Proposal of names for 329 higher rank taxa defined in the Genome Taxonomy Database under two prokaryotic codes

Identification of names lacking published taxonomic descriptions

To identify GTDB provisional names that are missing published taxonomic descriptions, we first screened the widely used taxonomic and nomenclature databases—NCBI and LPSN. A set of taxonomic names that stably occurred between release R01-RS80 and R07-RS207 were matched to the list of names obtained from NCBI and LPSN using a custom python script. Names applied to the rank of species (thus, including new combinations) were excluded. Names of higher taxa that appear only in GTDB, were treated as GTDB provisional names and were subjected to additional manual checks to ensure that they have not been published elsewhere by performing searches in Google Scholar and Web of Science. Additionally, GTDB names that were present in LPSN but missing in NCBI were manually checked for publication since LPSN has started to list names that appear in the literature without author citation (i.e. with unknown authorship such as Zakham et al. (2019), as described above). A final set of GTDB provisional higher taxa names was categorized by rank and nomenclature status of the genera that were selected earlier as provisional types for naming taxa.

Selection of type genome sequences

To select taxonomic names that can be proposed for validation under the SeqCode, names with the status Candidatus were screened for the presence of high-quality genomes that could serve as type material for the type species of the selected genera. For this, metadata containing assembly statistics, completeness and contamination, and genomic properties were compiled from the GTDB repository of release R07-RS207, for each selected Candidatus species. High quality genome sequences (estimated completeness > 90% and contamination < 5%) were checked for additional criteria recommended for a genome sequence to serve as type under the SeqCode, e.g. presence of 16S rRNA genes (Hedlund et al. 2022). Genomes meeting these minimal quality standards were then examined for agreement between genome and 16S rRNA-based taxonomic assignments via phylogenetic analyses.

16S rRNA gene phylogeny of type genome sequences

16S rRNA gene sequences were identified and extracted from the selected type genomes using nhmmer v3.1b2 with the 16S rRNA model (RF00177) from the RFAM database. The dataset was supplemented with additional sequences obtained from closely related species. Sequences were aligned using ssu-align v0.1.1, and trimmed using trimal v1.4.rev15 followed by a custom trimming script to remove low parsimony columns. Phylogenetic trees were constructed using IQ-TREE v.2.2.0.3 with the following parameters: iqtree -s -m MFP -T 10 -alrt 1000. Trees were rooted with the closest outgroup and visualized in Dendroscope v3.8.8. A detailed workflow is provided at https://doi.org/10.5281/zenodo.7992507.

Concatenated protein phylogeny

Although taxa described in this manuscript were delineated earlier as described in Parks et al. (2018) and phylogenetic trees are provided in the GTDB repository (https://gtdb.ecogenomic.org/downloads), additional genome-based subtrees were inferred to support taxonomic descriptions. For this, bacterial genomes corresponding to GTDB representative genomes of named species (type strains or selected type genome sequences) of each proposed taxon were obtained from RefSeq/GenBank release 207. Genome trees were constructed from a multiple sequence alignment of concatenated protein markers as described in Parks et al. (2018) using FastTree 2.1.11 (Price et al. 2010) with the WAG model as implemented in GTDB-Tk v2.1.0 (Chaumeil et al. 2022). These trees were then used as starting trees to infer final trees with IQ-TREE v.2.1.2 (Nguyen et al. 2015) using the PMSF approximation. Trees inferred with IQ-TREE were executed with the following parameters: -m LG + C10 + F + G -ft < starting_tree> -alrt 1000. Trees were rooted by the closest outgroup, visualized in ARB v.6.0.6 and beautified in iTOL (Letunic and Bork 2019).

Comparison of taxonomic assignments of type genomes

To verify the taxonomic assignment of type genome sequences, 16S rRNA gene phylogenies were compared to the concatenated protein phylogenies (inferred as described above) of the same taxa using the Tanglegram algorithm in Dendroscope v3.8.8, then annotated using Adobe Illustrator.

Nomenclatural consequences of the GTDB taxonomic framework

After searching the literature for effective publications of provisional GTDB higher taxon names, 284 bacterial and 45 archaeal taxa still lack formal nomenclatural proposals. Of these, 223 (68%) have the potential to be validated according to the ICNP because they are underpinned by type strains that are serving as type material for species of genera proposed here as nomenclature types of higher taxa (Oren et al. 2023; Table 1; Figure S1, Supporting Information). The remaining 106 (32%) provisional GTDB names are based on Candidatus taxa that lack cultured representatives and therefore are not covered by the rules of the ICNP (Murray and Schleifer 1994, Murray and Stackebrandt 1995). However, there is the potential to propose at least some of these names under the newly developed Code of Nomenclature of Prokaryotes described from Sequence Data, or SeqCode for short. SeqCode is an alternative code for prokaryotic taxa based on high quality sequence data rather than axenic cultures as type material providing the ability to validly name uncultured taxa (Hedlund et al. 2022, Whitman et al. 2022; https://seqco.de). Here, we designate type genome sequences, which satisfy the SeqCode data quality recommendations, for 23 published Candidatus species names to validate both genus and species names under the SeqCode, at which time the Candidatus prefix can be removed (Whitman et al. 2022; Table 2; Table S1 and Figure S1 and S19, Supporting Information). These provide the basis for 49 of the 106 (46%) provisional GTDB names based on Candidatus taxa (Table 3; Figure S1, Supporting Information). An additional 57 higher taxon names based on Candidatus species that do not satisfy the typification and other validation criteria of either code (e.g. no axenic cultures deposited in two culture collections in different countries, low quality genome sequence) are described as new Candidatus taxa (Table 4; Figure S1, Supporting Information). Note that we have chosen to propose higher taxon names under the ICNP where possible resulting in a much higher number of names proposed under this code than the SeqCode (223 vs. 49). This is because the SeqCode recognizes names validly published under the ICNP (Whitman et al. 2022), but not vice versa, thereby minimizing the possibility of future synonyms if SeqCode names are subsequently proposed under the ICNP.

The majority of new higher taxon names proposed in this study (excluding Candidatus taxa) are at the rank of family (54%) with only five new phylum names required (2%), reflecting the majority of GTDB reclassifications occurring at lower ranks (Parks et al. 2018). Collectively, these taxa belong to 26 of the 166 prokaryotic phyla recognized in GTDB release R07-RS207 with the majority belonging to the bacterial phyla Bacillota/‘Firmicutes’ (45%) and Pseudomonadota/‘Proteobacteria’ (23%). Names validly published under the International Code of Nomenclature for algae, fungi and plants (ICN) are now recognized by the ICNP (Oren et al. 2021a) and, therefore, we propose 10 new cyanobacterial higher taxa based on genera originally validly published under the ICN (Table 1). Another recent revision of the ICNP is the inclusion of the rank of phylum under its rules (Oren et al. 2021b), allowing us to propose four phyla based on genera with validly published names under the ICNP (Table 1). A number of existing phylum names occurring in an effective publication have already been orthographically corrected in GTDB (Table S2, Supporting Information), making them compliant with the recent emendation of the code (Oren et al. 2021b).

Finally, 44 bacterial and 20 archaeal (previously published in Rinke et al. 2021) GTDB higher taxon names were left as provisional either because (i) they lack type genera and species, or (ii) the original authors of the type species are expected to independently propose the high rank name(s) associated with that species (Table S3, Supporting Information). These include taxa belonging to the Patescibacteria (CPR) that were originally proposed as Candidatus phyla and whose names we subsequently modified to ranks below phylum in early GTDB releases. In order to validate names of such higher taxa, a type species with a corresponding genus stem will need to be designated and deposited according to the criteria of one of the nomenclature codes. Alternatively, these names can be replaced with new names based on newly designated types with validly published names.

In conclusion, the proposal of 329 higher taxon names will ensure that future releases of GTDB will provide a stable taxonomy to the research community largely in alignment with nomenclatural codes. It also illustrates how the ICNP and SeqCode can be used in a complementary fashion.