, a “big bang” model for the origin of the bacterial phyla (53
). Figure portrays the base of the bacterial tree as a “polytomy,” a star radiation clouded by the uncertainties of any estimation at that depth in the bacterial tree. It is common in phylogenetic analyses that particular sequences, for instance, those of some representatives of the Aquificales
, seem to branch more deeply in the bacterial tree than those of other phyla. This led to the popular notion that such organisms are particularly “primitive” divergences compared to other bacteria. However, the phylogenetic results that indicate greater or lesser depth of branching of the main bacterial phyla are based on only a few SSU rRNA residues and are not seen in analyses using other genes (6
). At this time, I do not think there is convincing evidence for particularly deeply branching phyla in the bacterial tree.
Any specific relationships among the main bacterial phyla to comprise “superphyla” are unclear at this time, although some results indicate such associations. For instance, some rRNA comparisons conflate the verrucomicrobia, chlamydiae, and planctomycete phyla to indicate deep affiliation, although sequence representation for these little-known groups of organisms is limited (32
). Sequence representation in phylogenetic analyses can significantly alter associations of sequences at the base of the bacterial tree, so some phylum shuffling will be inevitable as the databases expand. In general, however, associations seen only in rRNA or other molecular sequence comparisons are subject to uncertainties, as discussed above, and need to be confirmed by other analyses, such as pangenomic comparisons.