It should not be overlooked that the scoring can be affected by a number of artifacts because of its dependence on the selleck screening library underlying phylogenetic tree and the annotation of its leaves. For instance, LTP versions have sometimes selected the wrong sequence as, e.g., in the case of the type strain of Weeksella virosa [46]. But compared to the overall number of strains (Figure 3) these problems appear to be rare. Moreover, to avoid picking the wrong organisms in the GEBA project the 16S rRNA gene of each strain is resequenced after gDNA extraction, and the strain is put back if the sequence does not match database sequences annotated as being obtained from the same strain. Using a phylogenetic tree of some organisms instead of their taxonomic classification avoids a number of potential artifacts in taxon selection.

Even though it has only slowly been appreciated by taxonomists after Darwin, the sole possible goal of a taxonomic classification is to summarize the genealogy of the organisms [3,4]. For this reason, a taxonomic classification always contains less information than the empirical estimate of the phylogeny from which it was derived. But frequently classifications cannot even pretend to summarize the respective underlying genealogies because the classifications include non-monophyletic groups [3,4,47,48]. Current microbial classification contains a number of such taxa (e.g., Bacillus [15], Desulfotomaculum [49], Planctomyces [43], Spirochaeta [16] and Xanthobacteraceae [50]). Some of the problematic parts of the classification are due to missing phylogenetic analyses in the original description (e.

g., [15]), often because suitable character data or inference methods were simply lacking at the time when the taxon was GSK-3 described (e.g., [16]). But in other cases, such problematic taxa have been created due to conceptual shortcomings. For instance, the genus Schlesneria was introduced in a study [51] in which a tree was depicted that clearly showed that the placement of the new taxon causes another genus, Planctomyces, to become paraphyletic [43] (see [52] for algorithmically straightforward, character-independent definitions of the terms ��monophyletic��, ��paraphyletic�� and ��polyphyletic��). Clearly, such discrepancies are not due to preferring phenotypic traits (used as ��diagnostic�� characters) over 16S rRNA gene results because diagnostic characters are not necessarily synapomorphies. But only synapomorphies (or phylogenetic trees, of course [52]) can justify monophyletic groups [3,53]. For instance, it is easy to outline the diagnostic characters of reptiles that separate them from either mammals or birds, but nevertheless reptiles are the classical example of a paraphyletic group [3].