Genomic Metrics Applied to Rhizobiales (Hyphomicrobiales): Species Reclassification, Identification of Unauthentic Genomes and False Type Strains

Taxonomic decisions within the order Rhizobiales have relied heavily on the interpretations of highly conserved 16S rRNA sequences and DNA-DNA hybridizations (DDH). Currently, bacterial species are defined as including strains that present 95–96% of average nucleotide identity (ANI) and 70% of digital DDH (dDDH). Thus, ANI values from 520 genome sequences of type strains from species of Rhizobiales order were computed. From the resulting 270,400 comparisons, a ≥95% cut-off was used to extract high identity genome clusters through enumerating maximal cliques. Coupling this graph-based approach with dDDH from clusters of interest, it was found that: i) there are synonymy between Aminobacter lissarensis and Aminobacter carboxidus, Aurantimonas manganoxydans and Aurantimonas coralicida, “Bartonella mastomydis” and Bartonella elizabethae, Chelativorans oligotrophicus and Chelativorans multitrophicus, Rhizobium azibense and Rhizobium gallicum, Rhizobium fabae and Rhizobium pisi, and Rhodoplanes piscinae and Rhodoplanes serenus; ii) Chelatobacter heintzii is not a synonym of Aminobacter aminovoransiii) “Bartonella vinsonii” subsp. arupensis and “Bartonella vinsonii” subsp. berkhoffii represent members of different species; iv) the genome accessions GCF_003024615.1 (“Mesorhizobium loti LMG 6125T”), GCF_003024595.1 (“Mesorhizobium plurifarium LMG 11892T” ), GCF_003096615.1 (“Methylobacterium organophilum DSM 760T”), and GCF_000373025.1 (“Rhizobium gallicum R-602 spT”) are not from the genuine type strains used for the respective species descriptions; and v) “Xanthobacter autotrophicus” Py2 and “Aminobacter aminovorans” KCTC 2477T represent cases of misuse of the term “type strain”. Aminobacter heintzii comb. nov. and the reclassification of Aminobacter ciceronei as Aminobacter heintzii is also proposed. To facilitate the downstream analysis of large ANI matrices, we introduce here ProKlust (“Prokaryotic Clusters”), an R package that uses a graph-based approach to obtain, filter, and visualize clusters on identity/similarity matrices, with settable cut-off points and the possibility of multiple matrices entries.