See also
Microbial taxonomy
Phylogenetic
trees for prokaryotes
The goal of taxonomy has been described as classifying living organisms into a hierarchy of categories (taxa) that are useful (based on biologically informative traits) and monophyletic (consistent with the true phylogenetic tree) (Hennig, 1965), though this point of view is not universally accepted (Benton, 2000). (See Edgar 2018, Taxonomy annotation and guide tree errors in 16S rRNA databases for more details and references).
With microbial markers such as 16S rRNA, most organisms are known only from environmental sequencing, and in these cases predictions of taxonomy must necessarily be made from sequence evidence alone.
One
approach to predicting taxonomy for environmental sequences and to defining
higher groups for type strains is to infer a tree from 16S rRNA sequences,
on the assumption that the tree will be a reasonable approximation to the
phylogenetic tree. However, in practice this approach is problematic because
of pervasive branching order errors in
estimated trees. This problem has been disregarded or underestimated in
the recent literature.
Phylogenetic approaches are used by Greengenes and SILVA for predicting taxonomy of environmental sequences. Bergey's Manual defines many higher groups based on 16S rRNA similarity, often by construction of a tree. For example, sub-groups of Firmicutes were constructed by
"...maximum-likelihood analyses of a dataset comprising about 5000 almost full-length high-quality 16S rRNA sequences from representatives of the Firmicutes and another 1000 representing the major lines of decent of the three domains Bacteria, Archaea, and Eucarya. The topology was evaluated by distance matrix and maximum-parsimony analyses of the dataset". (Bergey's Manual, 2nd ed., vol. 3).
The curators of Greengenes, LTP and SILVA have stated that they consider consistency to be important (emphasis added):
"Greengenes is a dedicated full-length 16S rRNA gene database that provides users with a curated taxonomy based on de novo tree inference" (McDonald et al., 2012).
"[LTP is based on] comparative sequence analysis of SSU rRNA [which] has been established as the gold standard for reconstructing phylogenetic relationships among prokaryotes for classification purposes" (Yarza et al., 2008).
"SILVA predominantly uses phylogenetic classification based on an SSU guide tree … discrepancies are resolved with the overall aim of making classification consistent with phylogeny" (Yilmaz et al., 2014).
By contrast, taxonomy annotations for environmental
sequences in RDP are predictions made by an algorithm, the
Naive Bayesian Classifier, which does not use a
tree or otherwise explicitly consider phylogeny.