cyanobacterial phylogeny
Hover the menu buttons above to navigate this page. Each opens a sub-menu; some entries of which
carry a right-pointing arrow that permits the menu to be opened to the final level.
Clicking any of the second or third level entries moves to that part of the page.
Click the magnifier (left) to open a search box, or the logo to return to top of page.

If you arrived directly here by a web search (Google, Bing etc.), you may go to the home page of this site for full information.

Cyanobacterial Phylogeny and Taxonomy

Note on web browsers:

For early versions of Firefox on Windows 10, the user should go to 'Tools' > 'Options' > 'Applications'; scroll to 'Portable Document Format (PDF)', set 'Preview in Firefox'. The entry 'PDF' can be left as 'Use Adobe Acrobat Reader DC'. On Linux, the equivalent path is 'Edit' > 'Preferences' > 'Applications', and the 'PDF' entry is absent. Note that if pdf pages can no longer be opened, an application or a manipulation may have changed the 'Applications' item; simply reset it as above. This is often required after update of Firefox to a new version. You can revert to the built-in viewer at any time. For later Firefox versions (currently 91.2.0esr) the above changes do not apply. For Internet Explorer on Windows 10, by default on opening this page you may see a message "Internet Explorer restricted this webpage from running scripts..."; click "Allow blocked content", since we use java scripts. This behaviour can be changed via Tools > Internet options > Advanced; enable "Allow active content to run...". For the top horizontal menu of the Main page, entries in the third level do not always react; this can be overcome by right-clicking the entry and selecting "Open" from the pop-up menu. With these suggestions taken into account, Avast, Brave, Chromium, Firefox, Google Chrome, Internet Explorer 11, MS edge and Opera all behave correctly on Windows 10. We have no access to a Mac, and have not tested any browser on that operating system; comments on usability would be appreciated.

contact (write to mherdman1245(at)

It is advisable to clear the browser cache before navigating further on this site; the cache is normally used to speed-up browsing, but in the event of us changing pages, the old version of the pages will be shown.

Many of the pages on this site, especially the trees and some of their associated text files, are in PDF; these may be opened, searched, printed and downloaded by all major browsers. The main page is best searched with our search function but the built-in "find" function of the browser can also be used; this can be quickly opened by pressing the F3 button or simultaneously Ctrl+f. This is required for Internet Explorer, which does not recognize our search function, and also all browsers other than Firefox for the PDF documents. Firefox, however, seems unable to mark some entries in the trees, and is not recommended.

For Linux systems we suggest the use of the Chromium browser.

Note on screen size and printing:

All pages on this site are visible on screens as small as 10 inches diagonal (we have not tested smaller screens) such as small netbooks or tablets. For comfort, however, we recommend at least a 15 inch screen. All tree and text documents may be printed on US Letter or A4 paper in landscape mode. Since the tree expanded to the level of strains requires 18 printed pages, the user may wish to take a screenshot only of an area of the tree; this requires the very simple installation of a screenshot add-on for the browser (FireShot and Nimbus, for example, are available for most). We use Nimbus on Windows 10 and Linux because the built-in image editor functions well and produces figures similar to this, where Nodosilinea strains have been highlighted with boxes and arrows. The user will need to enlarge (using the zoom function of the pdf reader) the view before taking the screenshot.

The roots of the tree

Although the idea of creating a tree of life was suggested by Darwin (On the Origin of Species, 1859), Haeckel (General Morphology of Organisms, 1866) was the first (to the best of our knowledge) to publish a comprehensive tree (see below) and to coin the term "phylogeny".

The tree of Haeckel
Zoom of the Archephyta
Nostoc macrocolonies

Haeckel included Nostoc in that tree, grouped in the Archephyta with eukaryotic algae such as Ulva (see the zoom), most probably because the gelatinous macrocolonies of some Nostoc spp. are easily visible in the environment.

Nostoc is probably the first cyanobacterium to have been described, the name being derived from the word "Nostoch" coined by the 15th century scientist Paracelsus (Aureolus Philippus Theophrastus Bombastus von Hohenheim) to describe these gelatinous colonies (see Potts, 1997, for further details).

Progressive improvements in the quality of microscopes (first developed as a prototype in 1702 by Antonie van Leeuwenhoek, who described "animalcules") led to the discovery and description of "blue-green algae" (Cyanophyta). These were classified among the algae, differing in colour from all others. They were named according to the limited number of characteristics then available: cell shape and size, unicellular or filamentous, with or without sheaths and/or differentiated cells, habitat. The prokaryotic features of these microorganisms were established in the 1960s, giving rise to the name Cyanobacteria. The earlier elucidation of the structure of DNA by James Watson & Francis Crick (1953), drawing on X-ray crystallography data of Rosalind Franklin, led to modern gene sequencing technology and an explosion in the number of 16S rRNA and whole-genome sequences, forming the basis of modern phylogenetic trees. Major inconsistencies between the classical and genetic approaches became evident, as we shall see on this site.

Cyanobacterial phylogeny: the current status

A perusal of the phylogenetic tree shows the apparent chaotic state of cyanobacterial nomenclature, where organisms with a variety of names often fall together into the same generic cluster. Many such organisms are not available from Culture Collections for verification, and thus rampant misnaming persists. Even if available, the strains are not normally maintained in the axenic state, thus preventing any kind of chemo-taxonomic analysis. Consequently, the identification of cyanobacteria is traditionally based only on morphological and environmental considerations. Morphology may be influenced by many factors, including medium composition, phase of growth (leading to nutrient depletion in older cultures), light intensity, L/D cycles. Of many studies of this phenomenon, we list only four: Dextro et al., 2018 showed effects of light intensity, N and P concentrations on the differentiation of heterocysts and akinetes of Nostoc paludosum; Shafik et al., 2003 documented marked variation in morphology of Cylindrospermopsis raciborskii (now Raphidiopsis) under N- or P-limitation; Sukenic et al., 2013 demonstrated the effects of P concentration on the development of akinetes in Aphanizomenon ovalisporum; Zapomělová et al., 2008 performed elegant cross-table experiments showing major effects of nutrient concentration on the morphology of a wide selection of heterocystous organisms. If not carefully controlled in cultures, the effects are often dramatic. An example is the marine "Pseudanabaena" strain MBIC10772 (transferred to NBRC as NBRC 102910), isolated from seawater, Pacific Ocean, which falls into the heterocystous clade as species 1.3.7H, together with Halotia strains from saline habitats. The apparent loss of heterocysts in this strain is undoubtedly due to prolonged maintenance in N-rich medium (the Daigo IMK medium employed in the NBRC contains 200 ppm nitrate).

A further problem arises when the organism is a constituent of a non-axenic and non-clonal culture, often being seen as the only organism present. This appears to be the case for "Scytonema millei" strain VB511283 for which 3 draft genome sequences are available: the first two versions (JTJC00000000.1 and JTJC00000000.2) were of bacterial origin; the third (QVFW00000000) falls with strains named as Chroococcidiopsis, Cyanosarcina and relatives outside of the heterocystous clade. This applies not only to the 16S rRNA gene extracted from the genome sequence, but also to the whole genome.

Environmental information is often, but not always, available. Even when given, the details are of limited value.

Sequence quality, especially length, are often poor, as described in the Methods section.


Two websites are particularly informative for taxonomic studies.

Information on both prokaryotic cyanobacteria and eukaryotic algae is available from AlgaeBase.

LPSN (List of Prokaryotic names with Standing in Nomenclature), listing all prokaryotic organisms including cyanobacteria, is particularly useful, permitting rapid searches for genera and species, and providing information on (for example) phylogenetic inference, available from the "Navigation" item in the side menu.


All data on this site may be freely downloaded. In addition, the ARB database and supplemental files may be downloaded from Google Drive at this link without password.


A. 16S rRNA trees

A1. Sequences

16S rRNA and genomic sequences were obtained from the National Center for Biotechnology Information (NCBI), Joint Genome Institute (JGI) or publications on Dryad. Where the 16S rRNA genes were available within genomic sequences, they were extracted with the use of our SSU-finder script. The tree also contains over 200 unpublished 16S rRNA sequences of the Pasteur Culture Collection of Cyanobacteria (PCC), kindly provided by M. Gugger. Data collection ceased on 15.01.2017, except for new genera and genomes. There are currently over 50,000 cyanobacterial 16S rRNA sequences available from NCBI, increasing every year, as shown in the following figure:


A2. Sequence alignment

16S rRNA sequences were maintained in ARB (Ludwig et al., 2004) and aligned on the basis of secondary structure. If you are familiar with ARB, you will note that the sequence alignment shown below is much more compact than standard versions; we have replaced the helix and helix numbering scheme by those developed from the secondary structure model of the 16S rRNA molecule of Synechococcus elongatus PCC 6301. Alignments based on secondary structure are more accurate than those generated with e.g. ClustalX; the figures below show the ARB alignment (where subscripts "~" indicate perfect pairing on the two sides of a helix, "#" indicates a mismatch) of helix 6, where certain strains have an insert, and the ClustalX alignment of the same region for the same sequences.

helix 6

ARB is only available for Linux OS and OSX. If you are using MS Windows, you can install the virtualization software VirtualBox from and use it to run CyanoPhy, our Linux OS dedicated to phylogenetic studies; the latter may be downloaded (for 64-bit architectures only) here. Installation instructions are provided in the file "Install.doc" on the same site. CyanoPhy contains a version of ARB compiled for the OS to allow full functionality, together with databases of 16S rRNA and 23S rRNA sequences. The latter are old versions and should be replaced by the current versions, which may be downloaded as explained in the section "Data Availability".

A3. Sequence length

Note that many recent sequences, defining new genera and/or species, although complete at the 3' end, start at (or close to) only Escherichia coli position 359. The missing 5' region is essential for accurate determination of sequence identity and reconstruction of phylogeny, which should be performed only with positions common to all sequences in the alignment (in this case E. coli position 359 to the 3' end of the shortest sequence) and not on the full-length alignment (leaving the tree-building software to deal with the missing data). The missing region is also essential for the detection of chimeric sequences. The situation is even worse for the sequence of Spelaeonaias floccida (KY018918) of length 304 nt, starting at E. coli position 785. The latter implies phylogeny reconstruction on a very short sequence alignment. We plead for provision of at least one full-length (or as near as possible) sequence for each species described.

Substitution rate variability along the alignment, based on 721 non-identical sequences with 20 bases excluded at both ends, is shown in the figure below. The variable regions are the peaks with high relative rate. Both conserved and variable sites are required for accurate tree building. The vertical red line indicates E. coli position 359. At least 3 highly variable and 3 conserved regions are seen 5' to this point; these are missing in almost all recent sequences.


These short sequences impose the requirement of omitting a large part of the 16S rRNA sequence when building phylogenetic trees, and trees produced in this way are most likely inaccurate, as shown here. The dramatic change in topology of the tree inferred from truncated sequences is a consequence of the different phylogenetic distances calculated from full-length and shorter sequences; the figure below illustrates (using the almost full-length sequences employed for inference of the backbone tree described below) these distances (top panel) and the different number of nucleotide changes (bottom panel) in sequence sets which were used as full-length or trimmed to omit E. coli positions 1-358.


Such effects of 16S rRNA sequence length on phylogenetic distance have impacts on the grouping of strains even at the specific level. This is widespread throughout the tree, and we give only five examples for well-characterized species. Gaget et al. (2015) applied MLST with 16S rRNA, ITS and three housekeeping genes (gyrB, rpoC1 and rpoB) to a large number of strains of the genus Planktothrix, finding 7 strain clusters clearly separated at the specific level. We confirm this with analysis of almost full-length 16S rRNA sequences (covering E. coli positions 29-1542) and, where genomic sequences are available, their position in the genome tree. However, using the same 16S rRNA sequences truncated to omit E. coli positions 1-359, P. agardhii strain PCC 9239 exhibits a genetic similarity of 98.97% with P. paucivesiculata strain PCC 9018, artifactually combining the two species. The sequences of the two strains differ by 22 bases, of which 10 are located 5′ to E. coli position 359. Similarly, the complete 16S rRNA sequences extracted from the sequenced chromosomes of strains of the two species of Gloeobacter, G. violaceus PCC 7421 and G. kilaueensis strain JS1, differ in 20 positions along their entire (1485 nt) length, of which 8 occur in the E. coli 1-359 region. Since the uncorrected p-distance (D), is calculated as D=base difference/sequence length, we find D=0.013468 (equivalent to a similarity [S] of 98.65%) with full-length sequences, thus confirming the separation of the strains into two species. However, if the leading region is omitted (leaving 1159 bases), D=0.01035 (S=98.96%) and the strains would be considered as members of a single species. In contrast, the 16S rRNA sequences of two strains of Limnolyngbya circumcreta, CHAB5649 and CHAB4449, show a total of 20 base changes, only 2 occurring in the leading region; calculation of D with full-length (corresponding to E. coli positions 8-1542) sequences gives a value of 98.72%, which lies at the lower end of our species cutoff value, whereas calculation with truncated sequences yields a decreased value of 98.54%, suggesting that the two strains fit into two different species. This obviously affects the type species of the genus. A further example of the effect of 16S rRNA sequence length on distance estimates involves the genus Koinonema. The four sequences, all nearly full length, deposited by Buch et al. (2017) as representatives of the type species K. pervagatum, fall into two different species of genus in our tree. The three members of species share 98.99-99.12% sequence identity estimated from the full-length sequences, with the reference strain (63PC) in species showing 98.45-98.72% identity to the first species. Using truncated sequences, however, strain 63PC appears as a member of the same species as the others, showing increased D values of 98.97-99.31%. Finally, of the 10 species clusters containing strains identified as Nostoc commune (see sections on Type species and on Polyphyly), almost full-length (E. coli positions 26-1522) sequences are available for only 8: 1.8.1E, I, AE, L, N, O, Q and T. Truncating the same sequences to exclude the region 5′ to E. coli position 359 causes changes of D values ranging from -0.164 to +1.273, and results in the inability to discriminate species 1.1.1O, Q and T, leaving only 5 species, as shown in the figure below.

Nostoc commune

A4. Tree inference

Despite the above comments, we wished to retain the short sequences in order to include the new organisms. To overcome this limitation, a tree of 1500 sequences (almost full-length, lacking at most 50 nt at either end, with known and potential chimeric sequences excluded), was produced using FastTree (Price et al., 2009) under the GTR model of nucleotide evolution and the CAT approximation with 20 gamma classes. Support values were computed by FastTree with the SH test. The treefile was re-imported into the ARB database and shorter sequences were added as required with the ARB "quickadd by parsimony" method, which does not change the topology of the initial tree but may cause rotation of existing clusters. Click here to see an example of this method. Other treeing programmes such as PHYML or RAxML are not practical alternatives for the large dataset employed here. The final figures were produced by exporting the trees from ARB to XFig, re-exporting with multiple overlapping pages in postscript format, then converting to pdf with the ps2pdf utility of XFig. For convenience, we also provide a simple textual list of the strain tree, available here.

A5. Cutoff values employed for genus/species discrimination

Genera and species are discriminated in the tree on a phylogenetic basis. Organisms were divided into different species if their 16S rRNA sequence identity was lower than 98.7-99.0%, or into separate genera when they showed less than 96.5-96.9% 16S rRNA sequence identity. The identities were determined as uncorrected p-distances measured on comparable regions only, using, wherever possible, complete (or almost complete) sequences. The tree therefore provides a strictly phylogenetic classification. We derived these cutoff values from a thorough study of the literature and our phylogenetic trees. Where complete chromosomes are available, the 16S rRNA identity values were confirmed with measurements of in silico DNA-DNA hybridization (DDH) and average nucleotide identity (ANI). In addition to these traditional measures, we examined values produced by an alignment-free method employing Mash (Ondov et al., 2016) and also our extended marker set of 79 proteins (hereafter abbreviated as CGS, Core Gene Similarity, not to be confused with core genome). The latter is independent of 16S rRNA identities, because the 16S rRNA gene does not encode a protein product and is therefore excluded from the marker set. The methods employed are described in more detail in section B1 (for CGS only) or section B5 of this page for the other methods (will open in the same browser tab, use the back button of your browser to return here) and further information and graphics of all the above methods are given in the genome metrics document.

We repeatedly read statements such as "Stackebrandt & Goebel (1994) suggested that in prokaryotic taxa, those strains with less than 97.5% 16S rRNA sequence similarity should be considered to be separate species, while those with less than 95% similarity should likely be considered to be separate genera" and that "the value of 97.5% was subsequently (Stackebrandt & Ebers 2006) increased to 98.7-99%". Stackebrandt & Goebel (1994) in fact stated that "the phylogenetic definition of a species would generally include strains with approximately 70% or greater DNA-DNA relatedness and with 5°C or less ΔTm" and "Phenotypic characteristics should agree with this definition". They also wrote (reporting on the Ad Hoc Committee on Reconciliation of Approaches to Bacterial Systematics) "organisms which have 70% or greater DNA similarity will also have at least 96% DNA sequence identity". Finally "What the threshold value of 70% does not take into account is the possibility that the tempo and mode of changes differ in different prokaryotic strains" and "only complete sequences allow reliable phylogenetic comparison". Stackebrandt & Ebers stated "we now recommend a 16S rRNA gene sequence similarity threshold range of 98.7–99 % as the point at which DNA–DNA reassociation experiments should be mandatory for testing the genomic uniqueness of a novel isolate(s)." They recognized that the suggestion of 70% DDH was an arbitrary value. No value for genus separation is given in either paper, despite repeated statements to the contrary. Many papers have addressed this problem, summarized best in Kim et al. (2014), who suggested 98.65% 16S rRNA sequence similarity for delineation at the specific level; the figure in that paper shows that a value of 97% would be appropriate for genus separation. Yarza et al. (2014) suggested identities of 94.5% for genus separation and 98-99% for species separation. Our analysis gives a much higher value (96.5-96.9%) required for separation of genera, but agrees at the specific level. Indeed, a genus level cutoff value of 94.5% would unite many pairs of cyanobacterial genera into one, for example Anabaena of genus 1.2.10 with Nodularia, Synechococcus of genus 7.1.1 with Prochlorococcus. Yarza et al. (2014) wisely added "the 94.5% threshold for genera does not preclude the formation of genera that have sequence identities of 96% if it is supported by other phenotypic, genetic or environmental data".

In summary, the consensus value for specific delineation is 98-99% 16S rRNA sequence similarity, but a precise value for separation at the genus level has not been established. Almost all papers agree that only nearly complete 16S rRNA sequences give accurate measures of taxonomic diversity and that sequence quality (notably the error rate), alignment quality and misnaming of strains are major concerns. In addition, we add that all values have been obtained from comparisons of numerous bacterial phyla; as early as 1994, Stackebrandt & Goebel stated that "What the threshold value of 70% does not take into account is the possibility that the tempo and mode of changes differ in different prokaryotic strains".

Sadly, these criteria are rarely met with the available cyanobacterial 16S rRNA sequences, as discussed in the Methods section, and sequence length is often short. In addition, it seems that many authors have not thoroughly read the available literature. We cite below some of many examples illustrating this point.

León-Tejera et al., (2016): "similarity values should not be used as a primary criterion to define taxon levels in the Cyanobacteria".

González-Resendiz et al. (2018): "The thresholds set as clear evidence for recognition of new species or genera (Stackebrandt & Goebel 1994; Stackebrandt & Ebers 2006; Kim et al. 2014; Yarza et al. 2014), are not met in the case of Nunduva fasciculata. Indeed, all 16S rRNA sequences of members of Rivulariaceae sensu stricto are above 94.5% similar, suggesting that this very morphologically and ecologically diverse group could all belong to the same genus if only this criterion was used....We conclude that the 16S rRNA gene sequences were too similar to serve as a reliable or sole criterion for genus or species recognition in the Rivulariaceae".

Scotta Hentschke et al. (2016): "the "95% similarity between strains rule" to aggregate genera cannot be strictly applied for Nostocales".

Bohunická et al. (2015): "All Roholtiella species were above 97.9% similar, meaning they were above the similarity cut-off which is considered clear evidence of speciation (97.5%). The two representative sister taxa were 96.9–98.8% similar to Roholtiella species, which is above the similarity cut-off which is considered clear evidence for recognition of genera (95%)" and "the artificial limits set for species and genera (97.5% and 95% rRNA sequence similarity, respectively) proposed for prokaryotic taxa by Stackebrandt and Goebel (1994) are too low".

Kaštovský et al. (2014): "Cyanocohniella has 16S rRNA similarity above 95% for 29 different genera in the Nostocales, including taxa from five different families. If the 97.5% similarity cutoff were used to group taxa with Cyanocohniella, it would be the same species as .....".

Johansen et al. (2014): "...typical cut-offs for taxon recognition (<95% for genera, <97.5% for species; Stackebrandt and Goebel 1994, Ludwig et al. 1998) are much too conservative for recognizing taxonomic diversity in this clade.....We conclude that 16S rRNA gene similarity fails as a criterion for recognizing taxonomic diversity in the Nostocaceae".

The term "gold standard" for taxonomic purposes appears to have been first coined by Stackebrandt & Goebel (1994) as the rational for using DNA reassociation as the standard for species delineation, and has subsequently passed through various stages (e.g. being "tarnished": Stackebrandt & Ebers, 2006), before being completely transmuted to tin by Pietrasiak et al. (2019) in describing Myxacorys: "The prokaryotic microbiology "gold standard" based on DNA–DNA re-association levels and ΔTm (Wayne et al. 1987, Stackebrandt et al. 2002) should more aptly be called the "tin standard", and we may be undervaluing tin in this comparison". We criticize this statement on three grounds: (1) it implies that they were using genomic approaches, which is not the case; (2) if they were trying to be derogatory, they surely meant to write "overvaluing"; (3) neither Wayne et al. 1987 nor Stackebrandt et al. 2002 mention the word "gold". In the same paper, we read: "This study represents the first time cyanobacterial species can be diagnosed primarily by 16S rRNA gene sequence data".

B. Genome trees

B1. General

There are an increasing number of studies on the phylogenetic aspects of cyanobacterial genomes; some are mentioned below, others are described in the document correlating DDH, ANI, Mash, CGS and 16S rRNA similarity values. Criscuolo & Gribaldo (2011) created a phylogeny based on 61 genomic sequences and 191 concatenated protein markers derived mostly from Nostoc sp. PCC 7120, accepting the fact that some markers were absent from many genomes. This study, concerned mostly with the origin of chloroplasts within the cyanobacterial radiation, gave little useful taxonomic information, since genomes of organisms of many clades were not available at that time. Shih et al. (2013) expanded the coverage of the phylum by sequencing the genomes of 54 strains of different morphotypes and properties, recovering 29 complete and 25 draft sequences to give a total of 126. They employed a similar method and the same concatenated sequences of 31 conserved proteins (DnaG, Frr, InfC, NusA, Pgk, PyrG, RplA, RplB, RplC, RplD, RplE, RplF, RplK, RplL, RplM, RplN, RplP, RplS, RplT, RpmA, RpoB, RpsB, RpsC, RpsE, RpsI, RpsJ, RpsK, RpsM, RpsS, SmpB and Tsf) as Wu & Eisen (2008) in their study of the complete chromosomes of 578 bacterial species by blastp studies, in the absence of proof that these conserved proteins are all suitable for cyanobacterial studies. Calteau et al. (2014) used 29 unspecified conserved proteins selected from those of Wu & Eisen (2008) in blastp studies with 126 cyanobacterial genomes. Although a wealth of information was obtained by Shih et al. and Calteau et al., the taxonomic study was limited. Dvořák et al. (2014), using blastp studies of 70 genomes with 192 unspecified conserved protein sequences, constructed an elegant phylogeny with information on the time scale of evolution and HGT events, but phylogenetic analysis was restricted largely to a study of polyphyly of strains assigned to the genus Synechococcus. Moore et al. (2019) performed a study of a limited number of cyanobacterial genomes, comparing the position of chloroplast genomes to show the origin of the latter organelles, using 30 large and small subunit ribosomal protein sequences. Brito et al. (2020) employed a concatenated set of 1209 orthologous genes shared by Hyella patelloides strain LEGE 07179 and six other baeocyte-forming cyanobacteria, the resulting tree being limited to these seven strains. The branching order is similar to that shown in our tree, demonstrating the futility of using an excessive number of query genes in the analysis. Even though we show some additional strains, this interesting group clearly requires additional sequences. Chen et al. (2020) built a tree of 650 cyanobacterial genomes, including draft sequences of 163 strains of the FACHB culture collection, using a set of 31 unspecified universal single-copy genes "that has been adopted by many researchers". The tree includes genomes that are incomplete (e.g. Trichodesmium thiebautii H9-4 with only 44 of our 79 marker genes, Leptolyngbya valderiana BDU 20041 which contains 71 of 79) or misnamed (e.g. "Cyanobacteria bacterium USR001" should be Oscillatoriales cyanobacterium USR001) (this is also incomplete, with only 31 of 79 markers). Many strain numbers do not correspond to those given on the FACHB culture collection website, e.g. Cyanobacterium aponinum FACHB-4101 gives "no record"; Oculatella sp. FACHB-28 gives "Chlorella". The inclusion of incomplete sequences will cause major changes in the branching order of the tree, rendering it misleading and incorrect. The first detailed study of cyanobacterial taxonomy based on genomic sequences was that of Mareš (2018), who employed 23 conserved proteins (DnaG, Frr, InfC, NusA, Pgk, PyrG, RplB, RplC, RplD, RplE, RplF, RplM, RplP, RplT, RpoB, RpsC, RpsE, RpsI, RpsK, RpsM, RpsS, SmpB and Tsf) selected from those of Shih et al., mined from the complete chromosome of Synechocystis sp. PCC 6803. The phylogeny presented was inferred from tblastn analysis of the concatenated proteins and 191 genomic sequences (complete and draft) that contained all 23 markers. The detailed discussion of the state of cyanobacterial phylogeny, including polyphyly and a correlation of trees derived from genomic and 16S rRNA sequences, is well-worth reading. However, Mareš (2018) placed several strains, notably Chamaesiphon minutus PCC 6605 and Crinalium epipsammun PCC 9333, with Moorena producens strain 3L, rather than at the root of the heterocystous clade. Other studies, including that of Uyeda et al. (2016; 134 taxa, 137 unspecified amino acid marker genes plus 2 rRNA sequences) and our own, show that this configuration is unlikely.

The studies of Dvořák et al., Calteau et al. and Mareš gave rather similar phylogenetic trees, if we apply a degree of branch rotation; in contrast, the earlier work of Criscuolo & Gribaldo and Shih et al. placed the Nostocales clade in the centre of the tree, contrary to its position in the three other studies and in trees derived from analysis of 16S rRNA. Our own approach to the genomic phylogeny problem, started in 2017, is in good agreement with that of Mareš (2018); the methods are described in detail below. Other genomic trees are described in the genomic metrics document.

The trees of genomic sequences were produced from a set of 79 conserved proteins. These were obtained by tblastn studies of about 300 potential candidates from the complete chromosome of Gloeobacter violaceus PCC 7421 against the complete chromosome of Synechococcus elongatus PCC 6301, eliminating those that were found in 2 or more copies of similar e-value. Multi-copy candidates were retained only if the second and subsequent copies were found with e-values that would permit them to be screened out with suitable tblastn parameters. Genes that gave close to 100% identity with the most distant organisms (the heterocystous members) of the tree of complete chromosomes were also eliminated, since they were considered to be non-discriminatory; these included those encoding the constituents of the two photosystems. Also excluded were genes coding for products found only in certain genera (for example, gas vesicle proteins, toxins) or excluded from several genera (phycobiliprotein structural proteins and linker polypeptides). This left 79 conserved core marker genes essentially present in a single copy; these genes and their products are listed here. Although clustering of many of these genes is observed, many others are sufficiently spaced to permit coverage of the entire chromosome, as shown here. Their combined length is 19215 aa. Genome sequences were routinely examined with CheckM (Parks et al., 2015) to assess their phylogenetic position in the CheckM Bacteria tree, degree of completeness and contamination. The genomic sequences of Hassallia byssoidea VB512170 (JTCM00000000), one of two sequences of Mastigocoleus testarum BC008 (AXAQ00000000), Prochlorococcus marinus strains HNLC1 and HNLC2 (GL947594, GL947595) and Cyanobacteria bacterium 13-1-20CM (MNIV00000000) could not be placed as bacteria by CheckM, indicating that they were contaminated with a large proportion of non-bacterial DNA, and were therefore excluded from trees built (see below) with (a) complete chromosomes only and (b) all genomes (draft plus complete). Genomes were also checked with the help of a shell script to score the number of marker genes that they contain, the query cover and % identity of the corresponding HSP. This essential information is only obtainable because, unlike all previous publications, we do not concatenate the set of marker genes.

The trees of all and only complete genomes were produced with a shell script which concatenates the contigs of draft genomes into a single large contig. This is, of course, not ideal, but is required for further computation; however, duplicate copies of genes are screened out by our BLAST parameters, and a gene interrupted at one end of a contig will still have the same length after concatenation. The success of this method is amply demonstrated by the similarity of branching orders of the clusters in the trees built with complete chromosomes and with all genomes. The script then builds the BLAST database; uses tblastn to query the database with each marker gene amino acid product, screening out short and multiple hits; recovers and concatenates in order the corresponding HSPs from the individual genomes. For early versions of this site (up to and including version 9.6) alignment of the concatenated HSPs was performed with Muscle. Starting with version 10, muscle was replaced by MAAFT, running within the wrapper programme pagan2; the latter first produces an approximate alignment, then a guide tree, using the tree to refine the alignment. Both methods gave identical results. Regions of low similarity resulting from the alignment process were removed with Belvu, and FastTree (with the same settings as employed for 16S rRNA) produced the final tree. The genomic acronyms used to prepare the latter were converted to more complete labels (containing, in order, genus, species, strain, (type sp., ref. str. where appropriate), NCBI accession number and, if appropriate, the term metagenome) with (Taxnameconvert). The .pdf versions were produced from (XFig) following import and processing in ARB.

The genome-based trees were inferred de novo as new sequences became available, adding only a single sequence in each run. This was to avoid fixing the tree in a possibly incorrect configuration. If a sequence caused major topological changes it was removed, and subjected to further study. We use the term "circular" for genomes submitted to NCBI as "genome" (covalently closed, but probably not complete).

B2. Colour codes

The sequences are coloured in the genomic trees as:
complete chromosomes, red;
covalently closed, blue (unless type sp., ref. str., or cyanobiont);
draft genomes, violet (unless type sp., ref. str., or cyanobiont);
type species, green;
reference strains, black;
metagenomes, orange;
cyanobionts, pale blue. Note that we do not use this term for many organisms that are, for example, epizoic or epiphytic.

B3. Draft genomes

For draft genomes, those assessed as "Bacteria UID203" or more than 10% contaminated by CheckM, containing fewer than 77 of 79 marker genes (a gene being considered as present only if the query cover is greater than 40%), chimeric or arising from duplicate re-sequencing of the same strain (retaining the most complete sequence), were also excluded. Chimeric genomes containing segments from two or more cyanobacteria are very difficult to detect. A chimeric genome causes major changes in the tree topology; we extracted the core marker genes from each suspected chimeric genome and used them in a BLAST search for homology in the dataset, those finding more than 97.5% homology to genomes in dispersed clades being considered as representing chimeric genomes. Many may remain in the dataset, being undetected. To avoid overloading the tree, thereby making it more readable, we excluded many genomes whose clades contain a large number of sequences: these include many hundred Prochlorococcus marinus single-cell PCR sequences, metagenomic Microcystis spp, and 19 Fischerella sp. isolates from White creek, Yellowstone NP, USA. A complete listing of genomes excluded from the all-genome dataset is given in this table. For convenience, a simple textual list of the draft genome tree may be found here.

B4. Complete chromosomes

For inferring the tree based on complete chromosomes, we used only those listed in the NCBI as "complete genome" or "genome" (normally covalently closed with sequence gap), excluding WGS, providing they contain all 79 core marker genes. As for draft genomes, a gene was considered present only if the query cover was greater than 40%. Gloeomargarita lithophora D10 (CP017675) and Desikacharya piscinale CENA21 (CP012036) lack recN and ruvB, respectively; the organisms may genuinely lack these genes, or errors may have occurred in genome assembly. These genomes were omitted. Thermosynechococcus spp. strains BP-1 and NIES-2134 exhibit only 26% query cover (43.5% and 42.8% identity, respectively) with the marker recN and are annotated as having a "recN-like" protein. Curiously, the other members of Thermosynechococcus, strains CL-1, NK55a and PKUAC-SCTE542, show greater (91-100%) query cover and 45.5-46.7% identity with the marker gene. Dolichospermum sp. UHCC 0090 (previously named as Anabaena sp. 90) is unusual in possessing two circular chromosomes: chromosome 1 (CP003284, 4.33 Mbp) lacks 10 of the 79 core marker genes (efp, hemE, infC, mraW, nusA, pdhA, purM, ribF, rpsB, tsf), these being found on chromosome 2 (CP003285, 0.82 Mbp) together with a second copy of desC. The two chromosomes were concatenated for analysis and tree inference. A full listing of chromosomes excluded from the complete-chromosome dataset is given here, and a textual listing of the sequences here.

B5. In silico DDH, ANI and Mash-similarity

DDH was performed using the DSMZ genome-to-genome distance calculator (GGDC) available here. ANI values were calculated with JSpecies, the online version or the fastANI programme, which produces not only the ANI between two genomes but also a useful graphical summary. The Mash distances (MD) were calculated for all complete chromosomes using the wrapper script (Criscuolo, 2019), negating the default F81 correction by setting the -c option to 0.9 and using the -n option to stop the script after distance matrix construction. Other default values for sketch size and kmer size were retained after careful examination of the results obtained after varying each. The pairwise distances were extracted from the matrix with our own bash script, which converts them to percentage Mash similarity values, expressed as (1-MD)*100.

B6. Repeat sequences, core and accessory genomes

The genomic content of repeat sequences (direct, inverted, palindromes or shared between two organisms) was examined with the Repseek software. Core and accessory genomes were prepared by use of Spine.



Organism names are as submitted, except where changed in the NCBI flatfile Source or Organism fields or in later publications, especially if subsequently validated under the rules of the ICN. If applicable, the previous name is indicated in parentheses; in many cases the organism may have been identified only to the level of Order or Family. The type species of each genus is also shown.

16S rRNA genes from sequenced genomes:

The entries "rrnA" to "rrnG" indicate that the 16S rRNA sequence was extracted from the corresponding operon of a genome sequence, "rrnG" being employed for assembled complete genomes in which only a single operon exists (a common event near the root of the tree). Genomes are indicated as "complete" if they have been assembled; absence of this expression means that the genome is still in draft stage.

Colour codes:

16S rRNA sequences derived from PCR amplifications are indicated in blue except for those of known axenic strains (red) and type spp. (pale blue).

Sequences extracted from genomes of non-axenic strains are black;
those extracted from known axenic strains are in magenta;
those extracted from known metagenomes (identified as such in their NCBI flatfile, or known to be uncultured) are shown in orange.


The tree has been divided into higher ranks (from orders to families) at successive bifurcations. The orders 1-11 (designated as e.g. 1) are all divided into families (e.g. 1.1), except order 5, which is first divided into suborders (e.g. 5.1) and then families (e.g. 5.1.1) and order 11 (Gloeobacteriales). Each family is, if possible, further subdivided into genera (e.g. 1.1.1 or, for order 5 only, and most further into species (e.g. 1.1.1A or; if this is not possible (because the family contains only a single genus) the notation is (e.g) family 1.6/genus 1.6.1: Fortiea. Although many genera and their species were delineated by their bifurcation pattern within their family, many were obtained by applying strictly the 16S rRNA sequence identity cutoff values given above. This has created many genera and species containing only a single member, being sufficiently identical to others as to be included in the appropriate higher taxon but clearly separated from other members of that taxon. Such "singleton" or "loner" genera and species will be completed as sequences of related organisms become available. Loner genera are indicated by the term "(genus)", whereas loner species are numbered similarly to normal species. Although the clusters were originally labeled in strict numerical order (from top to bottom of the tree), addition of new sequences has caused some branch rotation and new genera have been given the next available number in their cluster; this has destroyed to some extent this numerical sequence. We are unable to re-number the clusters because the numbering scheme has been used in publications. A summary of the tree in strict numerical order is therefore provided.

Many genera are labelled as "monospecific genus". The monospecificity was established from 16S rRNA sequence identities; such genera usually contain several to many organisms identified as having many different specific (and, in some, generic) names. This liberal naming of organisms in the absence of phylogenetic data is very unwise, and contributes to the current chaotic state of cyanobacterial taxonomy. In many cases, we are unable to give a name to a specific or generic cluster, since it contains strains with many conflicting names; such clusters are represented only by their numbers.

Clusters of sequences are represented by polygons, in which the horizontal distances indicate the range of branch lengths and the vertical distance is a proportional representation of the number of sequences; the latter value is included in parentheses. If the trees are large, they are cut into multiple pages with an overlap of about 10 organisms.

The morphotypes tree:

We also provide a version of the tree in which the strains have been coloured in accordance with the Section (sensu Rippka et al., 1979) to which they are assigned; the clusters have been expanded only as far as required to obtain homogeneous groups. This tree is accessible from the Home page menu entry "Tree coloured by morphotype" or by clicking here. The colours employed are: Section I (unicells), red; Section II (baeocyte-formers), magenta; Section III (filamentous), pale blue; Sections IV and V (heterocystous), green; outgroup, gray. We have added the "Prochlorales" (Prochloron, Prochlorococcus) in orange, keeping Prochlorothrix separate (blue), and Acaryochloris (black). Examination rapidly shows the inadequacy of this approach. Although heterocystous cyanobacterial strains group into a single cluster, strains of Section V (stigonematacean cyanobacteria) are contained within the other heterocystous members of Section IV; distinction of the former at the ordinal level is therefore not maintained here. This is clearly shown in the tree of heterocystous cyanobacteria, again coloured according to morphotype (black for branching, red for non-branching forms), which is accessible via the link at the top of the morphotypes tree. Similarly, the order Prochlorales (see Lewin, 1989) is invalid: the unicellular Prochloron and Prochlorococcus strains, and the filamentous Prochlorothrix, all occur as genera within the cyanobacterial tree. Members of other Sections are often intermingled (e.g. unicellular strains of Section I mixed with filamentous isolates of Section III). The Section concept is thus only useful to denote a morphotype, not a true phylogeny. The doubtful sequences of Loriellopsis spp. and Scytonema millei strain VB511283, and the probable mutant strain "Pseudanabaena" MBIC10772, all described below, have been removed from this tree. Such non-heterogeneity has been remarked in many publications, the most extensive study (involving 1254 sequences) to date being that of Schirrmeister et al. (2011). However, the large treefile provided by the latter authors has the disadvantage of being readable (following download) only in specialized tree-viewing software.

Comments on the tree:

The final tree contains over 3000 sequences and has a total depth of 53 with a maximum distance to the root of only 0.453. Only bootstrap values greater than 50% are shown. We have chosen to employ such a large number of sequences, representative of all currently-available cyanobacterial taxonomic groups, in order to avoid misplacement of sequences caused by taxon sampling error (see e.g. Zwickl & Hillis, 2002). The latter is evident in many published trees showing cyanobacterial phylogeny, since the printed page permits a maximum of about 100 sequences to be included; to respect this limit, many organisms are excluded from the inference of phylogenetic trees for publication, causing taxon sampling errors and giving a false picture of phylogenetic relationships. Being too large for publication on paper, we present the tree on this site. We do not describe the tree here, leaving the reader to discover the current state of cyanobacterial phylogeny. We do, however, give some examples of conflicts and possible errors, in both this and linked documents.

Genera and species are discriminated in the tree on a strictly phylogenetic basis. The taxa defined are mostly in accordance with morphological (and, where available, other) characteristics. However, one should consider morphology as being highly plastic, as evidenced by the studies of Zapomělová et al. (2008) on strains of Anabaena crassa and A. circinalis (now Dolichospermum crassum and D. circinale). These species are normally differentiated on the basis of trichome width and the diameter of trichome coils. However, elegant cross-gradient experiments showed that these parameters were markedly affected by changes in phosphate concentration and light intensity such that, under a variety of conditions, the members of the two species were no longer distinguishable. Since both phosphate concentration and light intensity (as a result of self-shading) both change in the environment and during growth of cultures, we expect variation in cell dimensions of a single strain to occur. Additionally, morphology of the coccoidal Synechococcus elongatus strain PCC 6301 (earlier known as Anacystis nidulans) is easily changed to a filamentous form by mutation (Kunisawa & Cohen-Bazire, 1970), and Lehner et al. (2011) have demonstrated that a single point mutation in AmiC2, encoding cell wall amidase, is sufficient to change the filamentous Nostoc punctiforme strain ATCC 29133 to an aggregate-forming morphology. This closely resembles certain Cyanobacteria commonly named as "high temperature forms of Mastigocladus laminosus" or members of the genus Mojavia. Thus morphological characteristics must be interpreted with care and, in many cases, are without meaning.

The members of the phylum Cyanobacteria are all closely related phylogenetically, the most widely separated organisms (Leptolyngbya sp. strain 2LT21S03 and Synecococcus sp. strain JA-3-3Ab, excluding the environmental samples and bacterial outgroup at the base of the tree) sharing 82.9% 16S rRNA sequence identity (uncorrected p-distance). For comparison, the most distant members of the bacterial outgroup share only 71.6% identity. Nevertheless, the tree shows high internal heterogeneity, as evidenced by the mol %G+C values of the cyanobacterial 16S rRNA sequences that vary from 51.4 to 60.0, mean 54.6:


The peaks at or near 60% G+C correspond to some (but not all) extremophiles: (A) the Halothece members and Geitlerinema PCC 9228 (family 5.2.4) and (B) thermophilic "Synechococcus" of genera 10.1.1, 10.1.2, 10.2.1 and the unidentified thermophiles of genera 10.1.2, 10.2.1.

Of the chimeric sequences excluded, note in particular that the chimeric sequence of Umezakia natans strain TAC101 (AF516748) has been replaced by that of strain TAC611 (AB608023), Dolichospermum strain NIES 78 (AF313627) by AY701551 and Aphanizomenon strain NH-5 (AF425995) by AY196086. These substitutions resolve many uncertainties widely circulating in the literature.

The outgroups:

The outgroups of the 16S rRNA tree are shown in the outgroups tree, accessible from the button "The outgroups" on the Index page. The tree was rooted with a set of true bacterial 16S rRNA sequences. This cluster is separated from the genus Gloeobacter (genus 11.1.1) by two clusters of sequences obtained directly from the environment (environmental samples I and II). The positioning of numerous organisms such as Vampirovibrio chlorellavorus strain NCIMB 11383, Obscuribacter phosphatis strain Mlel_12 and others within the cluster "environmental samples I" indicates that this cluster most likely represents the Melainabacteria group. Further evidence is provided by the inclusion of sequences named as "uncultured bacterium" or "uncultured prokaryote". This also contains sequences named as "Cyanobacteria bacterium", "uncultured Gloeobacter" and "uncultured cyanobacterium". Further sequences with the latter name are found in the sister group "environmental samples II", together with a number of sequences named as "uncultured bacterium". The proposal (Soo et al., 2014) of the Melainabacteria as a class of the Cyanobacteria is not supported by their data, and Garcia-Pichel et al. (2020) argue strongly against this inclusion. The position in the bacterial outgroup of 9 cyanobacterial 16S rRNA sequences extracted from genomes (Anabaena aphanizomenoides LEGE 00250, Anabaena sp. MTD14b, Coleofasciculus sp. LEGE 07081, Cyanobacterium aponinum strain 0216, Leptolyngbya cf. ectocarpi LEGE 11479, Nodosilinea sp. LEGE 07088, Nodularia sp. LEGE 06071, cf. Phormidesmis sp. LEGE 11477 and Planktothrix mougeotii LEGE 06226) indicates the danger of sequencing the genomes of non-axenic strains. Nevertheless, the bacterial contaminant region of these genomes is, in some cases, sufficiently small to permit their inclusion in the genome tree based on our 79 marker genes; this applies to strains 0216 and MDT14b.

The clusters, based on 16S rRNA identities
and genomic analysis by
ANI, Mash-similarity, DDH and CGS

Click here to see the description. This document contains detailed data which are not documented on this page. We correlate, for clusters in which genomic sequences are available, the phylogenetic relationships inferred from 16S rRNA sequence analysis with the corresponding DDH and ANI values, and in many cases expand the analysis further with similarity values derived from Mash-similarity studies and from analysis by CGS. Several genera, including Synechocystis, the halophiles, the Prochlorococcus/Synechococcus cluster, and a few individual strains, are shown to be poorly supported by genomic comparisons, for reasons which are explained where possible.

Problematic organisms

Although all early publications defining cyanobacterial genera and/or species employed mostly 16S rRNA sequence identity, the trend in recent years is the much-heralded "polyphasic approach", in which sequence identity, morphology, ecological factors, secondary structure of the 16S-23S rRNA ITS (and rarely others) are supposedly taken into account. Often, however, one or more of these features is neglected, permitting definitions to be based on, for example, morphology alone or morphological plus ecological considerations, normally with a study of the ITS secondary structure, to the exclusion of 16S rRNA sequence data. Rarely, the latter data are employed alone, to the exclusion of other features. It seems that the "polyphasic approach" permits the inclusion or exclusion of features at the discretion of the authors, and papers with the same author(s) in common often employ such varied approaches. In the absence of a consensus, we have chosen to employ only 16S rRNA sequence identity in our definitions on these pages. Many of the problems encountered result from incorrect assumptions regarding the identity cutoff values required for separation of strains at the generic or specific levels, as explained in the Methods section of this page. We stress that, again as mentioned above, a sequence alignment is only correct if it takes into account the secondary structure of the small subunit RNA, and that distance measurements are accurate only if full-length sequences, of good quality, are used.

In addition, the printed page, unlike a website, imposes a limit on the number of strains that may be displayed in a phylogenetic tree. The trees normally attempt to illustrate all of the major clusters, condensed to fit the page; as a consequence, close relatives of a described taxon are often omitted, leading to taxon sampling error and completely false assumptions about the generic or specific assignment of that taxon. This could be alleviated by removing all irrelevant branches and expanding that under study.

We now give some examples of organisms which fail to meet the 16S rRNA sequence identity values or other criteria required for the separation of genera and species. They are listed, where possible, in alphabetic order, except for the members of genera 1.1.1 and 1.15.1, which are treated in the following paragraphs.

The planktonic heterocystous cyanobacteria are particularly problematic, as well described by many authors. We cite only three publications; further references may be found in each. These organisms fall into genus 1.1.1 of our tree, further divided into 14 species. Members carrying the generic names Anabaena, Aphanizomenon and Dolichospermum are clearly intermixed within these species. Species 1.1.1I, 1.1.1J and 1.1.1K contain exclusively members of Dolichospermum, but all contain strains carrying different specific epithets. Species 1.1.1H again contains Dolichospermum strains with different specific epithets, and a strain identified as An. cf. scheremetievi. Species 1.1.1L contains only Aphanizomenon strains and metagenomes, again with different specific names. Species 1.1.1A and 1.1.1C contain strains assigned to both genera. However, since these organisms are morphologically distinct, they should separate at the generic level. This has been the subject of extensive detailed studies, briefly described below, that reach no firm conclusion. Gugger et al. (2002) showed that planktonic members of the genera Anabaena and Aphanizomenon are intermixed in the phylogenetic tree, and that neither genus can be considered monophyletic. The 16S rDNA sequence similarity of members of the two genera was higher than 97.5%, which is well above the value we employ for genus discrimination. Rajaniemi et al. (2005) showed that benthic and planktonic Anabaena strains were intermixed and confirmed that both Anabaena and Aphanizomenon were not monophyletic genera. The 16S rRNA similarity values (again greater than 97.5%) between the strains indicated that the planktonic Anabaena and Aphanizomenon strains should be assigned to a single genus, divided into nine different species. Wacklin et al. (2009), further interpreting the results of Rajaniemi et al. (2005), transferred all planktonic morphospecies into the new genus Dolichospermum, with D. flos-aquae as type species, and suggested that different subclusters of the traditional genus Aphanizomenon also belong with Dolichospermum strains. This transfer was made because the description of the type species of the genus Anabaena, A. oscillarioides Bory ex Bornet et Flahault 1888, refers to benthic organisms. The authors concluded that the final taxonomic position of several other types of planktonic Anabaena is still unsolved, and that their work should be the basis for further studies. As may be seen from our tree, inferred 10 years later, progress has not, unsurprisingly, yet been made. We support and extend the division of this generic cluster into the nine species described by Wacklin et al., and are still unable to separate Aphanizomenon and Dolichospermum at the generic level despite their obvious morphological differences.

The Hapalosiphonaceae form genera 1.15.1 to 1.15.5 in our tree. Genera 1.15.2 and 1.15.3 both contain strains identified as Chlorogloeopsis spp., all originating from hotsprings (except Chlorogloeopsis fritschii PCC 6912, isolated from a terrestrial environment and PCC 6718 for which information is lacking). Genera 1.15.4 and 1.15.5 both contain only strains of Fischerella/Mastigocladus; where information on the habitat is provided, these were all again isolated from hotsprings, except Fischerella sp. PCC 9605 which has a desert origin. Genera 1.15.2 to 1.15.5 cause few apparent problems and are not discussed in detail here, except to confirm that they correspond to the generic clusters C2 (1.15.2), C1 (1.15.3) and M (1.15.4) in the tree of Casamatta et al. (2019), discussed below. Genus 1.15.5, containing only two strains, is not included in that tree. Genus 1.15.1 is problematic. We can confirm that this is a single generic entity by measuring the 16S rRNA sequence similarities of the strains therein: using only the longer sequences, the minimum similarity is 96.53%. This phylogenetically-defined genus contains strains with many different generic names (Fischerella, Hapalosiphon, Mastigocladus, Nostochopsis, Pelatocladus, Westiella, Westiellopsis) and is clearly in need of careful revision. The recently-described "genera" Neowestiellopsis (which lies within species 1.15.1G) and Reptodigitus (forming species 1.15.1E) add to the confusion rather than solving the problem, as described below.

Reptodigitus: this genus was erected by Casamatta et al. (2019) on the basis of a single strain, R. chapmanii strain SJRDC1, which falls as species 1.15.1E of our tree. As for Neowestiellopsis (below), we find no evidence in support of distinction of this strain at the generic level from the other members of our genus 1.15.1. Selecting only long sequences (this excludes members of species 1.15.1C, F and H, all represented by a single sequence of length 1292-1361 nt) for analysis, we find 16S rRNA similarity values of 96.53 to 98.41%, above our lower level genus cutoff value of 96.5%. The tree of Casamatta et al. shows generic level clades W (species 1.15.1A), N (which combines species 1.15.1B and 1.15.1C), R (which combines species 1.15.1D, 1.15.1E and 1.15.1F), F1 (species 1.15.1G), F2 (species 1.15.1I), H (species 1.15.1J) and P (species 1.15.1K). Species 1.15.1H of our tree is not represented. The tree shown by the authors is said to have been inferred from long sequences of length up to 1540 nt, considered to be more accurate than that inferred from shorter sequences of length 1170 nt. The use of longer sequences is encouraged elsewhere on this page. However, a length of 1540 nt for a 16S rRNA molecule is far greater than we expect. Regarding the sequences employed, that of Fischerella ambigua ISC 4 is of poor quality in containing many mismatches in paired regions of the 16S rRNA molecule and has been excluded from our tree; the sequencing errors explain the long branch leading to this strain in the tree of Casamatta et al. and the lower value (95.0%) of internal genetic identity shown on the Figure. In addition, the 16S rRNA sequences of Nostochopsis sp. strains with accession numbers KC854775-KC854779, which were deposited in NCBI by authors of which two appear on the paper of Casamatta et al., are attributed lengths of 1538-1753 nt in NCBI but are in fact short (1135-1186 nt, starting only at E. coli position 359), the excess being ITS; the latter appears to have been included in the analysis and would explain the low (97.3%) internal genetic identity value shown for cluster N. Regarding the delineation of species and genera, the cutoff values employed by the authors to discriminate at the specific level are correct, but the value cited (94.5% sequence identity; Yarza et al., 2014) is far too low. They appear to discriminate strains which share higher than 98.7% 16S rRNA sequence identity as members of the same genus, those with lower values belonging to separate genera. With our values, strains sharing more than 96.5% 16S rRNA sequence identity are members of the same genus, and those sharing greater than 98.7% are members of a single species. This implies that the morphological differences traditionally employed to discriminate cyanobacteria at the generic level may, in fact, be applicable only at the specific level, at least within the Hapalosiphonaceae.

Neowestiellopsis was created by Kabirnataj et al. (2018), with two species (N. bilateralis and N. persica), and said to be clearly distinguished from members of the genus Westiellopsis. Their 16S rRNA sequences are almost identical (99.72%), and thus represent a single species. N. persica (the type species) SA33 (reference strain) shares 96.83-100% 16S rRNA sequence identity to Westiellopsis spp. strains Ar73, AKM9 (both in our species 1.15.1A), SAG 19.93 (1.15.1G) and SAG 16.93 (1.15.1J), and is therefore not distinguished at the generic level from Westiellopsis. A third sequence, Neowestiellopsis sp. KHW5 (Singh & Mishra, unpublished), shows 99.51-99.79% identity to the others. All strains are found in species 1.15.1G of our tree. The nearest, but not identical, neighbour to the Neowestiellopsis strains in the tree of Kabirnataj et al. is Fischerella sp. strain 1711; we find a 16S rRNA sequence similarity of 100% between this and N. persica SA33. This discrepancy probably results from incorrect sequence alignment by the authors; this is further confirmed by our finding of 100% sequence identity between Pelatocladus maniniholoensis HA4357-MV3 and Hapalosiphon hibernicus BZ-3-1 (species 1.15.1K), which are separated by a long branch in the tree shown by the authors. The generic name Neowestiellopsis is, therefore, not valid: the strains are merely a single species of the large genus 1.15.1.

Alborzia: this genus was created by Nowruzi and Soares (2021) to accommodate a single strain, A. kermanshahica S2, which is the type species and reference strain. Unfortunately, the secondary structure of the single 16S rRNA sequence shows 28 mismatches in the paired stems of 16 helices. These are sufficient to produce a long branch for this strain in inferred phylogenetic trees, as shown in the tree of the authors, where it lies near to various Chroococcus spp. We are confident that re-sequencing will show this strain to fall within a cluster containing the latter organisms, and we have not added this organism to our own tree.

Aliinostoc: we show 17 strains which form 13 species of genera 1.3.4 and 1.11.1, with the type species A. morphoplasticum (reference strain: NOS) lying in species 1.3.4F. The sequences were provided by numerous authors and initially formed the coherent genus 1.3.4. However, Saraf et al. (2018) subsequently described two new species, A. tiwari and A. soli, that grouped loosely with the other Aliinostoc spp. in a phylogenetic tree containing only strains of this genus. Examination of the ITS showed minor differences of the secondary structure of the D1-D1′ helix between these strains, and major differences to that of the reference strain, these changes being used to confirm separation of the strains at the specific level. A. tiwari strain LI_PS and A. soli strain ZHI fall, in our tree, into genus 1.11.1 as two species (1.11.1G and 1.11.1D, respectively); we note that the sequence of strain LI_PS contains a paired insert in the stem structure of helix 6. The new sequences, like that of the type species, are almost full-length and show only 95.0-96.0% identity to strain NOS. The authors found similar values, but failed to employ them correctly. The ITS analysis is a good example of a subjective approach, where differences can be used to group strains into different species of a single genus when, in fact, they represent different genera. Kabirnataj et al. (2020) described 3 new isolates assigned to this "genus", each forming a separate species: A. catenatum strain SA24, A. magnakinetifex SA18 and A. constrictum SA30. In our tree, strains SA24 and SA18 fall into a single genus as species 1.3.4I and 1.3.4L, respectively, with 98.29% 16S rRNA sequence identity; strain SA30 is seen in genus 1.11.1 (species 1.11.1F), sharing 97.74% 16S rRNA sequence identity with strain Li_PS (species 1.11.1G) and does not contain the helix 6 insert present in the latter strain. The tree shown by these authors again contains (with one exception) only strains assigned to Aliinostoc, and it is quite clear from the tree topology that the strains represent at least two distinct genera. Lee et al. (2021) created the genus Pseudoaliinostoc, on the basis of a single strain, and transferred A. soli, A. tiwarii and A. constrictum to this genus. This may well resolve the problem within the Aliinostoc "genus", with cluster 1.3.4 being reserved for Aliinostoc strains and cluster 1.11.1 for Pseudoaliinostoc spp.

Members of the genera Anabaenopsis and Cyanospira, showing major morphological differences, have a minimum 16S rRNA sequence identity of 97.6%; they therefore represent several species of a single genus. Many of their sequences are almost full length (starting at E. coli position 28 and complete at the 3′ terminus). They are represented in the tree as species 1.3.2A (Cyanospira spp. and three strains of A. abijatae), 1.3.2B and 1.3.2C (containing exclusively Anabaenopsis spp.). The strains of C. rippkae and C. capsulata cannot be separated on the basis of 16S rRNA sequence homology, and the same applies to the strains identified as A. circularis, A. elenkinii and A. nadsonii. One additional sequence, seemingly mis-identified as C. rippkae, represents species 1.3.7F.

Members of the genus Capillus (Caires et al., 2018), subsequently renamed to Capilliphycus (Caires, 2019), fall into genus of the tree as 3 species, sharing 97.4-98.1% 16S rRNA sequence identity, the type species (C. salinus) being the only member of species Genus also contains strains named as three species of the genus Limnoraphis: L. cryptovaginata, L. hieronymusii and L. robusta (Komárek et al., 2013); these strains share 99.9% 16S rRNA sequence identity and fall into the single species of the tree. Where known, the strains named as Capilliphycus are from saline habitats, whereas the Limnoraphis isolates are of freshwater origin. Since Limnoraphis J.Komárek, E.Zapomělová, J.Smarda, J.Kopeckỳ, E.Rejmánková, J.Woodhouse, B.A.Neilan & J.Komárková, 2013 has clear priority over Capilliphycus, described 5 years later, the members of the latter "genus" should be renamed. The genus Limnoraphis itself requires emendation, with 2 of the 3 species being renamed to L. hieronymusii, the type species.

Chroakolemma. Members of this genus, assigned to two species (C. opaca and C. pellucida), were indistinguishable by analysis of their nearly full-length 16S rRNA sequences, but separable by their different ITS secondary structures (Becerra-Absalón et al., 2018). Since there is no evidence to suggest that the latter is valid as a single character to delineate taxa, we prefer to retain only a single species, C. opaca (the type species of the genus); the strains share 99.3-100% 16S rRNA sequence identity and fall into the mono-specific genus 6.3.21 in our tree. They do not form a sister clade to Scytolyngbya (mono-specific genus 6.3.31) as suggested by Becerra-Absalón et al. (2018), this being another example of incomplete taxon sampling.

Coleodesmium wrangelii Borzi ex Geitler: Flechtner et al. (2002) isolated one strain of Coleodesmium wrangelii, the type species. We show the sequences of two of the 3 clones of strain MC-JRJ1 (all unfortunately short, 1108 nt), which fall into species 1.14.1C with 99.37% 16S rRNA sequence identity. Two sequences (1112 nt) of the single strain (SEV2-5-Ca) of Tolypothrix distorta included in the study of Flechtner et al. are placed in species 1.14.1K of our tree. The published tree shows grouping of these strains, but possible separation at the generic level was not confirmed by values of 98.6-99.1% 16S rRNA sequence identity. One additional sequence, as Coleodesmium cf. scottianum ANT.LH52B.5 (Taton et al., 2006), is almost complete and is found here in species 1.14.1G, with 96.58% 16S rRNA sequence identity (using the shorter comparable region) with the strains of Flechtner et al. Using only the complete sequences available for our cluster 1.14.1, we can show that all strains form a single genus with a minimum of 96.63% 16S rRNA sequence identity. Coleodesmium wrangelii strain MC-JRJ1 and Coleodesmium cf. scottianum ANT.LH52B.5 are therefore members of the large genus Tolypothrix sensu stricto that contains all 17 known strains of T. distorta, the type species.

Compactonostoc shennongjiaensis: this genus was erected on the basis of a single 16S rRNA sequence, almost full length, by Cai et al. (2019b). Strain CHAB5781 showed 95.4-97.4% 16S rRNA identity with various strains of Cylindrospermum, Desmonostoc, Goleter, Komarekiella, Mojavia and Nostoc, and appears as an isolated branch, distinct from Komarekiella, in the tree presented. However, in our tree Compactonostoc shennongjiaensis CHAB5781 is seen only as a species of the Komarekiella generic cluster 1.8.5, sharing 97.1-97.2% 16S rRNA sequence identity with the other members, mostly represented by almost full length sequences. The genus Compactonostoc is therefore not valid.

Constrictifilum karadense: The genus Constrictifilum was created by Chavadar et al. (2021), based on the 16S rRNA sequence of a single strain (MKW3); C. karadense is the type (and only) species and MKW3 is the reference strain. Strain MKW3 lies in our genus 1.8.4 (species 1.8.4F), and was observed by the authors (and confirmed here) to cluster with Nostoc sp. CENA239 and Calothrix sp. NIES-2100 (species 1.8.4H and 1.8.4D, respectively). However, this genus also contains Calochaete cimrmanii CCALA 1012 (type species and reference strain) in species 1.8.4G and Calochaete sp. UIC 10574 (species 1.8.4I), the two type species sharing 97.24% 16S rRNA sequence identity. Since the genus Calochaete was validly published earlier (Hauer et al., 2013), it must be considered as having priority. C. cimrmanii CCALA 1012 is shown as an outlier to the 1.8.4 cluster in the tree of Chavadar et al. (2021), presumably illustrating poor sequence alignment.

Cryptochroococcus: Wang et al. (2021) described 4 strains (with sequences of multiple clones of each), all attributed to the type species C.tibeticus, with strain TP201716.4 as reference strain. These organisms lie in species of our 16S rRNA phylogeny. Species is comprised of the single sequence of the type species and reference strain Pseudochroococcus couteii PMC 885.14, the two reference strains sharing 98.83% 16S rRNA sequence identity. The latter organism was not shown in the tree of Wang et al. The genus therefore contains organisms named as two different genera; Pseudochroococcus couteii (Duval et al., 2021) published earlier, must have priority.

Cyanobacterium: strains named as members this genus (erected by Rippka & Cohen-Bazire, 1983) are found in 4 generic clusters in our tree ( [one strain], [6 strains], [2 species, 4 strains, including the type species C. stanierii and the reference strain PCC 7202] and [1 strain]). These clusters are separated at the level of 89.43-93.12% 16S rRNA identity. Moro et al. (2007) commented on the likely separation at the generic level of C. stanierii and C. aponinum strain CENA527, but continued to include both under the same generic name. Silva et al. (2014) showed clear separation of strain CENA169 from other members of Cyanobacterium, but again chose to retain the single generic name. As further documented on the page describing DDH values, the strains of genus require two specific names, and the generic name of C. aponinum should be changed.

Cylindrospermum: strains assigned to this genus fall into three generic clusters of our tree; genus 1.5.2 (as 2 species) and genus 1.8.2, both probably badly named, and genus 1.7.1. The latter is the main cluster of Cylindrospermum isolates, and divides into 6 species. Many of these contain organisms carrying different specific epithets: species 1.7.1A, for example, holds C. alatosporum, C. maius (the type species) and C. stagnale; species 1.7.1C contains C. catenatum, C. muscicola and C. pellucidum. The specific names employed within this genus deserve to be rationalized.

Dactylothamnos: this genus was erected by Komárek et al. (2015), and the reference strain D. antarcticus CENA410 was included in the study. Their tree shows this strain to cluster with isolates of Coleodesmium, Hassalia, Rexia and 6 "Microchaetaceae" CENA strains which are in our tree as Kryptousia microlepis (species 1.14.1B) and K. macronema (species 1.14.1H). D antarcticus CENA410, with two other strains deposited in NCBI by Fiore et al. and given in NCBI as unpublished, fall into species 1.14.1A; the sequences are all of good length (1412-1462 nt) and show 99.72-99.93% 16S rRNA sequence identity. Unfortunately, the tree of Komárek et al. has no scale marker, and distance values are not given in the publication. As explained in our discussion of Coleodesmium, the genus "Dactylothamnos" must be considered as a member of the genus Tolypothrix sensu stricto whose members are not separable at the generic level.

The genus Desikacharya was erected by Saraf et al. (2019a) on the basis of two strains: D. nostocoides BDU1-PS and D. soli BDU2-PS, together with a strain "Nostoc" thermotolerans 9C-PS which was transferred to the genus as D. thermotolerans 9C-PS. These formed a cluster with other Nostoc strains (CENA21, PCC 9426, TH1SO1, SAG 2306) and with Trichormus azollae Kom BAI/1983, which were subsequently transferred into the new genus. The cluster shown in the tree of Saraf et al. looks convincing, but the authors failed to take into account the 16S rRNA similarity values in assigning specific and generic names, stating that "the phylogenetic positioning of the strains should be given more importance rather than the percentage similarity values". The meaning of the latter statement is unclear to us, since a phylogenetic tree is based on similarity (or rather distance) values. The uncorrected p-values given by the authors support the establishment of the genus Desikacharya for strains BDU1-PS and BDU2-PS, but the transfer of N. thermotolerans 9C-PS is questionable since it shares only 96.2% 16S rRNA sequence identity with the type species D. nostocoides; strains N. piscinale CENA21 and T. azollae Kom BAI/1983 appear to be distinct genera. Curiously, data are not provided for strains PCC 9426, TH1SO1 or SAG 2306. As a result of the over-zealous transfer of strains, the members of this "genus" in fact form five genera in our tree, sharing low 16S rRNA sequence identities, all members of family 1.12. Four of the strains fall into genus 1.12.5 as three species. Genus 1.12.3 contains four species, including D. piscinale CENA21 (note that we have replaced the 16S rRNA sequence version of this strain (AY218832) by the sequence extracted from the complete chromosome of this organism), and three strains ("Calothrix dolichomeres" MBDU 013, "Nostoc" sp. strain PCC 9104 and "Nostoc" sp. strain MBR 210, represented by the 16S rRNA gene extracted from the draft genome) that were not included by Saraf et al. The single species of genus 1.12.4 contains two strains and the two other genera each contain only a single strain, D. thermotolerans 9C-PS and D. nostocoides BDU1-PS, respectively. The tree shown in the original publication omitted all near relatives of the Desikacharya strains, thereby showing serious taxon sampling error; the closest relatives, a cluster of Cylindrospermum isolates, being found as genus 1.7.1 in our tree. Note that the type species was not given in the original publication, an error corrected subsequently (Saraf et al. 2019b). In view of the above comments, this genus clearly requires emendation.

The salt tolerant strain Desmonostoc salinum CCM-UFV059 (KX787933), isolated from periphytic microbial mats in a saline–alkaline lake in Patagonia (de Alvarenga et al., 2018), falls into species 1.8.2E of the tree, together with isolates named variously as Desmonostoc sp., Dolichospermum flos-aquae, Nostoc sp., N. entophytum, N. linkia and N. piscinale, with which it shares 99.5-99.9% 16S rRNA sequence identity. Where known, these isolates are cyanobionts, epiphytes, or of freshwater or terrestrial origin. Strain CCM-UFV059 shows a minimum of 97.2% 16S rRNA sequence identity with other members of the genus Desmonostoc shown in the tree, and is undoubtedly a member of this genus; however, it would be unwise to name the entire cluster 1.8.2E as Desmonostoc salinum in the absence of information on salt tolerance of other members, particularly in view of the diverse habitats from which they were isolated. The specific epithet salinum should be considered as invalid. We note also that de Alvarenga et al., 2018 studied the fatty acid profile of this strain; we can have little confidence in the latter results, obtained from a non-axenic organism.

Drouetiella: of the three named species of this genus (Mai et al., 2018), D. lurida (the type species) and D. hepatica do indeed separate at the specific level, sharing 97.8% 16S rRNA sequence identity (species 6.2.7B and 6.2.7A, respectively). However, the single D. fasciculata strain shares only 96.2% identity with the D. lurida strain and 96.3% identity with the D. hepatica isolate and is therefore a member of a different genus. This clustering might change if full-length sequences become available; at present, those of D. lurida and D. hepatica are even shorter than normally provided by these investigators, being incomplete at the 3′ end.

Dulcicalothrix. This genus, with D. necridiiformans as type species and V13 as reference strain, was erected by Saraf et al. (2019c) on the basis of this single strain. It lies in genus 1.15.9 as the single member of species 1.15.9C. The authors transferred many species of Calothrix into Dulcicalothrix (shown as Clade 1 in their tree) and proposed the re-creation of the family Calotrichaceae. However, these strains all fall into genus 1.15.7, very different from the reference strain of Dulcicalothrix. Consequently, we have not renamed the Calothrix spp. strains in our trees. More recently-sequenced members of this genus fall into species 1.15.9B [Dulcicalothrix sp. Zw1-PS (Saraf et al. unpublished) and strain 25C-PS (Kumar et al. unpublished)] and are therefore members of the same genus as the type species. However, D. alborzica str. Alborz (Nowruzi & Shalygin, 2021) lies, like the Calothrix spp., in genus 1.15.7, as a result of this mistaken transfer.

Mareš et al. (2019) sequenced multiple clones of 7 strains, plus a single clone of 1 strain, and assigned them to the genus Gloeothece. They also transferred 3 strains previously identified as Cyanothece spp. to this genus. This produced a phylogenetic cluster containing divergent 16S rRNA sequences with identities as low as 92.6%. The authors chose to use a conservative approach, maintaining a single genus containing 14 organisms. Applying our cutoff values of 96.5-96.9% identity results in the division of this "genus" into 4 generic clusters, seen in our tree as genera,, and, with the additional genus containing unpublished sequences of PCC strains which were not available to the authors. Genus (Gloeothece I) is further divided into three species at our species-level identity cutoff of 98.7-99.0%, but species contains strains identified as G. membranacea and G. tepidariorum. The only two strains represented by almost full length sequences, G. membranacea PCC 6501 (starting at E. coli position 28) and G. tepidariorum PCC 6909 (starting at E. coli position 13, both ending at the 3′ terminus, neither sequenced by the authors), show 99.86% identity and differ by only 3 bases in the long comparable region of their 16S rRNA sequences, one difference resulting from an ambiguous base in strain PCC 6909. The sequences obtained by Mareš et al. start at best at E. coli position 551, thereby lacking a substantial part of the 5′ region. Our cutoff value separates each of the genera (Gloeothece III), and (Gloeothece IV) further into species that agree with the name appended. We note that the full-length 16S rRNA sequences of the two strains lying in genus were not obtained by the authors. The descriptions given by the authors for G. membranacea and G. tepidariorum reveal two strains with largely overlapping morphological properties (e.g. cells in groups of 2-8, up to 32 vs 2-4, up to 16; cell length 5-9 vs 5.6-9 μm) and minor differences in the secondary structure of the ITS (3- vs 5-base terminal loop of the D1-D1′ helix), which do not convincingly support their separation at the specific level. The difference in the D1-D1′ terminal loop results from a single base change. We describe above, in comments on the 16S rRNA tree, the dangers inherent in over-emphasizing morphological characteristics. The further transfer of strains previously identified as Cyanothece or Gloeothece to the new genera Rippkaea (with a single species) and Zehria (two species) is welcome and leaves only two members of the genus Cyanothece: C. aeruginosa (represented by the single strain SAG 87.79) and C. svehlovae (with only strain NIVA-CYA 258/1) in the tree of Mareš et al. The overall picture of these organisms therefore looks convincing, but we find many more strains identified as Cyanothece spp. elsewhere in our tree, as described in the section "Polyphyly". Strain SAG 87.79 is treated in the section on type species elsewhere on this page.

Hassalia was studied by Komárek et al. (2011), who created two new species (H. andreassonii and H. antarctica). The tree in this publication shows that Hassalia splits into 2 clusters as in our tree, one containing H. andreassonii CCALA954 and the identical Coleodesmium sp. ANT.LH52B.5 (sequenced by Taton et al., 2006), the second containing the 3 strains of H. antarctica and a single sequence of H. byssoidea (the type species) strain CCALA823, not obtained in this publication. The 3 clones of C. wrangeli mentioned above are closely related to the Hassalia clusters. No estimates of 16S rRNA sequence identities are given and, since the sequences are all short, this was a wise decision; however, the scale marker of the tree indicates that all strains must be members of a single genus. Their sequences fall into species 1.14.1F and 1.14.1G of our tree. An additional sequence of Hassalia, H. littoralis strain C76 (González-Resendiz et al., 2013), lies in species 1.14.1D. More sequences of Hassalia, again made available after the publication of Komárek et al. (2011), lie in species 1.14.1I, 1.14.1J, 1.14.1K and 1.14.1L. All available sequences of Hassalia, therefore, fall into a single phylogenetically-defined genus, Tolypothrix sensu stricto, together with the known strains of the type species T. distorta.

Crinalium and Hormoscilla: these genera were placed into the family Gomontiellaceae by Komárek et al. (2014) in the most recent taxonomic revision, together with Starria, Gomontiella (the type of the family is Gomontiella subtubulosa, for which no 16S rRNA sequence exists), Katagnymene and Komvophoron.

C. epipsammum was described by de Winder et al. (1990) and deposited in the SAG as strain SAG 22.89; the strain PCC 9333 (only one of four identical sequences extracted from the genome are shown here) is the same isolate, rendered axenic. Their 16S rRNA sequences, together with those of two of C. magnum (SAG 34.87 and Hg-6-6, the latter described by Mikhailyuk et al., 2019), three of H. pringsheimii (CCALA 1054 and SAG1407-1, the latter represented by two sequences) and one of H. cf. pringsheimii (Us-s-6-2) fall into family 2.2 as genus 2.2.3 in our tree. Excluding the shorter sequence (CCALA 1054), the sequences share 98.89-100% identity over their almost-complete lengths, and therefore appear as a single species of a single genus. Previous studies have found similar results, e.g. Bohunická et al. (2015) and Mikhailyuk et al. (2019). Bohunická et al. (2015) found that the strain C. magnum SAG 34.87 more closely fitted the desciption of Hormoscilla and, perhaps as a consequence, this strain is not currently available from the SAG. This shows that the distinction of the two genera, based on the available morphological descriptions, may not be clear. Starria was erected by Lang (1997) and assigned to the family Oscillatoriaceae. Starria zimbabweënsis SAG 74.90 is a member of a different genus (genus 2.2.4 in our tree), sharing 96.40-96-74% 16S rRNA sequence identity with members of the Crinalium/Hormoscilla cluster, and remains within the family Gomontiellaceae. A strain of Katagnymene spiralis (JWI1) was isolated by Lundgren et al. (2005) and transferred to the genus Trichodesmium; the 16S rRNA sequence is available from NCBI as T. pelagicum (sequence length only 584 nt). This strain clusters with Trichodesmium spp. in genus (see also Lundgren et al., 2005) and is therefore not a member of the family Gomontiellaceae. Komárek et al. (2014) maintained Katagnymene in the Gomontiellaceae, but pointed out that the genus may be problematic, in view of the affiliation of K. spiralis with Trichodesmium spp. Hašler et al. (2014a) found that the genus Komvophoron divided into two groups, the first (containing strains of K. hindakii and K. kgarii) being related to the members of the family Gomontiellaceae, the second (containing 3 clones of a single J. constricta strain), showing only 88% 16S rRNA sequence identity to K. hindakii, being more closely related to Spirulina spp. The latter cluster was renamed to Johansenia constricta, subsequently corrected to Johanseninema (Hašler et al., 2014b), and removed from the family Gomontiellaceae. Three sequences of J. constricta are shown in genus of our tree. Bohunická et al. (2015) agreed that K. hindakii and K. kgarii (the type species) are a sister group to the Gomontiellaceae strain cluster, but comment that, since "the 16S rRNA identity between Komvophoron and Gomontiellaceae was only 90%–91%...including Komvophoron inside the family...appears to be debatable". The Komvophoron strains are found in our tree not within the family Gomontiellaceae but in family 2.1 (the K. hindakii sequences representing 4 clones of a single strain), together with isolates of Chamaesiphon. The genus Ancylothrix was erected later than the revision of Komárek et al. (2014) and proposed by Martins et al. (2016) as a member of the Phormidiaceae. No close relatives of Ancylothrix are shown in the tree of the latter authors. Strains assigned to the genus Ancylothrix show low (90.25%, A. terrestris and 90.04%, A. rivularis) sequence identity to those of the Crinalium/Hormoscilla cluster and 88.24-89.27% sequence identity with Starria zimbabweënsis; the two proposed species of Ancylothrix share only 95.55-96.02% sequence identity and therefore must be considered as representing two distinct genera, shown in our tree as genera 2.2.1 and 2.2.2 of the family Gomontiellaceae. Thus with the removal of Katagnymene and Komvophoron, and the addition of "Ancylothrix", the family Gomontiellaceae contains the genera "Ancylothrix" (as 2 genera), Crinalium, Hormoscilla and Starria, on solid phylogenetic evidence.

Regarding Crinalium and Hormoscilla whose members, as stated above, must be considered to represent a single species, Crinalium W.B.Crow, 1927 clearly has priority over Hormoscilla Anagnostidis & Komárek, 1988 and, since it appears unwise to have two "genera" whose members show such high 16S rRNA identity values, Hormoscilla should, in principle, be invalidated. However, we await further evidence, notably a sequence of the genome of a strain identified as an authentic member of the genus Hormoscilla, which would permit DDH/ANI studies, before taking this step. Two genomic sequences were obtained (Schorn et al., 2019) from uncultured material enriched from tropical sponges of the family Dysideidae, which host a high abundance of a supposedly obligate symbiotic uncultured filamentous cyanobacterium now known as Hormoscilla spongeliae (Gomont) Anagnostidis & Komárek (previously Oscillatoria spongeliae Schulze ex Gomont). A further 4 metagenomic sequences, of which only 3 are shown in the 16S rRNA trees, were obtained by Podell et al. (2020) from a similar source. The 16S rRNA sequences extracted from these genomes, however, fall into genus 4.4.1 of our tree with other sequences of strains assigned to H. spongeliae, which form 4 species. These are the only members of family 4.4. This is clearly an unacceptable situation, since members of the "genus" Hormoscilla as now defined (to include H. spongeliae), lie in two different orders; the premature transfer of the marine symbiotic Oscillatoria spongeliae into the new genus Hormoscilla by Anagnostidis & Komárek (1988) was a therefore a serious error. Nevertheless, both H. spongeliae and Hormoscilla members of family 2.2 are classified as members of the family Gomontiellaceae in a single order in NCBI.

Hyella. Jung et al. (2021) renamed Chroococcidiopsis sp. strain PCC 6712 to Hyella disjuncta, and claimed that the strain clustered with Hyella patelloides LEGE 07179. In our 16S rRNA phylogeny, however, the two organisms are clearly members of different genera, sharing only 95.81% sequence identity.

Koinonema. Part of the description given here is briefly included in section A3 (sequence length) in the Methods, above. Four sequences, all nearly full length, were deposited by Buch et al. (2017) as representatives of the type species Koinonema pervagatum. They fall into two different species of genus in our tree. The three members of species share 98.92-99.07% sequence identity estimated from the full-length sequences, with the reference strain (63PC) in species showing 98.48-98.63% identity to the first species. These sequences are accompanied by those of strains ETS-05 and ElfPHct20, initially deposited as Phormidium spp., which lie in the third species of this genus and share 99.06% 16S rRNA sequence identity. A similar topology is observed in the tree of Buch et al., but these authors did not separate strain 63PC at the specific level.

Kryptousia: the genus, erected by Alvarenga et al. (2017), contains the species K. microlepis and K. macronema (the type species). Four sequences of K. microlepis fall into species 1.14.1B of our tree, showing 99.45-99.86% 16S rRNA sequence identity; 2 sequences of K. macronema, including that of the reference strain CENA338, lie in species 1.14.1H and exhibit 99.86% 16S rRNA sequence identity. The two species are related by 97.81-98.29% identity. The tree of Alvarenga et al. contains representatives of the genera Coleodesmium, Dactylothamnos, Hassalia, Rexia and Spirirestis; no distance estimates are provided but the scale marker of the tree indicates that generic separation of strains within this cluster is doubtful. All strains described in this publication lie in genus 1.14.1 (Tolypothrix sensu stricto) with a large number of strains identified as T. distorta, the type species of Tolpothrix.

Kyrtuthrix and Nunduva are problematic genera. A single 16S rRNA sequence of Kyrtuthrix (León-Tejera et al., 2016), K. huatulcensis (KT936560), was obtained from cultured material of a sample from a granitic coastal cliff, San Agustín bay, Mexico. No evidence is provided that the culture was uni-cyanobacterial, and the micrographs of cultures show a few cells in a pitifully unhealthy state; micrographs of the environmental samples can be interpreted as showing trichomes that are desiccated (as a result of osmotic shock?). The sequence (the consensus of three clones, for an unknown reason, but of reasonable length) cannot, therefore, be assigned unequivocally to the genus Kyrtuthrix. It appears in our tree as a member of the marine Rivularia clade (genus 1.15.15), sharing 98.0% and 97.7% 16S rRNA sequence identity with, for example, Rivularia sp. PCC 7116 and R. atra BIR KRIV1. The authors state that "Kyrtuthrix species demonstrate a complex morphology that involves variable cell shape and trichome disposition according to the life form (endolithic vs epilithic), habitat (supralittoral to mesolittoral), substratum (calcareous vs granite) and region of distribution (temperate vs tropical). There are representatives where irregular disposition of contorted filaments is very common and has been associated with the endolithic species K. dalmatica. In other populations, parallel disposition of filaments is predominant and can probably be related to an epilithic habit". This description further adds to our difficulties in discussing this genus. Many members assigned to the genus Nunduva (González-Resendiz et al., 2018) were found in several habitats in Mexico (San Agustín Bay, the same location as Kyrtuthrix, Playa Hermosa beach and Tangolunda bay), and several existing organisms were transferred to this genus: Microchaete grisea strain CCAP 1445/1 as N. britannica (brackish, estuarine, Newtown, Isle of Wright, UK and Rivularia sp. LEGE 07159 (intertidal, Portugal) as Nunduva sp. (with reservations). The populations were provisionally placed in Brasilonema based on morphology, but sequencing of both environmental and culture material assigned them to the same marine clade of the Rivulariaceae as Kyrtuthrix huatulcensis. This demonstrates the inadequacy of using a morphological approach alone. Note that K. huatulcensis is represented in the tree of González-Resendiz et al. by a collapsed clade labelled as containing 4 sequences, only one of which is available from NCBI. In our tree the Nunduva environmental and cultured organisms clearly form the species N. biania, N. britannica and N. fasciculata suggested by the authors although, unlike all other 16S rRNA sequences, that of N. britannica is short; however, the species N. kania is not separated from N. fasciculata, obviously being a member of the same species. All are related at the specific, not generic, level to Kyrtuthrix huatulcensis. Even though the authors employed 16S rRNA sequence data to position the organisms, they comment on the high sequence identity (about 98%) within Rivularian cyanobacteria and conclude that "similarity values should not be used as a primary criterion to define taxon levels in the Cyanobacteria". Our own interpretation is different: that the cluster 1.15.15, containing organisms from marine and brackish environments, is composed of a single genus with 6 species, despite wide morphological variability. This cluster contains two members of the type species of Rivularia (R. atra Roth ex Bornet et Flahault), which González-Resendiz et al. place into a different generic clade, equivalent to our species 1.15.14A. The creation of genera (here Kyrtuthrix and Nunduva) where phylogenetic data based on 16S rRNA sequences of good length show their members to separate only at the specific level, is clearly an error.

Lagosinema tenuis (Akagha et al., 2019) was isolated from the water column, Lagos Lagoon, Nigeria. The strain resembles morphologically Limnothrix planktonica but the polar structures of the cells could not be identified, being described as either gas vesicles or cyanophycin granules. Some rather simple experiments, including electron microscopic studies, would surely resolve this problem! Despite the morphological resemblance, the strain was found to be phylogenetically distinct from Limnothrix planktonica and the new generic name was proposed; we agree to such separation, Limnothrix planktonica falling into genus and Lagosinema tenuis into However, the short 16S rRNA sequences (two clones of a single strain) of Lagosinema tenuis show a separation of this "genus" from Persinema komarekii only at the specific level, sharing 97.1-97.4% 16S rRNA sequence identity. Since Persinema F.Heidari & T.Hauer, 2018 clearly has priority over Lagosinema S.C.Akagha & J.R.Johansen, 2019, the latter genus is invalid, being a species of Persinema in genus The tree shown by Akagha et al., (2019) suffers from poor taxon sampling, since it shows no potential relatives of Lagosinema except Limnothrix planktonica itself.

Members of the genus Laspinema (Heidari et al., 2018) fall into two species of genus in our tree; however, species contains strains named as L. thermale and L. lumbricale, sharing a minimum of 99.0% 16S rRNA sequence identity. The position of L. etoshii strain KR2008/49, whose short sequence was not obtained by the authors, in species, would benefit from verification.

Two strains of Leptochromothrix were described by Berthold et al. 2021) as L. engenei and L. valpauliae. It is clear from the tree shown by these authors that the two strains represent two different genera, clustering with two other new genera (Ophiophycus and Vermifilum). In our tree, the Leptochromothrix strains are well separated, L. engenei BLCC-M83 lying in genus and L. valpauliae BLCC-M82 in a different family (5.2.4) as a loner genus. The latter strain was chosen as the reference strain. It is unfortunate to choose the reference strain of the type species of a genus which falls outside the family (Vermifilaceae) that it is supposed to represent.

Leptoelongatus. The single strain, L. litoralis AP25 (Chakraborty et al., 2019), representing this genus would presumably fall into or near species 6.1.5A, which contains "unidentified LPP-group cyanobacterium" strain MBIC10012 and several other organisms from marine or saline habitats. However, this organism has been excluded from the tree because the sequence is chimeric, as found by UCHIME, with strains Calothrix sp. 96/26 LPP3 and MBIC10012 as nearest parental sequences.

Leptothoe. Members of this "genus" fall into two adjacent genera of our tree, one (strain TAU-MAC 1115) being a loner strain in family 6.1. The type species L. sithoniana, represented by the reference strain TAU-MAC 0915, is in species 6.1.6F. However, since the sequences are of poor quality (showing many mismatches in paired regions) and of medium length, their placement may not be exact. This is demonstrated by the placement of a second 16S rRNA sequence of strain TAU-MAC 1115, extracted from the draft genome sequence, into species 6.1.6B.

Meffert (1988) erected the genus Limnothrix with three species, L. planctonica, L. rosea and L. redekei, the type species. Unfortunately, 16S rRNA sequence data were not available. Although two sequences of L. rosea are available in NCBI, one (HQ916863) is too short to be of any practical value; the second, extracted from a draft genome (MRBY00000000), is named in NCBI as L. rosea but is held in the NIES culture collection as Oscillatoria rosea NIES-208. The sequence of this strain (AB003164) obtained by normal 16S rRNA-specific PCR reaction is available from NCBI as O. rosea. This strain falls into species, clearly separated from the other Limnothrix spp. strains, virtually identical to "Leptolyngbya" sp. PCC 7376 and accompanied by two species containing only strains assigned to the genus "Synechococcus". The two strains assigned to L. planktonica are found in species, and share 99.79% 16S rRNA sequence similarity. Sequences of strains assigned to the type species lie in two genera of the tree as two species: with the L. planctonica isolates and 5 strains identified only to the generic level, and with various Pseudanabaena species. The strains of these two genera are separated at the level of 88.34-88.54% 16S rRNA sequence similarity. Gkelis et al. (2005) showed a similar distribution of L. redekii strain sequences. It is therefore apparent that the genus Limnothrix requires extensive revision.

Microseira wollei (Farlow ex Gomont) G.B.McGregor & Sendall ex Kenins: the type species of Microseira falls into 2 species (A, 3 strains, and C, 5 strains) of genus 3.2.3. The species share a maximum of 97.62% 16S rRNA sequence identity. Genus 3.2.3 also contains 4 strains assigned to the genus Blennothrix, including the type species B. vermicularis, of which a 16S rRNA sequence was obtained from herbarium material by Palińska et al. (2015). Since this type species was designated in 1988 by Anagnostidis & Komárek, the genus Blennothrix clearly has priority over Microseira described by McGregor & Sendall (2015). The latter authors showed no near relatives of Microseira in their tree, the only strain of Blennothrix (PNG05-4) being distantly related and subsequently transferred to the genus Okeania (this is in a different order as genus of our tree). The genus Microseira is consequently invalid.

Minunostoc cylindricum, the type species of the genus, isolated from a wet rocky wall in China (Cai et al., 2019a) and represented in our tree by 16S rRNA sequences of all six clones of the two strains CHAB5843 and CHAB5844–1, forms genus 1.12.6. The six clones were reported as showing 16S rRNA sequence similarities of 99.6-99.9%, with which we agree from measurement of similarity with the full-length sequences. However, Cai et al. used only four of the sequences in their tree, and represented them only as a single collapsed clade. This clade is unusually long for virtually identical sequences, implying a much wider range of sequence identity. The tree was inferred with a variety of sequences of other genera, some of which are short; since the authors did not state the length and positions employed for their phylogenetic analysis, we are unable to comment further on this discrepancy; the shorter length available for comparison may have caused considerable problems. In addition, most of the sequences shown in their tree are also represented as collapsed clades, with no indication of the strains contained in each. The nearest relatives of Minunostoc appear to be a number of Trichormus strains, labeled only as "Trichormus (4 sequences)"; given the polyphyletic nature of strains currently named as members of the latter genus, the true position of Minunostoc in their tree is impossible to establish. Due to poor taxon sampling, many other potential relatives of Minunostoc were omitted from the tree shown in the publication, which is evidently inaccurate.

Myxacorys. Pietrasiak et al. (2019) provided 13 new 16S rRNA sequences, representing 7 strains, of organisms isolated from desert soil in North and South America, and named them as two species: M. californica and M. chilensis. Unfortunately, the sequences are all short, starting only at E. coli position 359. The authors transferred into this genus a large number of strains named in NCBI as Pseudophormidium sp., Leptolyngbya sp. or simply "cyanobacterium". The tree of Pietrasiak et al. appears to show convincingly the presence of the two species, claimed to show a maximum of 98.3% sequence identity. However, our sequence alignment, maintained in ARB and based on secondary structure, produces a generic cluster in which all strains show a minimum identity of 98.99%, even taking the most divergent strains of the tree of Pietrasiak et al., and therefore represent a single species. We show only 8 of over 60 sequences, falling into genus 6.3.7 of our tree. The authors state that "This study represents the first time cyanobacterial species can be diagnosed primarily by 16S rRNA gene sequence data", which is clearly incorrect. The two species were further justified by "consistent sequence divergence in Helix 21 of the 16S rRNA molecule" and "the invariant sequence difference in Helix 21 is what provides a diagnostic criterion for the two species", whereas we find a single compensatory base change in that region (which is, in fact, helix 22) in most, but not all, strains. Sequences of strains named as a third species, M. almedinensis, were subsequently deposited in NCBI (MK927050-MK927052); these show 96.43-97.12% 16S rRNA similarity to the reference strain of the genus, ATA2-1-KO14, and 99.01% to each other, thus probably justifying their separation at the specific, not generic, level. However, this is based on the short comparable sequence region available for these strains; the provision of longer sequences may modify this decision.

Nodularia: genus 1.3.1 of our tree, whose members show a minimum 16S rRNA sequence identity of 97.79%, is divided into six species. However, species 1.3.1A contains strains which carry different specific epithets: N. armorica, N. cf. harveyana, N. harveyana, N. moravica, N. sphaerocarpa and N. spumigena (the type species); isolates named as N. harveyana are also found in species 1.3.1C (4 strains), 1.3.1D (2 strains) and 1.3.1E (1 strain), with a strain of N. cf. harveyana lying in species 1.3.1F. It would appear wise to reclassify the Nodularia isolates into the six species defined phylogenetically.

The separation of Oculatella mojaviensis and O. atacamensis (in genus 6.2.1) as distinct species, as proposed by Osorio-Santos et al. (2014), is not supported by 16S rRNA sequence analysis, since their members share at least 99.4% identity (species 6.2.1D). Since the other proposed species (O. subterranea, O. kauaiensis, O. cataractarum, O. coburnii and O. neakameniensis) indeed separate at the specific level (identity values of 97.6% to 98.2%), this cannot be ascribed to lack of resolution of 16S rRNA analysis. However, the deposited sequences are extremely short, starting at E. coli position 359. The longer sequence of O. leona also places this organism in species 6.2.1D, since it shares 99.86% identity with another long sequence, Oculatella sp. FACHB-28. O. dilatativagina strain ACT687, again represented by a long sequence, shares only 97.89-97.96% identity with these strains, representing a different species (6.2.1H in our tree). These two long sequences were published by Becerra-Absalón et al. (2020), who recognized that they shared 98.9% identity, higher than the value (98.7%) employed for species discrimination. However, they considered that "Above this threshold, the percent identity is not useful for taxonomy", referring to the cutoff value of Yarza et al. (2014). The latter authors do not appear to have made this statement! O. leona strain ATE710 showed high 16S rRNA identity values with the other members of species 6.2.1D, from which it was separated in a tree inferred from an alignment of the ITS regions, which we consider to be highly ambiguous, and by differences in the D1-D1′, V2 and Box-B structures. This surely invalidates the use of ITS secondary structure in species determination.

The genus Odorella, with O. benthonica as type species, was erected by Shalygin et al. (2019a) based on the single strain CalAq792, originally isolated from the California Aqueduct near Los Angeles as a Hyella sp. The full-length 16S rRNA sequences of this strain lie in species of our tree, the two clones showing 99.73% 16S rRNA sequence identity. We find 99.67-99.83% 16S rRNA sequence identity with the nearest relative Pleurocapsa minor UAM 388 (JQ070059), similar to Shalygin et al. 2019a (who give the value for only one clone), although we note that this strain is represented by a relatively short sequence; 97.16-97.33% with P. minor HA4230-MV1 clone 2B and 97.58-97.76% with P. minor HA4230-MV1 clone 2C; and 98.45-98.62% with the short sequence of P. fuliginosa HA4302-MV1 (JN385285). Shalygin et al. 2019a find different identity values between these strains, perhaps indicating poor sequence alignment. Our values indicate that strains CalAq792 and UAM 388 are members of a single species, whereas strain HA4302-MV1 is placed into a different species ( of the same genus. P. fuliginosa HA4302-MV1 is the neotype strain and type species of the genus Pleurocapsa, proposed by Shalygin et al. 2019b in a later paper that does not mention Odorella. Shalygin et al. 2019a show that Odorella benthonica CalAq792 differs in 14 of 64 positions (including a 3 nt indel) in the ITS D1-D1′ region from that of P. fuliginosa HA4302-MV1 but only 4 positions from an unnamed clone of P. minor HA4230-MV1 (KC525080 or KC525081) both of which are in our species The authors conclude that "Even though secondary structures of the conservative domains of the ITS region of O. benthonica were quite similar to those of P. minor (Strain HA4230, GenBank KC525080), overall length of the ITS region and sequences within V2, Box-B, and V3 varied substantially", but do not give a detailed analysis. Since the ITS is subject to fewer mutational constraints than the 16S rRNA gene, such a low number of mutational changes cannot justify the separation of strain CalAq792 from the other isolates of Pleurocapsa of cluster; this entire cluster is named as Pleurocapsa sensu stricto in view of the neotypification of P. fuliginosa HA4302-MV1, thus invalidating the "genus" Odorella.

Two identical 16S rRNA sequences of strains Parakomarekiella sesnandensis coi00088998 and coi00088999 are shown in a tree of Soares et al. (2020). They are identical to Nostoc sp. strain 5N-02c and 99.76% identical with Nostoc sp. strain 1b-05. They show 98.31-98.47% sequence identity with six strains of Komarekiella atlantica, and therefore represent only a species of Komarekiella. The organisms are shown in our tree as species 1.8.5A (Komarekiella) and 1.8.5D (Parakomarekiella). The authors found similar 16S rRNA similarity values, but created the new genus for reasons that escape comprehension; the genus Komarekiella again has priority.

Pegethrix: the four species described by Mai et al. (2018) for this genus fall, on the basis of 16S rRNA sequence identity, only into two specific clusters, sharing 98.6-98.8% identity. Species 6.2.5A contains both P. bostrychoides (the type species) and P. olivacea, showing 99.0-100% identity; species 6.2.5B contains P. convoluta and P. indistincta (99.8-100% identity within the species). The sequences are all short, except for those obtained by other authors as Leptolyngbya frigida and transferred into the genus by Mai et al. The authors state that "Sequence identities of the 16S rRNA gene sequences in the genus do not provide evidence of species separation in this genus..... However, the phylogeny separates the species fairly well". We do not understand how the "phylogeny" (tree) can separate species that are not separable on the basis of sequence identity.

Persinema komarekii (Heidari et al., 2018), isolated from a radioactive thermal spring, Iran, and initially deposited in NCBI as Limnothrix planktonica, is sufficiently different from other filamentous cyanobacteria as to form a new genus within order 5.4.2. However, this is based on two cloned sequences (numbers 1 and 3) of a single organism; clone 3 is a chimera, the first 451 nt showing long inserts (totalling 60 nt) in helices 6, 10, 11 and 18, corresponding to various Paenibacillus spp. as shown by BLAST (100% identity, 100% query cover, E-value 0.0) and by UCHIME (breakpoint between positions 461-463). The almost full length 16S rRNA sequence of clone 1, starting at E. coli position 41, places the organism into genus The strain does not contain gas vesicles and is benthic, unlike Limnothrix planktonica. Surprisingly, an organism named as Klisinema sp. Espan (Koehler et al. unpublished) shares 99.59% 16S rRNA sequence identity with clone 1 of P. komarekii. Since a strain of K persicum (Heidari et al., 2018) clearly falls into genus of our tree, the Espan strain must be misidentified. We have commented above concerning the priority of the genus Persinema over Lagosinema, whose strains are both members of a single genus.

Turicchia et al. (2009), in a study focusing on oscillatorialean morphotypes, erected the genus Phormidesmis, mainly using "results from 16S rRNA gene sequencing as a basic criterion for the final taxonomic evaluation", with P. molle as type species and isolate 3CC04S05 as the type strain. Although this was a comprehensive study, the 16S rRNA sequences were not deposited in NCBI; the phylogenetic results cannot, therefore, be verified. The tree of the authors shows the nearest relatives of their five Phormidesmis strains to be Leptolyngbya frigida strain ANT.LH64B.1 (AY493577) and Phormidium autumnale strain UTCC471 (AF218371), both of which fall into genus 6.3.30 of our tree (named as Stenomitos spp.) and share 98.67% 16S rRNA sequence identity. The latter organism was initially submitted to NCBI as Pseudanabaena tremula (Casamatta et al., 2005) and is currently listed as Stenomitos tremulus, having been transferred into this genus by Miscoe et al. (2016). Our tree shows 31 strains assigned to the genus Phormidesmis by varied investigators. They fall into 9 generic clusters of two families (6.1.9, 6.1.10, 6.1.11, 6.3.1, 6.3.18, 6.3.19, 6.3.20, 6.3.23, 6.3.30), with two appearing as loner genera in family 6.3. Genus 6.3.1 is Leptolyngbya sensu stricto, containing the 16S rRNA sequences of 6 strains assigned to the type species L. boryana; this genus also contains the sequence of P. molle SAG 26.99. The six L. boryana strains are identical in 16S rRNA sequence, and share 99.93% 16S rRNA sequence identity with the P. molle strain. We therefore have a situation where a single genus contains strains assigned to the type species of two different genera, which is clearly nonsense.

Three strains assigned to the genus Porphyrosiphon by Yim (unpublished) fall into two genera in our tree: 3.3.2 (with strains named as Trichocoleus delicatulus and Coleofasciculus chthonoplastes), 3.3.3 (with a Trichocoleus sociatus strain) and one strain (Fukuoka et al., unpublished) falls into an unnumbered genus within family 2.1. In the absence of publications, we cannot comment further, except to note that the 16S rRNA sequences of Porphyrosiphon sp. strain ACKU_Y_DB3 in genus 3.3.2 and P. notarisii KHS17 (family 2.1) are extremely short.

Strains assigned to the genus Potamosiphon as the single species P. austaliensis by McGregor & Sendall (2019) fall into two well-separated species (3.2.6A and 3.2.6B) in our tree, showing only 98.1-98.4% 16S rRNA sequence identity; the sequences are almost full length, starting at worst at E. coli position 30 and complete at the 3' terminus. However, the sequence MH047863 (species 3.2.6B) differs from the others at 25 sites located exclusively in the 5-prime region up to position 166 (E. coli 227); exclusion of this region from p-distance calculation results in the sequences showing 99.7% identity, explaining the proposal of only a single species by McGregor & Sendall (2019) and showing the importance of using full-length sequences in tree inference. The short 5-prime region of sequence MH047863 shows higher identity (94.5%) to Microseira wollei YC1109 clone A45, described earlier by the same authors (McGregor & Sendall, 2015), than to the other Potamosiphon strains (84.9-86.7%), suggesting that this may be a chimeric sequence. Further investigation is therefore required.

Strains identified as Prochlorococcus marinus fall into 5 different specific clusters (7.1.3A-E); we show only full-length 16S rRNA sequences extracted from sequenced genomes. It is evident that 4 of the 5 clusters require different specific epithets. A start in this direction has been made, described in more detail in the document comparing 16S rRNA sequence identities with genomic metrics. Excluding these Prochlorococcus spp., most "OMF" strains cluster as 18 species on the basis of 16S rRNA sequence identity and DDH, as described in the same document. As in many other parts of the tree, these species contain strains with different generic epithets, which require renaming.

Rexia: This genus was erected by Casamatta et al. (2006). In a tree without scale marker, these authors show that a strain of the new species R. erecta clusters with Coleodesmium strain MC-JRJ1. The R. erecta strain was given as CAT4-SG4, but in NCBI the sequence of this strain is attributed to Scotta Hentschke et al. in a much later publication (2017), strain CAT 1M being attributed to Casamatta et al. The strain reportedly showed 96.2% 16S rRNA sequence identity with both Tolypothrix distorta (strain not stated, presumably strain SEV-2-5-Ca) and Spirirestis rafaelensis (strain again not stated, presumably strain SRS70), suggesting it to represent a different genus. We are unable to reproduce this result, and note that the sequence of R. erecta strain CAT 1M is very short (966 nt, claimed to be longer in the publication). Both sequences of R. erecta are shown in species 1.14.1G of our tree, and are members of the large genus Tolypothrix sensu stricto that contains all 17 known strains of T. distorta, the type species.

The genus Roholtiella, erected by Bohunická et al. (2015), was divided into 4 species, based on short 16S rRNA sequences starting at best at E. coli position 359. However, 3 of these (R. edaphica, R. fluviatilis, R. bashkiriorum) fall into the single species 1.8.3A, showing 98.8-99.3% 16S rRNA sequence identity and are therefore not separated phylogenetically. These strains show 98.5% 16S rRNA sequence identity with strain Tolypothrix PCC 7415, which is our lower limit for species distinction. A fourth species, R. mojaviensis, lies in our species 1.8.3B. One further strain, obviously mis-identified, is in family 1.3. Furthermore, distinction of Roholtiella and Mojavia (again represented only by short sequences) at the genus level, despite their morphological differences, is not supported: their members share a minimum of 97.2% 16S rRNA sequence identity and the Mojavia isolates fall into species 1.8.3C. This conclusion escaped Bohunická et al. (2015), who omitted Mojavia from their study. However, the available sequences of Mojavia are short; if the identity values above are confirmed with longer sequences, then Mojavia Řeháková & Johansen, 2007 would have priority over Roholtiella Bohunická, Pietrasiak & Johansen 2015.

The almost full length 16S rRNA sequences, of high quality, of Sinocapsa zengkensis provided by Wang et al. (2019) are a welcome addition and form genus 3.1.24. However, the sequences are those of multiple clones from a single strain, CHAB 6571 (deposited in NCBI as CHAB 6751). The different clones share 98.86% 16S rRNA identity (calculated from our alignment) and were found by the authors to exhibit 94.13% identity with the nearest relative, Aliterella antarctica strain CENA408, clearly showing a difference at the generic level. Although the naming of a new genus on the basis a single strain may be acceptable, the creation of a new family (Sinocapsaceae) is unwise, particularly in view of the high identity value given above and the subjective nature of comparisons of ITS secondary structure models. The sequences place this organism, in our tree, firmly into family 3.1.

Genus of our rRNA tree contains 10 strains assigned (Samylina et al., 2021) to the genus Sodalinema. Members of this genus were named as S. gerasimenkoae, S. orleanskyi, S. stali and Sodalinema sp., in addition to the existing S. komarekii. However, all strains show 99.22-99.66% 16S rRNA sequence identity, and are therefore members of a single species. These identity values are similar to those (99.0-100%) shown by the authors, and the minor differences in ITS sequence do not justify specific separation.

The genus Sphaerocavum was erected by Azevedo and Sant' Anna (2003) to include two strains isolated from reservoirs in Brazil that differed from members of the genus Microcystis by having only two planes of cell division and hollow colonies; two new combinations were proposed, S. leloupii and S. microcystiforme, although no phylogenetic evidence was provided. Rigonato et al. (2018) showed that 17 strains of "Sphaerocavum" are indistinguishable phylogenetically from strains of Microcystis aeruginosa. We show 5 such strains in the tree; they share 99.9-100% 16S rRNA sequence identity amongst themselves and a minimum of 99.6% with, for example, M. aeruginosa PCC 7941 (genus These strains should therefore be transferred into the genus Microcystis, as suggested by Rigonato et al. (2018), who proposed the name Microcystis brasiliensis and that the description of this genus be emended. While we agree with such emendation in order to include organisms dividing in two planes, we cannot accept the addition of another specific epithet to a cluster that represents, in fact, a single species on the basis of 16S rRNA sequence identity.

Amphiheterocytum and Sphaerospermopsis are problematic genera. Sphaerospermopsis was erected by Zapomělová et al. (2009) as Sphaerospermum, the name being corrected subsequently (Zapomělová et al., 2010), the type species being S. reniformis. Two strains assigned to S. reniformis are found in species 1.2.3A in our tree, but are not distinguished by 16S rRNA studies from strains named as S. aphanizomenoides, S. eucompacta, S. kisseleviana or S. oumiana. Strains named as S. aphanizomenoides additionally occur in species 1.2.3C with S. eucompacta and even a different genus (species 1.2.7B, a draft genome sequence). These various species require re-naming to be congruent with phylogenetic data. Species 1.2.3D contains two strains, sharing 99.38% 16S rRNA sequence identity, assigned to the new genus Amphiheterocytum, created by Sant'Anna et al. (2019) with A. lacustre as the single and type species; however, the tree shown by these authors reveals only a small difference between this genus and some members of Sphaerospermopsis, and 16S rRNA identity values demonstrated that the two "genera" cannot be separated. We find a minumum 16S rRNA sequence identity of 97.26% between all of these strains, similar to that reported by Sant'Anna et al. These authors chose to ignore the phylogenetic evidence, with the excuse that the standard cutoff values for species and genus delineation do not apply to the Nostocales, and stated that "no bootstrap or posterior probability values were achieved to justify their grouping in the same genus". The use of bootstrap values to delineate genera is not normal phylogenetic practice. The ITS secondary structures presented were inconclusive. It is apparent, therefore, that Amphiheterocytum is a species of Sphaerospermopsis, and cannot be separated at the generic level.

Spirirestis. This genus was erected by Flechtner et al. (2002) on the basis of two isolates (SRS6 and SRS70) with S. rafaelensis as type species and strain SRS70 as reference strain. The strains differ from all members of Scytonema and Tolypothrix in producing tightly-coiled filaments. We show only one of 3 clones of the reference strain, which lies in species 1.14.1K together with 1 strain sequenced by Vaccarino et al. (unpublished). In the tree of Flechtner et al., the 3 S. rafaelensis sequences have, as nearest relatives, 2 sequences of a Tolypothrix distorta strain said to be SEV2-5-BG (but in NCBI as SEV2-5-2Ca), and were reported to share 98.6-99.1% 16S rRNA sequence identity. Such a high level of identity firmly establishes the strains as members of a single genus, despite their clear differences in morphology. Genus 1.14.1 of our tree is Tolypothrix sensu stricto, containing all 17 known strains of the type species T. distorta and the isolates of Coleodesmium, Dactylothamnos, Hassalia, Kryptousia and Rexia described elsewhere on this page. None of these "genera" are separable phylogenetically.

Spirulina: S. major (the type species) occurs only in genus A second species, S. subsalsa, is seen in our tree as two generic entities (genera and, with an additional strain (probably misidentified) falling with the S. major strains, and 3 sequences of strains with no specific epithet are found with two strains of Halospirulina tapeticola (the type species of this genus) in genus A further strain (SAG 256.80), possibly misidentified as S. laxissima, lies remotely as species 6.1.1O. It is clear that all "Spirulina" species other than S. major should be transferred to new genera, those falling with Halospirulina being members of that genus.

Stenomitos: five named species of this genus, S. frigidus, S. hiloensis, S. kolaenensis, S. rutilans (type sp.) and S. tremulus, all except the first being represented by a single strain, lie in species 6.3.30A, B, C, E, F and H. Species 6.3.30A contains 5 strains named as Stenomitos sp., S. hiloensis, S. kolaenensis and S. tremulus which, in sharing 99.01-99.28% 16S rRNA sequence identity, must be considered as members of a single species. The remaining species shown in our tree, except for species 6.3.30D, appear to be well defined; we approve the use of quotation marks to discriminate strains assigned to S. frigidus senso stricto (species 6.3.30C) from those with the same name that require further revision. However, this genus also contains an organism named as Tildeniella nuda (strain Zehnder 1965/U140) in species 6.3.30D, described in the next paragraph, that shows 98.71% 16S rRNA sequence identity with, for example, S. frigidus ANT.LH53B.2. Thus a single genus, defined by 16S rRNA sequence identity, contains strains named as members of two genera, clearly an unfortunate situation. The strains and specific assignments are described by Shalygin et al. (2020) who, curiously, omitted the Tildeniella strain from their tree. We have renamed the organisms in our tree to match those given by Shalygin et al., with the previous names given in parentheses. The accuracy of the study of Shalygin et al. is called into doubt by their estimate (in the species description) of 97.75% 16S rRNA identity between the S. kolaenensis and S. tremulus strains; in our accurate alignment, the value is 99.01%. The estimate of 97.75% is not supported by their tree, which shows the S. hiloensis, S. kolaenensis and S. tremulus strains to be closely related; in the similarity matrix, they give a value of 99.0%. Using the incorrect genus-level cutoff of 94.5% (Yarza et al., 2014), the authors state that "16S genetic identity alone does not provide clear evidence that...the genus Stenomitos is separate from Neosynechococcus". This separation is easily achieved by using our more realistic cutoff of 96.5-96.9%; the full length 16S rRNA sequences of strains N. sphagnicola CAUP A 1101 and S. frigidus ANT.LH53B.2 show only 96.37% identity.

Tildeniella: two members of this genus, described by Mai et al. (2018) and represented by 3 short sequences, fall into the same (type) species, T. torsiva, with 99.9-100% 16S rRNA sequence identity (monospecific genus 6.2.20). However, a fourth strain, named as T. nuda, shows only 92.2-92.3% identity to the members of the type species, and falls into a distinct genus as a member of species 6.3.30D, as described in the paragraph immediately above. We note that Mai et al gave a sequence identity value of 99.1%, and are not sure how they arrived at that figure, particularly in view of the separation of T. nuda from T. torsiva in their tree. It is also surprising that these authors show one of two sequences of different operons of T. torsiva strain UHER 1998/13D clearly separated from the other, despite their high sequence identity.

Toxifilum. Although there is no doubt about the thorough study of Zimba et al. (2017), who created this genus for a single strain isolated from a marine habitat, the choice of name ("named for the production of anabaenopeptin and subsequent mortality of mysids") may be unfortunate. Anabaenopeptin G, the toxin found, has not been observed in any of the related strains in genus 6.1.14 (Leptolyngbya IV), all of which share 100% 16S rRNA sequence identity despite the diversity of their habitats (marine, estuarine, freshwater lakes). Although Lopes et al. (2012) successfully found the related toxins, anabaenopeptin A and D, in strain Nostoc sp. LEGE 06077, demonstrating the efficiency of their techniques, none were observed in the two members of this genus (strains Leptolyngbya sp. LEGE 06070 and LEGE 07319) studied. We have therefore refrained from renaming these strains and their generic cluster, pending further study.

Trichotorquatus: Of 12 members of this "genus" shown in the tree, 10 fall into monospecific genus 6.2.19. Despite their high 16S rRNA sequence identities, which place them into a single species, they carry 4 different specific names, including T. maritimus, the type species (represented by 4 clones). A single strain, Trichotorquatus sp. 5, represented by two sequences, lies in genus 6.2.26, and the final strain (T. andrei CMT-3FSIN-NPC33) represents a singleton genus within family 6.2. The species T. andrei therefore appears in two different genera; curiously, strain T. andrei CMT-3FSIN-NPC33 is not shown in the 16S rRNA phylogeny of the authors. Since the 4 species were separated on the basis of their ITS sequences, the occurrence of the same species within two genera indicates the inadequacy of this approach.

Assorted cave dwellers

In the case of Loriellopsis cavernicola (Lamprinou et al., 2011), a cave rock scraping was examined for morphological features of the dominant organism, which was described as a heterocystous, true-branching cyanobacterium. A 16S rRNA gene sequence was obtained from the bulk sample, rather than from an isolated strain, and assumed to be that of the dominant organism. There is no certainty that the sequence was that of the desired organism. It was compared to a very limited number of sequences of true-branching heterocystous cyanobacteria, an uncultured hypolithic organism and several heterocystous cave dwellers, in a phylogenetic tree with extreme taxon sampling error. The nearest relative included in the tree was uncultured cyanobacterium clone HAVOmat106, which in our tree is found outside of the heterocystous clade (order 1) in family 3.1. A number of sequences have subsequently been found to show high identity with the original L. cavernicola LF-B5: 3 sequences of uncultured hypoliths from marble, quartz and carbonate rocks in the Mojave desert where the most abundant organisms were Chroococcidiopsis-like cyanobacteria (99% sequence identity to LF-B5 by NCBI blast, Smith et al., 2014) and one sequence (OTU 28) from an Arctic soil crust predominantly containing Leptolyngbya spp. (98.6% sequence identity by the Neighbor-joining method, Pushkareva et al., 2015). Finally, Fig. 2 of Ragon et al. (2012) shows that the L. cavernicola LF-B5 sequence clusters loosely with Chroococcidiopsis thermalis PCC 7203 and Chroococcidiopsis sp. SAG 2025. This is confirmed in our tree, where the uncultured "Loriellopsis" sequences are found in species 3.1.15A and OTU28 in species 3.1.15B, as a sister group to the Chroococcidiopsis sp. strains in genus 3.1.14. Surprisingly, Kilgore et al. (2018) described the sequence of an uncultured organism as that of Loriellopsis sp., despite low (92.8%) identity and very loose grouping with the sequence LF-B5; this was shown in their 16S rRNA tree and deposited in NCBI (KY924323) under that name. This sequence is found in family 6.3 in our tree, clearly unrelated to the others. Since “Loriellopsis cavernicola” is one of only two only supposed heterocystous representatives to fall completely outside of the Nostocalean clade (to which the genus was assigned by Lamprinou et al. despite sharing very low sequence identity with other members) when additional sequences are included in the tree, we conclude that the sequences assigned to “Lorelliopsis” represent a cosmopolitan non-heterocystous organism, probably Chroococcidiopsis-like, found in caves, as desert hypoliths and in Arctic soil crusts (and possibly other habitats). This genus should therefore be renamed (since it is based on the genus Loriella) and redefined. A second organism, incorrectly named as Scytonema millei (strain VB511283, QVFW00000000), also lies outside of the Nostocalean clade and is described in detail in the genome metrics document.

Sequences assigned to representatives of the genus Geitleria were obtained by Kilgore et al. (2018) from the limestone walls of a cave in The Great Smoky Mountains National Park, Tennessee, named as G. appalachiana and assigned to a new family, the Geitleriaceae. This genus is unusual in that its members show true branching, but no heterocysts are observed. The organisms were not cultured, but purified as much as possible by careful cleaning of filaments by micromanipulation; however, it is possible that the thick sheath prevented DNA extraction and that the sequences represent, in fact, an unknown contaminant. In the phylogenetic tree shown in this publication many clades are collapsed, being labelled with e.g. "Gloeotrichia/Nostoc/Desmonostoc/Trichormus/Goleter etc. 34 OTUs". No indication of the sequences employed for tree inference is given, and thus no verification of the tree is possible. The five short 16S rRNA sequences show 99.1-100% identity, and are therefore members of a single species. A simple study by NCBI blastn of the sequence of clone 3 shows as nearest relatives (identity 95%, 100% query cover, E value 0.0) 2 clones of Brasilonema sp. PTS-MK70 (KY365509-10), 3 strains of Calothrix (HF678500, HQ847563, CP011382) and Phyllonema tangolundensis strain C76-M18-K59 (MG754003), none of which are true-branching organisms. These may be found in our tree in genera 1.19.4, 1.15.6 and 1.15.17, respectively. Note that Phyllonema does not appear to be included in the tree of Kilgore et al.. Although the blastn results show some variabilty, our tree inference using the 16S rRNA sequences clearly places the uncultured Geitleria as genus 1.13.5, firmly fitting into family 1.13 as a sister genus to Iningainema pulvinus, Petalonema (strain HA4277-MV1) and Scytonema stuposum, with which its members share 92.4-94.0% 16S rRNA sequence identity. The latter organisms are all capable of producing heterocysts, and are placed in the family Scytonemataceae; they do not appear to be included in the tree of Kilgore et al., even though the Petalonema strain was described in a paper which had one author in common. The creation of the family Geitleriaceae would therefore appear inappropriate, and the sequences may not represent the genus Geitleria.

Toxopsis was erected by Lamprinou et al. (2012) with the single species, T. calypsus, to include a type of cave-dwelling heterocystous cyanobacterium. Sequences of five clones of only a single strain (PLF) were provided, are virtually identical and nearly full-length. Two are found in genus 1.14.3 in the tree. Members of this genus resemble closely in morphology members of Godleya alpina (Novis & Visnovsky, 2011), isolated from a high altitude mountain site in New Zealand, although these were not included in the study of Lamprinou et al. The sequences of strains assigned to these genera share 98.6-98.9% 16S rRNA sequence identity, placing them as a single species of a single genus, for which the name Godleya clearly has priority. T. calypsus and G. alpina are both type species, the former obviously being invalid. We note that members of genus 1.14.3 are probably common at high altitudes, since the uncultured members were, without exception, amplified from high mountainous regions in Nepal and Peru.

Spelaeonaias floccida, isolated from a cave, was described as resembling members of the genus Symphyonemopsis by Lamprinou et al. (2016), but with the subaerophytic growth of the ensheathed hormogonia as the autapomorphic feature. There is no guarantee that this feature is sufficient to merit distinction at the generic, rather than specific, level. Indeed, the single 16S rRNA sequence (KY018918) for this organism falls into species 1.19.5B, with that of Symphyonemopsis sp. strain VAPOR1 (AJ544085), also isolated from a cave, lying in species 1.19.5A. Spelaeonaias therefore appears to be a later synonym of Symphyonemopsis Tiwari, G.L. & Mitra, A.K. (1969 '1968'). However, the sequence provided is extremely short (304 nt); trees produced from alignments containing such short sequences are undoubtedly inaccurate, which would explain the failure of Lamprinou et al. (2016) to find any close relationship to Symphyonemopsis. We added the sequence to our backbone tree, built only with long sequences, as described above, and are fairly convinced of the accuracy of its position. Further decision regarding the invalidation of the genus Spelaeonaias should nevertheless await the publication of longer sequences.

Members of the genus Iphinoe, first described by Lamprinou et al. (2011) from caves, similarly fall into species 1.19.5A in our tree; the 16S rRNA sequence of a member of the type species (I. spelaeobios) was sufficiently long (1446 nt) to be included in the backbone tree. The available 16S rRNA sequences show 99.1-99.3% identity with that of the Spanish cave-dweller Symphyonemopsis sp. strain VAPOR1 (AJ544085) over E. coli positions 8-1421, and 98.7-98.9% with the sequence of the uncultured clone HAVOmat 34 (EF032787), from a cave in Hawaii. These organisms are therefore all members of a single species, and Iphinoe is therefore a later synonym of Symphyonemopsis Tiwari & Mitra (1969 '1968').

Other "genera" isolated from caves are listed below.

  • Chalicogloea cavernicola CCALA 975 (JQ967037), described by Roldan et al. (2013) from a Spanish cave, should not have generic status, being found only as species, with morphologically similar Gloeocapsa strains from the PCC (species and Chroococcus strains (species as close relatives. All 16S rRNA sequences described here are nearly full length.

  • From Hawaii, USA, Miscoe et al. (2016) named four isolates at the generic level, all of which are found as species of existing genera:
    • Trichormus sp. HA7287_LM1 (represented by a short sequence, KF417428) is species 1.3.7B among three species of Halotia, whose 16S rRNA sequences are almost complete.
    • Trichormus anomalus HA4352_LM2 (also represented by a short sequence, KF417426) falls as species 1.12.1B within the genus "Nostoc" IX.
    • Pelatocladus maniniholoensis HA4357-MV3 clone Ftcon (JN385293, full-length sequence) lies in species 1.15.1K together with Hapalosiphon hibernicus.
    • Stenomitos rutilans HA7619_LM2, represented by a short 16S rRNA sequence (KF417430) is found as species 6.3.30H with seven other species (A-G) containing organisms with the generic names Tildeniella and (previously) Leptolyngbya, Phormidesmis and Phormidium. The full-length sequence of Stenomitos sp. strain Ru-0-2 (Mikhailyuk et al., unpublished; MH688849) lies in species 6.3.30A. The specific names appended to members of the genus Stenomitos, together with the transfer of numerous organisms into this genus, are described in the "problematic organisms" section of this page.

    "Loner" sequences

    None of the "loner" sequences show near relatives by BLAST (21/05/17). They may form new clusters when more sequences become available.

    Problematic Type Species

    Particularly disturbing is the distribution in the tree of type species listed in AlgaeBase ( under the Botanical International Code of Nomenclature (ICN).

    Although many such organisms fall correctly into a single generic and specific cluster of our tree, others are found as different generic or specific entities. Such dispersal of the sequences may result from mis-identification, or may indicate intragenomic heterogeneity in different rrn operons of a single strain. Some examples are given below, the organisms being arranged, as far as possible, in their alphabetic order; the names of the investigators listed are only for illustrative purposes and are not exhaustive. Reference strains, where designated and sequenced, are included. A complete list of type species and reference strains is available here, for the genera listed on this site.

    Affixifilum. Lefler et al. (2021) described the occurrence of Affixifilum spp. and Neolyngbya spp. in South Florida, USA. The Affixifilum strains form a coherent generic cluster ( in our rRNA phylogeny), sharing 99.13-99.90% 16S rRNA sequence identity. The strains of A. granulosum show 99.81-99.90% identity, and those of the type species, A. floridanum, 99.52-99.23% identity. Similar identity values were reported by the authors. All strains are therefore members of a single species, despite their different specific names. Although the authors were able to separate the two species on the basis of pair-wise distances and secondary structures of the ITS region, the 4 base differences in secondary structure between the species are limited to the terminal unpaired loop of the D1-D1′ helix, Box-B and V3 being unchanged. Since unpaired regions are relatively free to mutate, and alignment of the ITS is prone to error, we do not consider this as sufficient evidence for specific distinction. Sequence length (nt): 1036.

    Anabaena oscillarioides Bory ex Bornet & Flahault: members studied by Rajaniemi et al. (2005) and Gupta et al. (2012) occur in species 1.1.1L (2 strains), species 1.2.5A (1) and species 1.10.1B (1). Members of the 3 different genera share 93.27-95.33% 16S rRNA sequence identity; those within genus 1.1.1L are 99.86% identical. Sequence length (nt): 1417-1462.

    Amazonocrinis nigriterrae: described by Alvarenga et al. (2021) as the type species of the genus; strains assigned to this type species lie in 4 species of genus 1.9.4. The need for further division of this species was easily seen in the tree provided by the authors. Sequence length (nt): 1364-1489.

    Aphanizomenon flos-aquae Ralfs ex Bornet & Flahault; type designated in Kirchner, O. (1898): strains studied by Gorham, Rippka, M.M. Watanabe or Sivonen fall into two species (1.1.1A and 1.1.1L) that show a maximum of 98.61% 16S rRNA sequence identity. Within each species, the strains exhibit 99.44-100% sequence identity. The four strains and one metagenome of species 1.1.1A share 99.79-99.93% 16S rRNA sequence identity with Dolichospermum flos-aquae strain 37 (AJ293102). This strain was originally named as Anabaena flos-aquae, and transferred to Dolichospermum as D. flos-aquae, the type species of the genus, in later revisions. Together with one strain of Aph. flos-aquae, strain 37 was identified by K. Sivonen who demonstrated the taxonomic confusion in this group (Gugger et al., 2002), which is further documented in the section "Problematic organisms" on this page. Sequence length (nt): 1249-1487.

    Arthronema africanum (Schwabe & Simonsen) Komárek & Lukavsky: data from publications by Friedl or Turner show that this taxon falls into genera 6.3.1 (1 strain) and 6.3.17 (2 sequences of a single strain), with only 88.82% 16S rRNA sequence identity). Sequence length (nt): 1444-1460.

    Arthrospira jenneri (type species) and the genus Limnospira (type species L. fusiformis) deserve extended mention. Although deposited in NCBI as sequences of "Uncultured Arthrospira sp.", obtained as environmental material from a Polish reservoir by Nowicka-Krawczyk et al. (2019), the organisms are named as the type species of Arthrospira in their publication. Tree inference showed 17 highly similar sequences to fall as a sister taxon of the Planktothrix cluster. The Arthrospira spp. strains that are normally found in that position appeared to fall elsewhere in the tree, and were assigned to the new genus Limnospira, with SAG 85.79 (CCALA 026, UTEX 2340) as reference strain. Arthrospira sp. PCC 8005 was renamed as the reference strain of L. indica. However, numerous publications demonstrate that all known Arthrospira isolates form a sister group to Planktothrix, as also shown in our tree (see Arthrospira spp. in genus, Planktothrix spp. in genus, where these genera are closely related and members of the same order). Our alignment of the 16S rRNA region (ITS excluded) of the sequences of the uncultured A. jenneri followed by tree inference places them into a different order, as genus (note that we show only 2 of the 17 sequences available). This genus also contains 2 sequences (which show 99.86% 16S rRNA identity) of the single strain Microcoleusiopsis ganfuensis CHAB4138, the type species, related to the A. jenneri sequences by 98.71-99.52% 16S rRNA identity. We are unable to understand how Nowicka-Krawczyk et al. (2019) produced the tree shown in their publication, unless the 1800 bp fragment that they used included the ITS region of certain strains, absent from others. That this is likely is confirmed by the fact that the normal length of a 16S rRNA gene is around 1490 nucleotides; comparison of sequences lacking certain regions violates the principles of phylogenetic analysis and leads to spurious results. The study does, however, show that organisms resembling Arthrospira occur in the environment but have not yet been correctly named. The genus Limnospira, incorrectly replacing Arthrospira, should be invalidated, being based only on uncultured organisms, and on the grounds of faulty phylogenetic tree inference and the close relationship with the strains of M. ganfuensis. If retained, the genus would have priority over Microcoleusiopsis, described two years later (Geng et al., 2021); note that these authors excluded near relatives from their phylogenetic tree (a simple NCBI blast returns the uncultured organisms with the highest identity). Sequence length (nt): 1335 (A. jenneri), 1494-1496 (M. ganfuensis).

    Cephalothrix komarekiana C.F.S. Malone et al.: identified by Fiore, Genuario or Malone, this taxon is found in 2 species (A and B) of genus 3.2.1, the sequence of the reference strain CCIBt3277 being in species 3.2.1A. The strains share only 98.47-98.67% 16S rRNA sequence identity between species, 99.87-100% identity within species. Sequence length (nt): 1434-1483.

    Chroococcopsis gigantea Geitler, ref. str. suggested by Shalygin et al. (2019), falls as a single strain of a loner genus in family 5.1.5. Sequence length (nt): 1462.

    Cronbergia siamensis (Antarikanonda) Komárek, Zapomelová & Hindák: strains of this type species fall into two distinct genera, 1.7.1 and 1.18.1, with two strains named as Cronbergia sp. additionally lying in genus 1.3.6. Errors in sequencing are apparent, since the same strain (SAG 11.82) is represented by two sequences (KM019950 and GQ389643) that fall into both genera 1.7.1 and 1.18.1. Sequence length (nt): 1254-1482.

    The situation in the genus Cyanothece requires a longer description. Sequences of Cyanothece aeruginosa (Nägeli) Komárek (Neotype Mareš et al. 2019), the type species of Cyanothece, deposited by Bohunická et al. (unpublished), Friedl et al. (unpublished) and Mareš et al. (2019), fall into two different orders (3 and 5) and families (3.1 and 5.1.1). The sequence of the designated reference strain SAG 87.79 is found in both, in genera 3.1.21 and That deposited by Friedl et al. falls with Chlorogloea purpurea SAG 13.99, Eucapsis minor SAG 14.99 and Cyanothece sp. PCC 6910 in genus, whereas the sequences in genus 3.1.21 (Bohunická et al. and Mareš et al.) are identical to a strain named as Chlorogloea microcystoides (SAG 10.99). NCBI BLAST of the sequence in finds mainly strains assigned to the genus Microcystis at low (94.5%) identity; this is genus in our tree. Representative sequences of genus 3.1.21 find few close relatives in BLAST, except for the 8 clones of C. aeruginosa provided by the authors, a single sequence deposited by Bohunická et al. and the two sequences of Chlorogloea microcystoides SAG 10.99 (KM019955, E value 0.0, identity 99.91%, query cover 99%, and KC572079, 0.0, 99.91%, 92%), which are found in the same generic cluster in our tree. These were deposited by two different authors; curiously, they do not appear in the tree of Mareš et al. (2019). At lower identity (but greater than 94.0%), BLAST finds 6 clones of C. svehlovae NIVA-CYA 258/1 and "coccoid cyanobacterium" strain CMT-3FBIN-NPC3 clone PCI148B (here in genus 3.1.22) and many other strains submitted as "coccoid cyanobacterium" of which only 3 are shown here (genus 3.1.23). It is highly unlikely that the sequences of the 16S rRNA gene of a single strain fall into different orders, even if sequence heterogeneity (which was not detected in the 8 clones of the strain in genus 3.1.21) is considered. The SAG culture would therefore appear to contain two different organisms. This requires further investigation. Sequence length (nt): 1143-1163 (strain SAG 87.79 in is 1462).

    Desikacharya nostocoides Saraf & Prashant Singh (type designated by Saraf et al., 2019b) is present in our tree as the single member of an unnumbered genus of family 1.12. However, four other genera (1.12.3, 1.12.4, 1.12.5 and another unnumbered) are also named as Desikacharya. Sequence length (nt): 1486.

    Desmonostoc muscorum (C.Agardh ex Bornet & Flahault) Hrouzek & Ventura: papers of Friedl, Palińska, Rajaniemi, Ventura and others show this species in 3 genera (1.8.1, species F, 1.8.2, species A, B, C and H, 1.9.4) of two families. The sequence of the reference strain NIVA-CYA 818 is in species 1.8.2A with 3 other identical (100% 16S rRNA sequence identity) D. muscorum strains. Sequence length (nt): 936-1463.

    Dolichospermum flos-aquae (Brébisson ex Bornet & Flahault) P.Wacklin, L.Hoffmann & J.Komárek: strains given in publications of Fiore, Friedl, Honda, Rajaniemi, Zapomelova and many others are seen in four species of genus 1.1.1 (A, E, H and J), containing 2, 5, 2 and 2, respectively, and also in three other families as species 1.2.3A (1 strain), 1.8.2E (3 members) and genus 1.10.1 (4 strains). The strains in species 1.1.1A are almost identical in 16S rRNA sequence to those of 4 isolates and 1 metagenome of Aphanizomenon flos-aquae, the type species of the genus, as described above. This is obviously an unfortunate situation, and is further documented in the section "Problematic organisms" on this page. Sequence length (nt): 1344-1479.

    Dulcicalothrix necridiiformans: the genus was erected by Saraf et al. (2019c), with D. necridiiformans as the type species and strain V13 (deposited in metabolically inactive form in the National Centre for Microbial Resource as MCC 3324) as reference strain. The single 16S rRNA sequence of this strain forms species 1.15.9B, a member of the large Calothrix spp. generic cluster. The strains of this cluster show a minimum 16S rRNA sequence identity of 96.95% and a maximum of 98.71%. The authors transferred all members of that cluster to Dulcicalothrix; we think it unwise to do so until further members of this genus are isolated and sequenced. Sequence length (nt): 1476.

    Fischerella thermalis Gomont: strains isolated principally by Miller fall into 3 species (A, B and C) of genus 1.15.4. The major cluster is species 1.15.4A, whose 18 members possess identical 16S rRNA sequences; species B (1 strain) and C (2 members) share only 97.56% and 98.47% identity with the first. Sequence length (nt): 1459-1508.

    Foliisarcina bertiogensis D.O. Alvarenga, J. Rigonato, L.H.Z. Branco, I.S. Melo & M.F. Fiore: identified by Fiore and colleagues, the 4 strains assigned to this species are seen in a single genus as species and, showing only 97.10-97.17% 16S rRNA sequence identity; the sequence of the reference strain CENA333 is in genus Sequence length (nt): 1416-1418.

    Halomicronema excentricum Abed, Garcia-Pichel & Hernández-Mariné: identified by Garcia-Pichel, this organism lies in 2 genera (6.1.4, 6.1.5), sharing only 92.74% 16S rRNA sequence identity; the sequence of the reference strain TFEP1 is in genus 6.1.4. Sequence length (nt): 1348-1487.

    Hassallia byssoidea Hassall ex Bornet & Flahault; Type designated in Geitler, L. (1942): treated by Hauer and D. Singh, this taxon lies in 2 species (E and F, 16S rRNA sequence identity 98.58%) of genus 1.14.1. Sequence length (nt): 1129-1486.

    Koinonema pervagatum strains, identified and 4 sequences deposited in NCBI by Buch et al. (2017), are found in two distinct species (A and B) of genus, the species showing 98.45-98.52% 16S rRNA sequence identity. Sequence length (nt): 1482-1485.

    Komarekiella atlantica Hentschke, J.R.Johansen & Sant’Anna: in papers by Fiore or Scotta Hentschke this strain may be found in species of 2 genera (1.8.1, 1 strain, and 1.8.5A, 5 strains), the sequence of the reference strain CCIBt 3483 being in species 1.8.5A and sharing only 95.54% 16S rRNA sequence identity with the strain in the first species. Sequence length (nt): 1164-1482.

    Leptolyngbya boryana (Gomont) Anagnostidis & Komárek: this was studied by Hirose, Johansen, Osorio-Santos, Strunecký, Turner and others, and lies in 2 genera (6.3.1, 6 strains, and 6.3.5, 1 strain). The strains of genus 6.3.1 are identical in 16S rRNA sequence, and show only 96.50% identity to that of genus 6.3.5. Sequence length (nt): 971-1488.

    Limnothrix redekei (Goor) Meffert: described in papers by Sivonen, Ventura. M.M. Watanabe and Wilmotte, this species lies in species of 2 genera (, 4 strains and, 2 strains). Within each species, the strains are almost identical in 16S rRNA sequence, but the members show only 89.07-89.27% identity between species. Sequence length (nt): 1438-1475.

    Mastigocladus laminosus Cohn ex Kirchner: described by Castenholz, Johansen and others, sequences of this may be seen in 2 genera (1.15.1, 1 strain, 3 species (A, B and D) of genus 1.15.4, 8 strains) plus one in the bacterial outgroup. Sequence length (nt): 1286-1493.

    Myxacorys chilensis Pietrasiak et Johansen, identified by Pietrasiak et al. (2019), with ATA2-1-KO14 as reference strain, lies in genus 6.3.7. The 3 members assigned to this species are virtually indistinguishable from strains of M. californica in the same generic cluster. Sequence length (nt): all 1164.

    Nodosilinea nodulosa (Z.Li & J.Brand) Perkerson & Casamatta: isolates identified by Brandt, Johansen or M.M. Watanabe, are seen in species 6.1.1C (3 strains) and 6.1.1D (1 strain) of a single genus. The sequence of the reference strain UTEX 2910 is in species 6.1.1C, sharing 98.75-99.00% 16S rRNA sequence identity with the two other members of that species and 98.50% with N. nodulosa PCC 7104 in species 6.1.1D. Sequence length (nt): 1429-1488.

    Nostoc commune Vaucher ex Bornet & Flahault; Type designated in Geitler, L. (1942): strains listed in Arima et al. (2012), Fiore et al. (2005), Katoh et al. (2012), Nazifi et al. (2015) and Řeháková et al. (2007) fall into 10 different specific clusters (1.8.1B, E, I, L, O, Q, T, AD, AG and AK). We add to this list the strain CCIBt 3485, deposited in NCBI as Komarekiella atlantica (KX638486), the type species of that genus, but named in the publication of Scotta Hentschke et al. (2018) as Nostoc commune, which lies in species 1.8.1G. Dvořák et al. (2019) more recently performed metagenomic analysis of two herbarium exsiccatae of material which they identified as N. commune and as N. flagelliforme. Unfortunately, the genomes are only 80-83% complete as shown by CheckM and do not contain sufficient marker genes to be included in the genome tree, but each possesses one copy of the rrn operon. The 16S rRNA gene sequence extracted from the genome of the herbarium type material identified by the authors as N. commune on the basis of thallus morphology lies in species 1.8.1O, which may therefore represent the true phylogenetic position of the type species. Sequence length (nt): 1189-1495.

    Strains assigned to the type species of Oscillatoria, O. princeps Vaucher ex Gomont emend Muehlsteinova et al., fall into two species (A and B) of genus 3.2.5, separated by 97.66-97.87% 16S rRNA sequence identity. Sequence length (nt): 981-1460.

    Persinema komarekii Heidari & Hauer; this taxon, studied only by Heidari & Hauer, falls into 2 species (B and C) of genus, the sequence of the reference strain TM/S1 is in both; the TM/S1 clone 3 sequence has inserts resembling bacterial segments in helices 6, 10, 11, 18 and is detected as a chimera by UCHIME (a Geitlerinema-like organism and the bacterium Paenibacillus), the other clone not being chimeric. Sequence length (nt): 1452 (chimer excluded).

    Phormidesmis molle (Gomont) Turicchia, Ventura, Komárková & Komárek is the type species of the genus, with isolate 3CC04S05 as the type strain. Although treated by Turicchia et al. (2009), 16S rRNA sequences were unfortunately not made available. Strains assigned to this species are seen in two genera (6.3.1 and 6.3.18), with one forming a loner genus in family 6.3. Note that genus 6.3.1 is Leptolygbya sensu stricto, containing 6 strains of the type species L. boryana, in addition to the P. molle isolate. The situation is further complicated by the fact that one strain (PACC 8140) named as P. molle by Stoyanov et al. (2014) falls into the Nostocales. Sequence length (nt): 757-1471.

    Potamosiphon australiensis G.B.McGregor & Sendall: this "species" lies in 2 species (A and B) of genus 3.2.6. The members of species 3.2.6A share 99.59% 16S rRNA sequence identity, and strain FHC0914.02 in species 3.2.6B shows only 98.08-98.36% identity to the others. Sequence length (nt): 1461-1464.

    Prochlorococcus marinus Chisholm, Frankel, Goericke, Olson, Palenik, Waterbury, West-Johnsrud & Zettler 1992: the type strain of Prochlorococcus, identified by Chisholm, Rocap, Urbach and many others, falls into all 5 specific clusters of genus 7.1.3, the sequence of the reference strain CCMP1375 being in species 7.1.3C. Four of these species require renaming. Sequence length (nt): 1478-1485.

    Richelia intracellularis J.A.Schmidt: members of this taxon, examined in the laboratory of Zehr, lie in 2 species (A and B) of genus 1.15.18. The sequences shown are all extracted from draft genomes. Those in species 1.15.18A share 99.26% 16S rRNA sequence identity and only 97.0% identity with strain RC01 in species 1.15.18B. Sequence length (nt): all 1490.

    Stanieria cyanosphaera (Komárek & Hindák) Komárek & Anagnostidis: strains identified by e.g. Friedl, Turner or M.M. Watanabe, fall into 2 genera ( and sharing only 94.92% 16S rRNA sequence identity. Sequence length (nt): 1354-1493.

    Synechococcus elongatus (Nägeli) Nägeli: isolates named as this species by Fiore, Friedl and many others fall into 3 genera (, 1 strain, 7.2.1, 6 strains and 8.2.1, 1 strain). Many other sequences are available. S. elongatus sensu stricto appears to comprise genus 7.2.1. Sequence length (nt): 618-1489.

    Synechocystis aquatilis Sauvageau: members identified by Friedl or Ventura may be seen in 3 genera (,, Sequence length (nt): 1269-1461.

    Tenebriella Hauerová, Hauer et Kaštovský: strains named as members of the type species, T. amphibia, lie in the mono-specific genus, showing a minimum 16S rRNA identity of 98.81% to strains named as T. curviceps. The two "species" should be combined. Note that the sequence of Tenebriella sp. SERB 31 (KM982580, not sequenced by the authors) contains many mismatches in the paired stem of many helices, giving a long branch in the tree; we have omitted this sequence, although it is present in the tree shown by the authors. Sequence length (nt): 1143-1461.

    Tolypothrix distorta Kützing ex Bornet & Flahault; Type designated in Geitler, L. (1942): strains identified in publications of, for example, Becerra-Absalón et al. (2019), Hauer et al. (2014) and Flechtner et al. (2002) are found in six different specific clusters of genus 1.14.1, Tolypothrix sensu stricto. Using only the longer sequences, isolates within each species show a minumum of 98.70% (species 1.14.1I) to 99.59% (species 1.14.1M) 16S rRNA sequence identity. Sequence length (nt): 1112-1484.

    Two strains each of Toxopsis calypsus Lamprinou, Skaraki, Kotoulas, Economou-Amili & Pantazidou and Godleya alpina Novis & Visnovsky, both type species, fall as 2 species into the single genus 1.14.3. The strains share 99.54% and 99.81% 16S rRNA identity, respectively, within species and 98.61-98.89% between species. This is clearly an impossible situation, and Godleya alpina Novis & Visnovsky should have priority, having been described first. Sequence length (nt): 1455 (Toxopsis) and 1110-1120 (Godleya).

    Trichormus variabilis: strains carrying this name (some renamed from Anabaena variabilis in NCBI) are found in species 1.1.4B and monospecific genus 1.10.1. The two clusters share only low (around 95.4%) 16S rRNA sequence identity. Sequence length (nt): 1207-1489.

    Umezakia natans M.Watanabe: members identified by Neilan and colleagues occur in 2 species (A and C) of genus 1.3.3, sharing 97.69% 16S rRNA sequence identity. Sequence length (nt): 910-1318.

    Westiella intricata Borzì: identified by Micallef et al., this taxon is found in family 1.13 and genus 1.15.1. These clusters each contain a 16S rRNA gene extracted from the genome of W. intricata strain UH HT-29-1; such macro-heterogeneity may be real, or may result from inclusion of foreign DNA during assembly. The gene in species 1.15.1J is identical to that (KJ767016) obtained from 16S rRNA-specific DNA polymerase reaction. Sequence length (nt): 1439-1491.

    Westiellopsis prolifica Janet: sequences studied by e.g. Friedl, Hoffmann or Mareš lie in 3 species (A, G and J) of genus 1.15.1, sharing a maximum of 97.8% 16S rRNA sequence identity. Sequence length (nt): 1136-1347.

    Wollea saccata Bornet & Flahault: strains identified by Komárek or Mareš are seen in 2 genera (1.2.5 and 1.2.7, separated by 96.32% 16S rRNA sequence identity). Sequence length (nt): 1050-1447.


    The findings briefly described above illustrate difficulties encountered in the current system of Botanical taxonomy, in which descriptions of type species are often based on drawings and/or dried or fixed material in herbaria. Of almost 300 genera for which 16S rRNA sequences are available (listed here), 38 (31%) are problematic (see above), either because those of several independently-sequenced strains fall into different generic or specific clusters or because sequences lie in an existing genus with a type strain whose name has priority. A further 51 genera are represented by a single sequence of the type species; it may be anticipated that some of these will become problematic when further sequences become available. Of course, some of these strains will have been mis-identified, but we cannot escape the conclusion that many type species are ambiguously described. This situation is clearly not acceptable.

    A possible solution to this problem is to sequence herbarium exsiccatae of type material, although given the likely "metagenomic" nature of such specimens, it will be difficult to be certain that the type species itself has been sequenced. In addition, both Mareš et al. (2019) and Shalygin et al. (2019) have given examples where the herbarium material was impossible to obtain. Palińska et al. (2006) and Palińska et al. (2015) demonstrated the feasibility of extracting DNA from dried herbarium exsiccatae, and sequenced the 16S rRNA genes of 9 samples; these are labelled in our tree as "(herbarium)". Two of these have been transferred to different genera in later revisions: Microcoleus chthonoplastes HUW153 (DQ460700, species 3.1.4A) to Coleofasciculus, Nostoc muscorum HUW888 (DQ460702, species 1.8.2A) to Desmonostoc. Dvořák et al. (2019) significantly extended these attempts by performing metagenomic analyses of two herbarium exsiccatae from the Berlin Botanical Garden and Botanical Museum under catalogue numbers B 400001728 (hereafter sample S1) and B 400008238 (hereafter sample S2), which the authors identified as Nostoc commune and N. flagelliforme, respectively. In both metagenomes, the most abundant cyanobacterium was a Nostoc sp., reportedly followed by Scytonema millei. However, the accession number given for the latter is that of an outdated sequence version of this organism which we find to fall into the bacterial outgroup. S1 also contained, at low abundance, Microcoleus vaginatus (bin6), and both S1 and S2 contained proteobacteria (3 and 9, respectively, in addition to the misidentified Scytonema sequence). This confirms the mixed nature of herbarium specimens. Unfortunately, both of the Nostoc spp. genomes are less than 85% complete, contain fewer than 77 of our 79 core marker genes and cannot be included in our genome tree, but they each contain a single copy of the rrn operon. The 16S rRNA gene sequences extracted from these genomes with our SSU-finder script are found in species 1.8.1O (N. commune) and 1.8.1AD (N. flagelliforme). We find that the Microcoleus vaginatus (bin6) sequence is only 22.6% complete and contains no rrn operons; it is not included in any of our trees.


    Many "genera" are polyphyletic, e.g.:

    • Anabaena: 85 strains are shown in the tree, often found with Trichormus, in 8 families and 16 genera: 1.1.1, 1.1.2, 1.2.2, 1.2.3, 1.2.4, 1.2.5, 1.2.6, 1.2.7, 1.2.9, 1.3.1, 1.3.3, 1.8.2, 1.9.2, 1.10.1, 1.11.1 and 1.13.4.

    • Calothrix: the 63 members included fall into 8 generic clusters in families 1.4 (2 genera), 1.5 (1 genus), 1.8 (2 genera) 1.11 (1 genus), 1.12 (1 genus) and 1.15 (6 genera). Other (probably misnamed) Calothrix strains are scattered throughout order 1. No members of the type species, C. confervicola, have been sequenced.

    • Chroococcidiopsis: 23 members shown fall into genera 3.1.14 (15 strains), 3.1.16 (1), 3.1.17 (3), 3.1.18 (2), (1) and (1).

    • Chroococcus: the 19 strains shown fall into genera,,,, and

    • Cyanothece: even after the revision of Mareš et al. (2019), described under "Problematic genera" and "Type species" above, where organisms carrying this name are found as two genera (3.1.21 and 3.1.22), other strains identified as Cyanothece lie in genera,,, and 8.2.2.

    • Geitlerinema: strains assigned to the type species, G. splendidum, are all found in family 4.5/genus 4.5.1. Recognizing that another major cluster of Geitlerinema strains exists (family 5.1.9/genus, Strunecký et al. (2017) erected the genus Anagnostidinema to contain them, thus removing one cluster of Geitlerinema strains. However, isolates named as Geitlerinema are still found in genera (2 strains, both marine), (2, freshwater), (2, freshwater and salt marsh), 6.2.8 (1, hot spring), 6.3.29 (1, hot spring) and strains P-1104 (genus, PCC 7407 (genus 6.1.15) and PCC 9228 (genus whose 16S rRNA genes were extracted from the sequenced genomes. Eight strains of genus were renamed to Baaleninema spp. (Samylina et al., 2021) with B. simplex as the type species and PCC 7105 as the reference strain. However, these authors divided the members of species into two species: Baaleninema simplex (2 strains) and Baaleninema sp. (3 strains) which are in fact con-specific, showing a minimum of 99.66% 16S rRNA sequence identity. Many more 16S rRNA sequences of strains named as Geitlerinema are available, but not included in the tree. These organisms require further study.

    • Gloeothece: 23 strains are shown, falling into the closely-related genera,, and, which share no more than 93.72-96.22% 16S rRNA sequence identity.

    • Hydrocoleum: the 11 strains included fall into genera (with Okeania strains), (with Dapis), and Both of the latter genera contain exclusively isolates named as Hydrocoleum. Genus holds a 16S rRNA sequence obtained from herbarium material of H. lyngbyaceum; genus contains a sequence from herbarium material of H. brebissonii. Unfortunately, the sequences obtained from herbaria by Palińska et al. (2015) are too short to permit calculation of accurate 16S rRNA identity values, and their placement in the tree may be inexact. Nevertheless, this genus is clearly polyphyletic. We find one additional sequence of a strain (SAG 38.87) named as Hydrocoleum sp. in the distant genus; this strain is probably mis-identified. Palińska et al. (2015) similarly show a wide dispersion of Hydrocoleum spp. into 4 clusters.

    • Jaaginema: the 16 strains shown in the 16S rRNA phylogeny fall into 4 genera (genus 3.1.19, 4 sequences; species A and B, both containing 3 sequences;, 3 sequences and 6.2.20, 3 sequences). They are therefore assignable to 3 orders and 3 different families. The strains of species are virtually identical in 16S rRNA sequence to strains named as Limnothrix redekei, the type species of Limnothrix.

    • Komarekiella: Nine strains assigned to this genus by Scotta Hentschke et al. (2018), with K. atlantica as type species, are placed in species 1.8.5A (we show only 5 members). An additional strain (CCIBT 3485) is given in this publication as Nostoc commune but may be found in NCBI as K. atlantica (KX638486); this falls into Nostoc species 1.8.1G and is therefore indeed a member of a different genus, separated from the main cluster by several other genera.

    • Leptolyngbya: more than 150 sequences presented in the tree fall into at least 49 generic clusters in 4 families (5.1, 6.1, 6.2 and 6.3), plus 2 renamed (Perkerson III et al., 2011; da Silva Malone et al., 2020) to Nodosilinea (genus 6.1.1) and Monilinema (genus 6.3.5). Leptolyngbya sensu stricto (L. boryana) is found as genus 6.3.1, although isolates within this cluster, sharing a minimum of 99.2% 16S rRNA sequence identity, are named as 5 different genera and 6 species. The strain L. boryana LG4 Rupite (in genus 6.3.5) shares 99.97-99.90% 16S rRNA sequence identity to the 3 strains of Monilinema alkalinum (which share 99.90-100% identity between themselves), and is therefore a member of the same species. This strain, not shown in the tree of da Silva Malone et al., requires re-naming. Leptolyngbya sp. PCC 7376 (in genus is treated in the document describing the clusters, based on 16S rRNA identities and genome metrics.

    • Lyngbya: the 14 strains named as Lyngbya spp. shown in the tree fall into 7 different generic clusters (3.1.3, 3.3.1,,,, and 6.1.5) with one (in renamed to Limnoraphis. A strain named as a member of the type species, L. confervoides (strain BDU141951), lies in genus 6.1.5 and is represented by a draft genomic sequence. Other members of this cluster are currently named as Halomicronema excentricum, Leptolyngbya sp., Phormidium lucidum and Pseudanabaena sp.

    • Macrochaete: strains named as Macrochaete lie in two genera, 1.15.11 and 1.15.12, separated by 95.28-95.42% 16S rRNA sequence identity. The first contains 7 strains named as M. psychrophila, the type species, whereas the second contains strains identified as M. lichenoides and M. santannae.

    • Merismopedia: the 16S rRNA sequences of six strains assigned to this genus are available. The strains fall into species (1 strain), genus (2 strains), genus 7.1.1K (2 strains), with one as a singleton in family 5.1.7.

    • Microcoleus: of 20 strains shown, one is seen in genus 3.1.5 (Wilmottia), four in genus 3.1.8, two in each of genera 3.1.10 and 3.1.11, one in each of genera 3.1.12 and and most (9) in genus

    • Nostoc: more than 200 members are shown in the tree, falling into at least 16 generic clusters in the Nostocales, in 9 families (1.2, 1.3, 1.6, 1.8, 1.9, 1.10, 1.11, 1.12 and 1.13). Five generic clusters have been renamed from Nostoc to Aliinostoc (Bagchi et al., 2017, genus 1.3.4), Halotia (Silva et al., 2014, genus 1.3.7), Desmonostoc (Hrouzek et al., 2013, genus 1.8.2), Komarekiella (with data from Scotta Hentschke et al., 2017, genus 1.8.5) and Desikacharya (Saraf et al., 2019a,b, five genera in family 1.12). Members of the type species, N. commune, fall into 10 different specific clusters of genus 1.8.1 as described above.

    • Oscillatoria: Hauerová et al. (2016) created the genus Tenebriella with two named species (T. amphibia [type sp.] and T. curviceps), and transferred many Oscillatoria strains into the new genus; these are placed in genus of our tree, removing this cluster from the long list of genera occupied by strains named as Oscillatoria spp. Strains identified as members of the type species, O. princeps, lie in two different species of our tree (3.2.5A and 3.2.5B), sharing 97.66-98.27% 16S rRNA sequence identity. Other strains identified as Oscillatoria spp. are found in family 3.1 and genera 3.2.1, 3.3.1, 3.3.2,, (2 species),,, (3 species), (2 species), 6.1.1 (2 species), 6.1.15 and 6.3.1 (Leptolyngbya boryana).

    • Petalonema: two sequences of a single strain (isolated from a limestone rock, Slovakia) are the only members of genus 1.19.10. Two other strains named as Petalonema spp. are found in species 1.13.1A (isolated from a waterfall, Hawaii) and genus 1.16.3 (isolated from Antarctica). The sequences in genus 1.19.10 were assigned to the type species, P. alatum, by Mareš et al. (2015), who also observed the placement of the two other strains in distant genera and proposed that they be transferred to other genera.

    • Phormidesmis: 31 strains bearing this name are found in 9 different genera of two families (6.1.9, 6.1.10, 6.1.11, 6.3.1, 6.3.18, 6.3.19, 6.3.20, 6.3.23 and 6.3.30). Only two members of the type species, Phormidesmis molle (Turicchia et al. 2009, who separated the genus from Phormidium), are available: strains PACC 8140 and NIES-2126. The former falls into the heterocystous clade and has been omitted from the tree; the latter forms one of the two loner genera in family 6.3. Other sequences of the type species were reported by Turicchia et al. (2009), but were not made available. In a recent review of the genus, Raabová et al. (2019) created two new species (P. arctica and P. communis) and transferred Leptolyngbya nigrescens into P. nigrescens. The authors studied 26 strains of Phormidesmis originally designated as Leptolyngbya, Phormidesmis or Phormidium. The phylogenetic tree presented shows a unique cluster of these organisms, but no close relatives were included. They stated that the nucleotide similarity between species within Phormidesmis varied between 92% and 98%; it is therefore evident that the strains studied do not form a single genus. Indeed, P. arctica is seen both in genus 6.3.19 and in genus 6.3.20; one strain of P. communis falls into genus 6.3.18 with a second strain as a loner genus; strains of P. nigrescens fall into a single species (6.3.18B) but with strains of P. priestleyi; isolates of the latter may also be found in genera 6.1.10 and 6.1.11 (credited in NCBI to Taton et al., 2006, but deposited only in 2014). The remaining members of this "genus" shown in our tree are identified only to the generic level.

    • Pleurocapsa: 17 members are shown, falling into genera (4 strains), (7 strains), (the type species and reference strain), (2 strains), (a single strain). The 16S rRNA gene extracted from the genome of strain Pleurocapsa sp. PCC 7327 lies as a loner genus in family 5.1.3; this is not the result of poor assembly of the genome, since it is identical to gene sequences obtained from 16S rRNA-specific PCR reactions in two independent laboratories.

    • Pseudanabaena: the 34 members shown fall into genera 1.3.7 (a het- mutant),,, and, with other, probably mis-identified, strains in genera 6.1.1, 6.1.4, 6.1.5, 6.1.8 and 6.1.13.

    • Rivularia: of 22 strains represented, two form genus 1.15.10, twelve lie in genus 1.15.15 (isolates from marine/saline habitats) and eight in genus 1.15.16, whose members were isolated from desert environments.

    • Schizothrix: the 7 strains identified as members of this genus and shown in the tree fall into three clusters: 4 named as Schizothrix cf. calcicola in genus 6.2.18, 1 (as Schizothrix arenaria) is a "loner" strain in the same family (6.2) and 2 named as Schizothrix sp. in the different family 6.3 (genus 6.3.25).

    • The 59 sequences of Scytonema spp. included in the tree fall into 7 generic clusters in families 1.11 (1 strain only), 1.13 (3 strains), 1.16 (1 strain only), 1.17 (1 strain only) and 1.19 (5 clusters and 4 dispersed strains). Only a single strain, PCC 7110, assigned to the type species (S. hofmannii) has been sequenced, falling into genus 1.19.6. Two of the clusters of family 1.19 are described in more detail in the section on heterogeneity, below. Strains previously assigned to Scytonema crispum or S. cf. crispum were transferred to a new genus, Heteroscytonema, as H. crispum, in the thorough study of Sendall & McGregor (2018); these, of which we show only 7, form genus 1.16.5. They represent a single species, with a range of 16S rRNA sequence identity from 99.3 to 100%.

    • Stanieria: the 9 members shown for this genus fall into genera (3 strains), (4 strains) and (2 strains). Strains identified as the type species, S. cyanosphaera, are seen in genera (2) and (2).

    • Stigonema: of 26 strains shown, 24 lie in genus 1.16.1 and 2 in genus 1.16.4.

    • Synechococcus: 192 members, originally described under this name, are shown in our 16S rRNA phylogeny. They are dispersed into at least 10 clusters in 7 orders (families 2.1, 5.1.7, 5.4.1, 6.1, 6.3, 7.1, 7.2, 8.1, 8.2 and orders 9, 10). Komárek et al. (2020) made a small step in attempting to rectify this situation, as described below.

      • Genus We show 17 strains in 2 sub-clusters. Walter et al. (2017) renamed Synechococcus sp. PCC 7002 to Enugrolinea euryhalinus; Salazar et al. (2020) named the entire cluster to Limnothrix, based on comparisons of the identity of 14 genomic sequences, with strain PCC 7002 as "type genome". Most strains fell into a single species, L. euryhalinus, and two further unnamed specific clusters were found. The name Limnothrix, employed on the basis of the clustering of 10 unicellular strains "with two described Limnothrix genomes", was clearly an unfortunate choice, since it was previously validly published to describe filamentous, gas-vacuolate, cyanobacteria (Meffert, 1988). In addition, the low (72.26% ANIb) value of genetic identity between strains PCC 7002 and Limnothrix sp. IAM M-220 (Oscillatoria rosea NIES-208) clearly shows separation at the generic level. The above changes were not valid under the rules of the ICN. Komárek et al. (2020) created and validated the generic name Picosynechococcus for this cluster, but showed only 6 strains in their 16S rRNA phylogeny; they omitted many of the strains represented by complete chromosomes and did not report 16S rRNA identity values. Our analysis shows that the 2 sub-clusters ( with 11 strains and with 5 strains) correspond to 2 distinct species, sharing 99.50-100% and 99.70-100% 16S rRNA sequence identity, respectively, internally and separated by values of 97.40-97.71%. The type species and reference strain P. fontinalis FG-1 lies in the second specific cluster. Although incompletely treated by Komárek et al. (2020), we have renamed all members of this generic cluster to Picosynechococcus, excluding Leptolyngbya sp. PCC 7376 and Limnothrix rosea IAM M-220. The latter strains show only 71.91% and 72.26% ANIb with the complete chromosome of strain PCC 7002; the genome of the reference strain is not available, but that of strain NIES-970 in the same species gives 71.70% ANIb with strain PCC 7376 and 71.97% with strain IAM M220. The genomic evidence there clearly indicates that the two filamentous strains are not members of this genus. The diagnosis given by Komárek et al. (2020) implies that motility is a general characteristic of this genus; however, the strains which we have been able to verify (those of the PCC) are immotile. Furthermore, the use of "Picosynechococcus" as a generic name would appear to be unwise, especially since the authors were aware that the term "Pico" was in widespread use to describe the small-celled (0.4-0.8 µm in width) Synechococcus-like organisms now classified as Parasynechococcus. The type sp. of Picosynechococcus was descibed as having cells of width 0.8–2.0 μm, and the strains which we have verified possess cells of 1.2-2.0 μm width. Cells of strains assigned to the genus Synechococcus are 1.0-1.2 μm wide, within the range of Picosynechococcus.
      • Genus 7.1.1. More than 100 members are shown in our tree, forming 32 spp. on the basis of 16S rRNA sequence identity. Coutinho et al. (2016a) assigned the generic name Parasynechococcus to 15 strains of this cluster, and subsequently (Coutinho et al., 2016b) assigned specific names to each of the 15 strains. The genus Parasynechococcus was not validly published under the rules of either the ICN or ICNP, but was validated later by Komárek et al. (2020). The tree of these authors shows only 9 strains and a single named species, the type species P. marenigrum Coutinho, Tschoeke, Thompson et Thompson. The reference strain, WH8102, lies in cluster 7.1.1Y of our tree. The valid names of 14 species (excluding P. indicus strain CB0205, for which only fragments of the rrn operon could be extracted from the genome) are shown in our phylogenetic trees, which still contain many strains with the original name. Walter et al. (2017) retained the generic name Parasynechococcus for the species africanus (strain CC9605), chilensis (CC9902), marearabicus (WH8109), marenigrum (WH8102) and nordiatlanticus (BL107), and created 4 new genera (Pseudosynechococcus, Magnicoccus, Regnicoccus and Inmanicoccus), based on genomic comparisons. Salazar et al. (2020), again based on genomic comparisons, expanded the coverage of this cluster with a study of many new genomes, and added the generic name Lacustricoccus. These authors retained the genus Parasynechococcus with the 5 named species of Walter et al. (2017), P. africanus strain CC9605 being the "type genome". None of the above generic or specific names have been validly published, and we have retained the original names in our phylogenetic trees.
      • Genus 7.1.5. The 11 strains of this cluster, mostly originally named as Candidatus Synechococcus spongiarum, form a single species, sharing 98.71 to 100% 16S rRNA sequence identity. Salazar et al. (2020) renamed this cluster to Synechospongium, based on genomic studies. Since this was not validly published, and these organisms were not treated by Komárek et al. (2020), we have retained the original names in our trees.
      • Genus 7.2.1. This cluster contains 14 strains of the genus Synechococcus, representing a single species with a minimum of 99.79% 16S rRNA sequence identity. The type species, S. elongatus, and reference strain (PCC 6301) were accepted by Komárek et al. (2020). A seemingly misnamed strain carrying the specific epithet elongatus falls into genus 8.2.1. We have retained genus 7.2.1 as Synechococcus sensu stricto. S. lacustris strain Tous, lying in genus 7.1.2 with 3 strains named as Synechococcus sp., was renamed to Lacustricoccus lacustris by Salazar et al. (2020). This change was not validly published.
      • Genus 8.1.1. The genus Thermosynechococcus was initially invalidly published (Katoh et al., 2001; type species T. elongatus), but was validated by Komárek et al. (2020). We include 18 strains of T. vestitus, T. vulcanus and Thermosynechococcus spp. in genus 8.1.1, which is divided into 3 specific clusters. The latter share internally 98.89-100% 16S rRNA sequence identity and are separated by values of 96.87-98.04%. All strains are of thermal origin. Komárek et al. (2020) show 6 Thermosynechococcus strains of species 8.1.1A in their 16S rRNA phylogeny and one additional strain (PCC 6715, species 8.1.1B) in a genome-based tree. The type species is Thermosynechococcus vestitus (Copeland) Komárek et Strunecký comb. nov., reference strain: BP-1, lying in cluster 8.1.1A. We have renamed all strains of generic cluster 8.1.1 to Thermosynechococcus.
      • Genus 10.1.1. This genus is divided into 2 species; species 10.1.1A contains 8 strains with a minimum of 99.86% 16S rRNA sequence identity and 10.1.1B is represented by 9 strains which show 99.03-99.93% sequence identity. The species separate with around 96.5% rRNA sequence identity. All members were isolated from thermal environments. Walter et al. (2017) renamed strain JA-3-3Ab (in species 10.1.1A) as Leptococcus yellowstonii and JA-2-3B'a (in 10.1.1B) as L. springii. These names were not validly published.
      • Genus 10.1.2. The generic cluster is divided into 5 species, separated by 95.09-98.63% 16S rRNA sequence identity and containing only 8 strains. Komárek et al. (2020), on the basis of the single strain Rupite RUP-VU-1 with a short (1070 nt) 16S rRNA sequence, renamed this cluster to Thermostichus, with strain Rupite RUP-VU-1 as the type species and reference strain. These authors did not report 16S rRNA identity values and, unfortunately, included the members of both genera 10.1.1 and 10.1.2 in the single genus Thermostichus. Awaiting further analysis, we have retained the members of both generic clusters as Synechococcus spp., reserving the name T. vulcanus only for the 3 members of species 10.1.2C. Synechococcus sp. strain PCC 7336 is also shown as a member of the genus Thermostichus in the genomic tree of Komárek et al. (2020). This is clearly a major error, since the strain lies in genus 9.1.1 of our tree.
      • Genus 10.2.1. The nine members of the 4 species of this generic cluster, again isolated from thermal environments, were not included in the study of Komárek et al. (2020), and have been retained here under their original names. Curiously, the single 16S rRNA gene extracted from the genome sequence of Gloeomargarita lithophora strain Alchichica-D10, isolated from the highly alkaline volcanic crater Lake Alchichica (Moreira et al., 2017), also sits in this generic cluster in the 16S rRNA phylogeny (species 10.2.1D, sharing 98.23% 16S rRNA sequence identity with, for example Synechococcus sp. strain ATCC 700244 in species 10.2.1A), but the genome sequence forms a distinct branch in the genomic tree. The 16S rRNA sequences of two uncultured Gloeomargarita-like organisms, occurring in the same lake, are virtually identical to that of strain D10. The phylogeny of these organisms deserves further study, and more precise details of the temperature of the sampling site (not known to have high temperatures) are important.

      In conclusion, most of these "Synechococcus-like" cyanobacteria still show an extreme polyphyly and their nomenclature requires further extensive revision. More than 150 strains remain to be considered.

    • Synechocystis: the 52 strains fall into genera 2.1.1 (1 strain), 2.1.3 (1), 3.4.2 (3), (1), (1), (5), (2), (3), (1), (2), (30) and (2). The type species, S. aquatilis, lies in genus which additionally contains only four isolates, all named as Stanieria.

    • Tolypothrix: We show only 43 sequences of strains assigned to this genus. They fall into 5 families (1.4, 1.8, 1.9, 1.11 and 1.14) of the Nostocales (order 1) as 6 generic clusters; the members of 5 such clusters are clearly misnamed. All strains of the type species, T. distorta, are seen in genus 1.14.1, but as 6 distinct specific clusters (1.14.1A, F, I, K, M and O), with the majority (10 of 17 strains) in species 1.14.1I.

    • Trichormus: Komárek and Anagnostidis (1989) transferred Anabaena variabilis (and many other species) into this genus with T. variabilis as type species, without phylogenetic study. Strains named as members of this genus (often renamed in NCBI) group in at least 2 generic clusters (1.1.4 and 1.10.1). Single strains are also found in 1.2.6B (Sherwood et al., unpub), 1.3.7B (Miscoe et al., 2016), 1.4.1A (Johansen et al., unpub), 1.12.1B (Miscoe et al., 2016) and 1.13.4 (Rajaniemi et al., 2005), making a total of 7 generic clusters in 7 families. Rajaniemi et al. (2005) remarked on the polyphyly of this "genus".

    16S rRNA gene heterogeneity

    Except for certain genera (e.g. Gloeobacter, Gloeomargarita, most Prochlorococcus strains), cyanobacteria possess multiple copies of the rrn operon, ranging in number from 2 to 7. A study of the 16S rRNA genes extracted from completed chromosomes with our SSU-finder script permits comparison of the sequence variation of the various copies of a single strain. Normally, the multiple copies within a single organism are identical in sequence (e.g. the 4 genes of Nostoc punctiforme PCC 73102 or the 5 of Nostoc sp. PCC 7524). In some cases they exhibit microheterogeneity. For example, Nostoc flagelliforme strain CCNUN1 is represented by a complete chromosome sequence (CP024785) with six 16S rRNA genes but only 5 rrn operons (rrnA-rrnE) whose almost completely identical (99.8-100%) 16S rRNA sequences fall into species 1.8.1AE (whose members show 99.8-100% 16S rRNA sequence identity); only one (rrnA) is shown in the tree based on 16S rRNA sequences. The additional 16S rRNA sequence (labelled as "(no operon)" in place of an operon designation, since it is not part of a rrn operon) falls into species 1.8.1AC; it shows 99.2-99.4% 16S rRNA sequence identity with the genes of operons A-E, but also 99.1-99.3% with other members of species 1.8.1AC. All sequences of species 1.8.1AC, including the CCNUN1 "no operon" sequence, differ at 4 paired sites in the stem structure of helix 37 from members of species 1.8.1AE; members of the two species are separated by around 98.1% 16S rRNA sequence identity. The atypical 16S rRNA gene of this organism is not carried on any of the 8 plasmids of this strain; it is directly followed on the chromosome by a transposase. None of the other completed chromosomes show higher variation. Among the draft genomes, one (in operon rrnD, falling into family 1.13) of the three 16S rRNA genes of Nostoc sp. strain 996 differs from the others (rrnB and rrnC, genus 1.3.7, only one being shown here) in 156 positions, giving 89.74% identity and placing the organism into a different genus; however, this is a metagenomic assembly and suggests the presence of DNA from at least two organisms. The two 16S rRNA genes of Westiella intricata UH HT-29-1 (see Micallef et al., 2014 for genome description) show 156 differences; NCBI BLAST results indicate that gene B (genus 1.15.1) is identical to the 16S rRNA gene of other members of the true-branching heterocystous type, whereas gene A (in family 1.13) shows no identity with members of this cluster. This again may suggest the presence of DNA from two organisms. The "nearly-complete" genome (the term used by the authors, meaning the genome has been assembled to a single contig but not closed) of the axenic strain Sctyonema sp. HK-05 (NIES-2130) carries 4 identical chromosomal copies of the 16S rRNA gene plus an incomplete rrn operon on plasmid 4, containing one copy each of the 16S and 23S rRNA genes but no 5S rRNA; the 16S rRNA gene on the plasmid (in genus 1.19.11) differs from the chromosomal copies (which fall into genus 1.19.1, only one being shown) at 120 positions (91.94% identity), placing the organism into a different genus if this gene is recovered in a 16S rRNA-specific PCR reaction. The differences are dispersed along the entire length of the gene, therefore it is not a chimera. In the few other examples available (Leptolyngbya boryana NIES-2135 in genus 6.3.1 and Microcoleus sp. PCC 7113 in genus 3.1.10, both complete chromosomes from axenic organisms and carrying 3 chromosomal copies of the rrn operon), the plasmid-borne copy of the 16S rRNA gene, in a complete rrn operon, is identical to the chromosomal versions (only one of which of each organism is included in our tree). To the best of our knowledge, plasmid-borne 16S rRNA genes are rare in other bacteria; we have found only 3 (Deinococcus geothermalis DSM 11300, CP000358; Mesorhizobium sp. CCNWGX035, DQ273666; Rhodococcus equi strain 28, DQ150571). These are identical to the chromosomal copies. Please write to mherdman1245(at) to inform us of others that you know of.


    In a detailed study, Johansen et al. (2017) cloned a large number of the PCR amplification products from the total DNA purified from individual strains of Scytonema hyalinum and sequenced the clones repeatedly until a disparate sequence was found. Sequences from the separate clones of each of 20 organisms were found in two different generic clusters (here 1.19.1 and 1.19.11). We show only five of these (strains CMT-1BRIN-NPC32, CMT-1SWIN-NPC17, HAF2-B2-c1, MX1 and WJT9-NPBG6A, indicated by red arrows in the figure to the left), the others being strains ATA-SAL-RM1, CMT-1BRIN-NPC30, CMT-1BRIN-NPC31, CMT-1BRIN-NPC39, EM3-HA15, EM3-HA19, EM3-HA20, EM3-HA25, EM3-Z1-c02, FL17-MK96, HA4185-MV1, HTT-U-KK4, WJT4-NPBG1, WJT9-NPBG6B, FI5-JRJ03. The secondary structure of the disparate gene is essentially unchanged, thus the gene may be functional.

    The HK-05 plasmid 16S rRNA sequence described above falls into genus 1.19.11, whereas the chromosomal copies lie in genus 1.19.1. The plasmid copy is clearly distinct, sharing only 91.94% 16S rRNA sequence identity with the chromosomal genes. The existence of a promiscuous plasmid capable of transferring to a variety of Scytonema strains may explain the results of Johansen et al. We cannot exclude the possibility that the strains concerned are not unicyanobacterial, but harbour two different organisms, both of which were sequenced and represent the two clones reported for each strain. Assuming that this is not the case, the 2 clones of, for example, strain CMT-1BRIN-NPC32 (T1 and T2) share only 91.7% 16S rRNA sequence identity over the comparable sequence region (1420 nt), and thus differ at 118 positions. These positions are again dispersed along the entire length of the gene, excluding the possibility of chimera formation. Such a high level of intragenomic heterogeneity, if common in other cyanobacterial genera, would make the placement of many strains in the tree doubtful, since their position would depend on the operon sequenced. Further studies of transcription and sequencing of more plasmids are required to resolve this situation.



    As for the 16S rRNA sequences, we have used the names given by the authors at the time of release of the genomes, changing these if the strain has been re-named and validated under the rules of the ICN; note that some have been re-named in NCBI. In particular, "Synechococcus" sp. LL became Vulcanococcus limneticus LL (NQLA00000000; di Cesare et al., 2018). The latter change implies that the organism was isolated from a thermal environment; although it was indeed found in a lake in a volcanic crater, the lake contains a wide variety of freshwater fish and is suitable for swimming. It is not surprising, therefore, that this strain clusters with other freshwater organisms in the genome tree. Many metagenomes of Vulcanococcus spp. were sequenced from freshwater samples taken from the Chesapeake and Delaware Bays, USA, not known to be volcanic.In a thorough study of genomic sequences, Coutinho et al. (2016a) separated a marine group of "Synechococcus" (the sister group of Prochlorococcus) from all other "Synechococcus" clusters by erecting the genus Parasynechococcus, based on 15 strains, of which strain WH8102 is the type. These strains were subsequently (Coutinho et al., 2016b) each assigned a specific epithet. While this would be an excellent approach to reducing the number of generic clusters named "Synechococcus" (see above), these strains are invalid under the rules of the ICN, since a type species was not designated, and under the ICNP, are still in NCBI as "Synechococcus", and we have retained them as such in the genome trees. We employ the term "metagenome" only for those genomes obtained from environmental samples. In principle, this term should also apply to those from non-axenic, non-clonal cultures.

    Although we tend to treat bacterial genome sequences as the new gold standard for, among many other purposes, the study of phylogeny, this concept cannot at present be applied to Cyanobacteria. Sequences derived from metagenomic assemblages may be useful for many investigations, but they may in fact be chimeric, containing parts of the genomes of several organisms; their phylogenetic use should be treated with caution. Examples of such heterogeneity are shown here. Draft genome sequences are, by definition, incomplete; many most likely lack parts of the genome of the organism, may be chimeric and may contain duplicated segments. Unfortunately, this type of genome is the most frequent in the current data set. Even complete chromosomes should be treated with care: some are, in fact incomplete (see the Methods section, above); many are derived from non-axenic cultures and are therefore likely to contain foreign DNA. The only genomic sequences we can trust are those from an axenic culture of the organism, prepared and carefully checked for contamination by a trained microbiologist before DNA extraction. Only two culture collections currently offer a large number of axenic isolates: the PCC and the NIES; there are very few other genomes from axenic strains (e.g. Dolichospermum sp. UHCC 0090 CP003284/5 and Synechococcus elongatus UTEX 3055 CP033061).

    We present two phylogenetic trees constructed from genomic sequences. That with entirely complete chromosomes is accessible from the button "Complete only" on the Home page navigation menu or by clicking here; the tree built with complete plus draft genomes is opened by the "Complete and WGS" button on the Home page, or here. Both trees have the genome logo at the top; clicking this opens the appropriate list of excluded genomes.

    Type species and reference strains

    Of over 800 genomes (more than 180 complete) presented in the trees, less than 200 are type species (most of these represent only two genera: Microcystis and Prochlorococcus) and few represent reference strains. Less than 10 of the latter are completed chromosomes.

    Genome properties

    Taking only the genomes of axenic isolates into account, the Cyanobacteria show an enormous range of both DNA base composition (30.8 to 68.7 mol% G+C for Prochlorococcus marinus PCC 9511 and Cyanobium gracile PCC 6307, respectively) and genetic complexity (from 1.44 to 11.06 Mbp for Candidatus Atelocyanobacterium thalassa ALOHA and Calothrix sp. NIES 4071, respectively), as summarized in the following figure.


    These values are similar to those reported (in pre-sequencing days) by Herdman et al., 1979a,b (35 to 71 mol% G+C and 2.03 to 10.94 Mbp), Prochlorococcus spp. isolates not being available at that time. The completed genomes of unicellular cyanobacteria presented here (excluding Prochlorococcus spp.) range in size from 2.11 to 6.50 Mbp (pale blue in the figure); those of baeocyte-forming organisms from 4.99 to 7.70 Mbp (orange); filamentous non-heterocystous strains exhibit a range of 3.15 to 9.67 Mbp (green) and the heterocystous strains (dark green) from 5.29 to 11.06 Mbp. There is, therefore, a general trend of increasing genome size with increasing morphological complexity, with a certain degree of overlap of genetic complexity between these morphotypes. The Prochlorococcus spp. appear as 5 species in the 16S rRNA tree and show two distinct size distributions: 4 species (7.1.3A-7.1.3D) vary from 1.64 to 1.93 Mbp, showing extensive genome streamlining (see e.g. Yooseph et al., 2010), whereas members of the fifth (7.1.3E, strains MIT9303 and MIT9313 being the only organisms represented by complete chromosomal sequences) contain much larger genomes (2.68 and 2.41 Mbp, respectively). The genome sizes of the latter are similar to those of their sister group OMF-type "Synechococcus" (2.22 to 3.34 Mbp). Unfortunately, there are no completed chromosomal sequences available for other members of the "Prochlorales", but both the 16S rRNA and draft genome trees clearly illustrate the now generally-accepted division of these organisms into 2 orders and 3 distinct families containing, respectively, the marine unicellular picoplanktonic Prochlorococcus spp. (genus 7.1.3), the unicellular endosymbiotic Prochloron didemni (genus and the filamentous Prochlorothrix hollandica (genus 7.3.1).

    genome-repeats Many genomes contain a high proportion of repeat sequences, which may be direct, inverted or palindromes. Since it seems pointless to examine this property in the draft genomes, the repeats are shown only for complete chromosomes in the figure to the left.

    Key to symbols:
    Magenta, two strains (PAL-8-15-08-1 and JHB) of Moorena producens; red, Microcystis aeruginosa, the high values are strains NIES-102, NIES-843 and FACHB-1757; orange, Arthrospira platensis; black, Planktothrix agardhii; green, Prochlorococcus spp.; pale blue, "Synechococcus" relatives of Prochlorococcus; dark blue, all other cyanobacteria.

    Environmental aspects

    Of the other Cyanobacteria, only the heterocystous organisms form a distinct phylogenetic cluster (order I in the 16S rRNA trees, containing 18 families), the only exception being the organism known as Scytonema millei strain VB511283 (QVFW00000000) which is described in the genome metrics document. The other morphotypes are often intermixed within a single family (see the 16S rRNA tree coloured by morphotype). Although morphotypic information is given for almost all strains, or is evident from their generic name, environmental data are often scanty. The information available for strains whose chromosome has been completely sequenced is partly documented below, the generic numbering system being taken from the 16S rRNA trees.

    • Organisms from thermal environments are scattered throughout the trees but mostly fall into a number of discrete clusters, represented by: Fischerella thermalis (genus 1.15.4), Gloeocapsa sp. (genus 3.1.16), Cyanobacterium aponinum (genus, Leptolyngbya sp. (genus 6.2.8), Thermosynechococcus elongatus (genus 8.1.1) and "Synechococcus" strains of genus 10.1.1. Unfortunately, genome sequences are not available for many interesting clusters, notably Cyanocohniella calida (genus 1.3.5), Ewamiania thermalis (genus 1.19.12), Jaaginema sp. (genus, Onodrimia javanensis (genus 6.3.14) and "Synechococcus" strains of genus 10.2.1.

    • Strains isolated from hypersaline environments are again scattered throughout the trees but are poorly represented by complete chromosome sequences. We can list only Dactylococcopsis salina (genus, Euhalothece natronophila and Halothece sp. (genus, the halophilic Geitlerinema of genus, and Halomicronema hongdechloris (genus 6.1.17). Again, it would be interesting to obtain complete genomic sequences for some clusters, for example Halospirulina tapeticola (genus and the Halomicronema members of genus 6.1.4.

    • Alkalophilic organisms are very poorly represented by strains whose chromosome has been completely sequenced: Cyanobacterium stanieri (genus and Gloeomargarita lithophora (genus 10.2.1) are the only examples.

    • Planktonic cyanobacterial organisms are much more numerous than the thermophilic and halophilic members, and some genera are well represented by complete chromosome sequences. Microcystis aeruginosa (genus, with 10 complete genomes) and Planktothrix spp. (genus, 2 complete genomes) are good examples. Completed chromosomal sequences of organisms such as Raphidiopsis raciborskii (genus 1.2.1) and many others are limited in number.

    • Cyanobionts, whether facultative or obligate, occur throughout the trees. Strains whose chromosome has been completely sequenced include Nostoc azollae 0708 (genus 1.2.5), Nostoc punctiforme PCC 73102, Nostoc sp. N6, Nostoc sp. 5183 and 10 other members of genus 1.8.1. This genus harbours the majority of cyanobionts, which profit from small-celled, motile, hormogonia to colonize higher plants. Although we might expect baeocytes, if motile, to perform the same function, very few symbiotic forms have been sequenced. We exclude from the group of cyanobionts organisms such as the Hormoscilla "symbionts" of marine sponges, since they merely seem to profit from the "host" as substrate. The unidentified organisms EtSB and RgSB (genus are spheroid bodies found as obligate symbionts of Epithemia turgida (Nakayama et al., 2014) and Rhopalodia gibberula (Nakayama and Inagaki, 2017), respectively. Unlike other cyanobionts, which can be cultured, these are non-photosynthetic, providing the host with nitrogenous material as a result of N2 fixation.


    Despite the obstacles discussed in the opening paragraphs above, it seems possible to obtain accurate phylogenetic trees of the cyanobacterial phylum. This is demonstrated by the congruence between the tree of Mareš (2018) and those presented on this site. The only observable difference in these trees is the position of the branch containing Chamaesiphon minutus PCC 6605 (CP003600) and Crinalium epipsammum PCC 9333 (CP003620). The position is well-supported in both trees, and further studies are required to elucidate this problem. However, the 16S rRNA data support the placement of these strains at the base of the Nostocales clade, as found in our genomic trees, and Calteau et al. (2014) in their study find concordant results. The congruence of the trees from both studies also gives us confidence in the use of the FastTree software, which works with profiles of internal nodes instead of a distance matrix, for tree inference. The use of protein, rather than DNA, sequences as markers gives higher resolution, but the number to employ is open to debate. Mareš (2018) used 23, of which 21 are used in our set of 79 sequences. The two (DnaG [DNA primase] and RpoB [RNA polymerase beta subunit]) sequences of Mareš not employed here may be a good addition. In both studies, the markers are well dispersed on the genomes studied, as shown in these figures. The studies of Mareš and our own both show the unusual behaviour of strain Leptolyngbya sp. O-77 where, of the two 16S rRNA genes extracted from the genome sequence (AP017367), one falls into the bacterial outgroup of trees inferred from 16S rRNA sequences; we have been unable to show that this sequence is chimeric, but it does contain many mis-matched bases, and therefore assume inclusion of foreign DNA during assembly of the genome. The position of this strain in the genomic trees may therefore be open to doubt. We have observed similar misplacement of a 16S rRNA gene copy from the genome sequence of Cyanobacterium aponinum strain 0216 (WMIA00000000), where one falls into the bacterial outgroup, and from the metagenomic sequences of both Anabaena spp. MDT14b (LJOV00000000) and WA113 (LJOS00000000), both also seen in the bacterial outgroup.

    Note that the genomic trees built de novo and that extracted directly from the ARB 16S rRNA database are not identical. Most of the differences involve simple branch rotations, which cause some clusters to invert in the tree; an exception is the position of the Pseudanabaena clade, which cannot be resolved in this way. Phylogenetic inference by any method will only find the tree that best fits the data and parameters supplied, and this is rarely the "best" tree. We have used the same parameters for inferring both trees, but they may still suffer from taxon-sampling errors.


    Akagha, S.C., Johansen, J.R., Nwankwo, D.I. and Yin, K. (2019). Lagosinema tenuis gen. et sp. nov. (Prochlorotrichaceae, Cyanobacteria): a new brackish water genus from Tropical Africa. Fottea 19: 1-2.

    Alvarenga, D.O., Andreote, A.P.D., Branco, L.H.Z. and Fiore, M.F. (2017). Kryptousia macronema gen. nov., sp. nov. and Kryptousia microlepis sp. nov., nostocalean cyanobacteria isolated from phyllospheres. International Journal of Systematic and Evolutionary Microbiology 67: 3301–3309.

    Alvarenga, D.O., Andreote, A.P.D., Branco, L.H.Z., Delbaje, E., Cruz, R.B., Varani, A. and Fiore, M.F. (2021). Amazonocrinis nigriterrae gen. nov., sp. nov., Atlanticothrix silvestris gen. nov., sp. nov. and Dendronalium phyllosphericum gen. nov., sp. nov., nostocacean cyanobacteria from Brazilian environments. International Journal of Systematic and Evolutionary Microbiology 71. Epub ahead of print. DOI: 10.1099/ijsem.0.004811.

    Anagnostidis, K. and Komárek, J. (1988). Modern approach to the classification system of cyanophytes. 3. Oscillatoriales. Arch Hydrobiol Suppl. 80: 327–472.

    Arima, H., Horiguchi, N., Takaichi, S., Kofuji, R., Ishida, K.I., Wada, K. and Sakamoto, T. (2012). Molecular genetic and chemotaxonomic characterization of the terrestrial cyanobacterium Nostoc commune and its neighboring species. FEMS Microbiology Ecology 79: 34-45.

    Azevedo, M.T. and Sant' Anna, C.L. (2003). Sphaerocavum, a new genus of planktic Cyanobacteria from continental water bodies in Brazil. Algological Studies 109: 79–92.

    Bagchi, S.N., Dubey, N. and Singh, P. (2017). Phylogenetically distant clade of Nostoc-like taxa with the description of Aliinostoc gen. nov. and Aliinostoc morphoplasticum sp. nov. International Journal of Systematic and Evolutionary Microbiology 67: 3329–3338.

    Becerra-Absalón, I., Johansen, J.R., Muñoz-Martín, M.A. and Montejano, G. (2018). Chroakolemma gen. nov. (Leptolyngbyaceae, Cyanobacteria) from soil biocrusts in the semi-desert Central Region of Mexico. Phytotaxa 367:201.

    Becerra-Absalón, I., Muñoz-Martín, M.A., Montejano, G. and Mateo, P. (2019). Differences in the Cyanobacterial Community Composition of Biocrusts From the Drylands of Central Mexico. Are There Endemic Species? Frontiers in Microbiology 10: 937.

    Becerra-Absalón, I., Johansen, J.R., Osorio-Santos, K. and Montejano, G. (2020). Two new Oculatella (Oculatellaceae, Cyanobacteria) species in soil crusts from tropical semi-arid uplands of Mexico. Fottea 20: 160–170.

    Berthold, D.E., Lefler, F.W. and Laughinghouse, H.D. (2021). Untangling filamentous marine cyanobacterial diversity from the coast of South Florida with the description of Vermifilaceae fam. nov. and three new genera: Leptochromothrix gen. nov., Ophiophycus gen. nov., and Vermifilum gen. nov. Molecular Phylogenetics and Evolution 160: 107010.

    Bohunická, M., Mareš, J., Hrouzek, P., Urajová, P, Lukeš, M. and 5 other authors (2015). A combined morphological, ultrastructural, molecular, and biochemical study of the peculiar family Gomontiellaceae (Oscillatoriales) reveals a new cylindrospermopsin-producing clade of cyanobacteria. Journal of Phycology 51: 1040–1054.

    Bohunická, M., Pietrasiak, N., Johansen, J.R., Berrendero-Gómez, E., Hauer, T., Gaysina, L.A. and Lukešová, A. (2015). Roholtiella, gen. nov. (Nostocales, Cyanobacteria)—a tapering and branching cyanobacteria of the family Nostocaceae. Phytotaxa 197: 84-103.

    Brito,A., Vieira, J., Vieira, C.P., Zhu, T., Leão, P.N. and 5 other authors (2020). Comparative Genomics Discloses the Uniqueness and the Biosynthetic Potential of the Marine Cyanobacterium Hyella patelloides. Frontiers in Microbiology 11: 1527.

    Buch, B., Martins, M.D. and Branco, L.H.Z. (2017). A widespread cyanobacterium supported by polyphasic approach: proposition of Koinonema pervagatum gen. & sp. nov. (Oscillatoriales). Journal of Phycology 53: 1097-1105.

    Cai, F., Li, X., Geng, R., Peng, X. and Li, R. (2019a). Phylogenetically distant clade of Nostoc-like taxa with the description of Minunostoc gen. nov. and Minunostoc cylindricum sp. nov. Fottea 19: 13-24.

    Cai, F., Li, X., Yang, Y., Jia, N., Huo, D. and Li, R. (2019b). Compactonostoc shennongjiaensis gen. & sp. nov. (Nostocales, Cyanobacteria) from a wet rocky wall in China. Phycologia 58: 200-210.

    Caires, T.A., Lyra, G .de M., Hentschke, G.S., da Silva, A.M.S., de Araújo, V.L., Sant’Anna, C.L. and de C. Nunes, J.M. (2018). Polyphasic delimitation of a filamentous marine genus, Capillus gen. nov. (Cyanobacteria, Oscillatoriaceae) with the description of two Brazilian species. Algae 33: 291–304.

    Caires, T.A. (2019). Capilliphycus gen. nov.; validation of "Capillus T.A.Caires, Sant’Anna & J.M.Nunes," inval. (Oscillatoriaceae, Cyanobacteria). Notulae algarum 95: 2 pp.

    Calteau, A., Fewer, D.P., Latifi, A., Coursin, T., Laurent, T. and 5 other authors (2014). Phylum-wide comparative genomics unravel the diversity of secondary metabolism in Cyanobacteria. BMC genomics 15: 977.

    Casamatta, D.A., Gomez, S.R. and Johansen, J.R. (2006). Rexia erecta Gen. et sp. nov. and Capsosira lowei sp. nov., Two Newly Described Cyanobacterial Taxa from the Great Smoky Mountains National Park (USA). Hydrobiologia 561: 13–26.

    Casamatta, D.A., Johansen, J.R., Vis, M.L. and Broadwater, S.T (2005). Molecular and morphological characterization of ten polar and near-polar strains within the Oscillatoriales (Cyanobacteria). Journal of Phycology 41: 421–438.

    Casamatta, D.A., Villanueva, C.D., Garvey, A.D., Stocks, H.S., Vaccarino, M., Dvořák, P., Hašler, P. and Johansen, J.R. (2019). Reptodigitus chapmanii (Nostocales, Hapalosiphonaceae) gen. nov.: a unique Nostocalean (Cyanobacteria) genus based on a polyphasic approach. Journal of Phycology: Epub ahead of print. DOI: 10.1111/jpy.12954.

    Chakraborty, S., Maruthanayagam, V., Achari, A., Pramanik, A., Jaisankar, P. and Mukherjee, J. (2019). Euryhalinema mangrovii gen. nov., sp. nov. and Leptoelongatus litoralis gen. nov., sp. nov. (Leptolyngbyaceae) isolated from an Indian mangrove forest. Phytotaxa 422: 58–74.

    Chavadar, M., Saraf, A., Suradkar, A., Mishra., D, Kumar, N. and Singh, P. (2021). Constrictifilum karadense gen. et sp. nov., a new Nostocalean genus from Maharashtra, India. FEMS Microbiology Letters 368: fnab057.

    Chen, M.-Y., Teng, W.-K., Zhao, L., Hu, C.-X., Zhou, Y.-K., Han, B.-P., Song, L.-R. and Shu, W.-S. (2020). Comparative genomics reveals insights into cyanobacterial evolution and habitat adaptation. The ISME Journal: Epub ahead of print. DOI: 10.1038/s41396-020-00775-z.

    Coutinho, F., Tschoeke1, D.A., Thompson, F. and Thompson, C. (2016a). Comparative genomics of Synechococcus and proposal of the new genus Parasynechococcus. PeerJ 4: e1522.

    Coutinho, F., Dutilh, B., Thompson, C. and Thompson, F. (2016b). Proposal of fifteen new species of Parasynechococcus based on genomic, physiological and ecological features. Arch. Microbiol. 198: 973–986.

    Criscuolo, A. (2019). A fast alignment-free bioinformatics procedure to infer accurate distance-based phylogenetic trees from genome assemblies. Research Ideas and Outcomes 5: e36178.

    Criscuolo, A. and Gribaldo, S. (2011). Large-Scale Phylogenomic Analyses Indicate a Deep Origin of Primary Plastids within Cyanobacteria. Molecular Biology and Evolution 28: 3019–3032.

    da Silva Malone, C.F., Genuário, D.B., Vaz, M.G.M.V., Fiore, M.F. and Sant’Anna, C.L. (2020). Monilinema gen. nov., a homocytous genus (Cyanobacteria, Leptolyngbyaceae) from saline‐alkaline lakes of Pantanal wetlands, Brazil. Journal of Phycology: Epub ahead of print 2020. DOI: 10.1111/jpy.13106.

    de Alvarenga, L.V., Vaz, M.G.M.V., Genuário, D.B., Esteves-Ferreira, A.A., Almeida, A.V.M. and 6 other authors (2018). Extending the ecological distribution of Desmonostoc genus: proposal of Desmonostoc salinum sp. nov., a novel Cyanobacteria from a saline–alkaline lake. International Journal of Systematic and Evolutionary Microbiology 68: 2770–2782.

    de Winder, B., Stal, L.J. and Mur, L.R. (1990). Crinalium epipsammum sp. nov.: A filamentous cyanobacterium with trichomes composed of elliptical cells and containing poly-Β-(1,4) glucan (cellulose). Journal of General Microbiology 136: 1645–1653.

    Di Cesare, A., Cabello-Yeves, P.J., Chrismas, N.A.M., Sánchez-Baracaldo, P., Salcher, M.M. and Calieri, C. (2018). Genome analysis of the freshwater planktonic Vulcanococcus limneticus sp. nov. reveals horizontal transfer of nitrogenase operon and alternative pathways of nitrogen utilization. BMC Genomics 19: 259.

    Dextro, R.B., Moutinho, F.H.M. and Nordi, C.S.F. (2018). Growth and special structures production of Nostoc paludosum (Nostocaceae, Cyanobacteria) under nutrient starvation and different light intensities. Ambiente e Agua - An Interdisciplinary Journal of Applied Science 13: 1–16.

    Duval, C., Hamlaoui, S., Piquet, B., Toutirais, G., Yéprémian, C., Reinhardt, A., Duperron, S., Marie, B., Demay, J. and Bernard, C. (2021). Diversity of cyanobacteria from thermal muds (Balaruc-Les-Bains, France) with the description of Pseudochroococcus coutei gen. nov., sp. nov. FEMS Microbes 2: xtab006.

    Dvořák, P., Hašler, P. and Poulíčková, A. (2019). New insights into the genomic evolution of cyanobacteria using herbarium exsiccatae. European Journal of Phycology DOI: 10.1080/09670262.2019.1638523.

    Dvořák, P., Casamatta, D.A., Poulíčková, A., Hašler, P., Ondřej V. and Sanges, R. (2014). Synechococcus : 3 billion years of global dominance. Molecular Ecology 23: 5538-5551.

    Fiore, M.F., Neilan, B.A., Copp, J.N., Rodrigues, J.L.M., Tsail, S.M., Lee, H. and Trevors, J.T. (2005). Characterization of nitrogen fixing cyanobacteria in the Brazilian Amazon floodplain. Water Res. 39: 5017-5026.

    Flechtner, V.R., Boyer, S.L., Johansen, J.R. and DeNoble, M.L. (2002). Spirirestris rafaelensis gen. et sp. nov. (Cyanophyceae), a new cyanobacterial genus from arid soils. Nova Hedwigia 74: 1-24.

    Gaget, V., Welker, M., Rippka, R. and Tandeau de Marsac, N. (2015). A polyphasic approach leading to the revision of the genus Planktothrix (Cyanobacteria) and its type species, P. agardhii, and proposal for integrating the emended valid botanical taxa, as well as three new species, Planktothrix paucivesiculata sp. nov.ICNP, Planktothrix tepida sp. nov.ICNP, and Planktothrix serta sp. nov.ICNP, as genus and species names with nomenclatural standing under the ICNP. Systematic and Applied Microbiology 38: 141-158.

    Garcia‐Pichel, F., Zehr, J.P., Bhattacharya, D. and Pakrasi, H.B. (2020). What’s in a name? The case of cyanobacteria. Journal of Phycology 56: 1–5.

    Geng, R., Li, W., Chao, A., Guo, X., Li, H., Yu, G. and Li, R. (2021). Establishment of a New Filamentous Cyanobacterial Genus, Microcoleusiopsis gen. nov. (Microcoleaceae, Cyanobacteria), from Benthic Mats in Open Channel, Jiangxi Province, China. Diversity 13, 548.

    Gkelis, S., Rajaniemi, P., Vardaka, E., Moustaka-Gouni, M., Lanaras, T., and Sivonen, K. (2005). Limnothrix redekei (Van Goor) Meffert (Cyanobacteria) strains from Lake Kastoria, Greece form a separate phylogenetic group. Microbial Ecology 49: 176–182.

    González-Resendiz, L., Johansen, J.R., Alba-Lois, L., Segal-Kischinevzky, C., Escobar-Sánchez, V., Garcia, L.F.J., Hauer, T. and León–Tejera, H. (2018). Nunduva, a new marine genus of Rivulariaceae (Nostocales, Cyanobacteria) from marine rocky shores. Fottea 18: 86–105.

    González-Resendiz, L., León-Tejera, H.P., Díaz-Larrea, J., Alba-Lois, L. and Segal-Kischinevzky, C. (2013). Hassallia littoralis sp. nov. (Cyanobacteria, Microchaetaceae) from Mexico’s marine supralittoral based on morphological and molecular evidence. Phytotaxa 137: 35–47.

    Gugger, M., Lyra, C., Henriksen, P., Couté, A., Humbert, J.-F. and Sivonen, K. (2002). Phylogenetic comparison of the cyanobacterial genera Anabaena and Aphanizomenon. International Journal of Systematic and Evolutionary Microbiology 52: 1867–1880.

    Gupta, V., Natarajan, C., Chaudhary, V., Kumar, A., Sharma, E., Sharma, J., Bhatnagar, A.K. and Prasanna, R. (2012). Analyses of diversity among fungicidal Anabaena strains. Journal of Applied Phycology 24: 1395-1405.

    Hašler, P., Dvořák, P. and Poulíčková, A. (2014a). A new genus of filamentous epipelic cyanobacteria, Johansenia. Preslia 86: 81–94.

    Hašler, P., Dvořák, P. and Poulíčková, A. (2014b). Johanseninema, a corrected name for a recently described genus of filamentous epepelic cyanobacteria. Preslia 86: 293–294.

    Hauer, T., Bohunická, M., Johansen, J.R., Mareš, J. and Berrendero-Gomez, E. (2014). Reassessment of the cyanobacterial family Microchaetaceae and establishment of new families Tolypothrichaceae and Godleyaceae. Journal of Phycology 50: 1089-1100.

    Hauer, T., Bohunická, M. and Mühlsteinová, R. (2013). Calochaete gen. nov. (Cyanobacteria, Nostocales), a new cyanobacterial type from the "páramo" zone in Costa Rica. Phytotaxa 109: 36–44.

    Hauerová, R., Hauer, T., Kaštovský, J., Komárek, J., Lepšová-Skácelová and O., Mareš, J. (2021). Tenebriella gen. nov. – The dark twin of Oscillatoria. Molecular Phylogenetics and Evolution 165: 107293.

    Heidari, F., Zima, J., Riahi, H. and Hauer, T. (2018). New simple trichal cyanobacterial taxa isolated from radioactive thermal springs. Fottea:18:137–149.

    Herdman M., Janvier M., Waterbury J.B., Rippka R., Stanier R.Y. and Mandel, M. (1979a). Deoxyribonucleic acid base composition of cyanobacteria. J Gen Microbiol 111: 63–71.

    Herdman M., Janvier M., Rippka R. and Stanier R.Y. (1979b). Genome size of cyanobacteria. J Gen Microbiol 111: 73–85.

    Hrouzek, P., Lukešová A., Mareš, J. and Ventura, S. (2013). Description of the cyanobacterial genus Desmonostoc gen. nov. Including D. muscorum comb. nov. as a distinct, phylogenetically coherent taxon related to the genus Nostoc. FOTTEA 13: 201-213.

    Johansen, J.R., Bohunická, M., Lukešová, A., Hrčková, K., Vaccarino, M.A. and Chesarino, N.M. (2014). Morphological and molecular characterization within 26 strains of the genus Cylindrospermum (Nostocaceae, Cyanobacteria), with descriptions of three new species. Journal of Phycology 50: 187–202.

    Johansen, J.R., Mareš, J., Pietrasiak, N., Bohunická, M., Zima, J. Jr., Štenclová, L. and Hauer, T. (2017). Highly divergent 16S rRNA sequences in ribosomal operons of Scytonema hyalinum (Cyanobacteria). PLoS ONE 12: E0186393.

    Jung, P., D’Agostino, P.M., Brust, K., Büdel, B. and Lakatos, M. (2021). Final Destination? Pinpointing Hyella disjuncta sp. nov. PCC 6712 (Cyanobacteria) Based on Taxonomic Aspects, Multicellularity, Nitrogen Fixation and Biosynthetic Gene Clusters. Life 11: 916.

    Kabirnataj, S., Nematzadeh, G.A., Talebi, A.F., Tabatabaei, M. and Singh, P. (2018). Neowestiellopsis gen. nov, a new genus of true branched cyanobacteria with the description of Neowestiellopsis persica sp. nov. and Neowestiellopsis bilateralis sp. nov., isolated from Iran. Plant Systematics and Evolution 304: 501-510.

    Kabirnataj, S., Nematzadeh, G.A., Talebi, A.F., Saraf, A., Suradkar, A, Tabatabaei, M. and Singh, P. (2020). Description of novel species of Aliinostoc, Desikacharya and Desmonostoc using a polyphasic approach. International Journal of Systematic and Evolutionary Microbiology 70: 3413–3426.

    Kaštovský, J., Gomez, E.B., Hladil, J. and Johansen, J.R.(2014). Cyanocohniella calida gen. et sp. nov. (Cyanobacteria: Aphanizomenonaceae) a new cyanobacterium from the thermal springs from Karlovy Vary, Czech Republic. Phytotaxa 181: 279–292.

    Katoh, H., Furukawa, J., Tomita-Yokotani, K. and Nishi, Y. (2012). Isolation and purification of an axenic diazotrophic drought-tolerant cyanobacterium, Nostoc commune, from natural cyanobacterial crusts and its utilization for field research on soils polluted with radioisotopes. Biochim. Biophys. Acta 1817: 1499-1505.

    Katoh, H., Itoh, S., Shen, J.R. and Ikeuchi, M. (2001). Functional analysis of psbV and a novel c-type cytochrome gene psbV2 of the thermophilic cyanobacterium Thermosynechococcus elongatus strain BP-1. Plant Cell Physiol 42: 599–607.

    Kilgore, C., Johansen, J.R., Hauer, T., Casamatta, D. and Sheil, C.A. (2018). Molecular characterization of Geitleria appalachiana sp. nov. (Nostocales, Cyanobacteria) and formation of Geitleriaceae fam. nov. Fottea, Olomouc 18: 150-163.

    Kim, M., Oh, H.-S., Park, S.-C. and Chun, J.(2014). Towards a taxonomic coherence between average nucleotide identity and 16S rRNA gene sequence similarity for species demarcation of prokaryotes. International Journal of Systematic and Evolutionary Microbiology 64: 346–351.

    Komárek, J. and Anagnostidis, K. (1989). Modern approach to the classification system of cyanophytes. 4 - Nostocales. Archiv für Hydrobiologie Supplement. 82: 247–345.

    Komárek, J., Genuário, D.B., Fiore, M.F. and Elster, J. (2015). Heterocytous cyanobacteria of the Ulu Peninsula, James Ross Island, Antarctica. Polar Biology 38: 475–492.

    Komárek, J., Johansen, J.R., Šmarda, J. and Strunecký, O. (2020). Phylogeny and taxonomy of Synechococcus-like cyanobacteria. Fottea 20: 171–191.

    Komárek, J., Kaštovský, J., Mareš, J. and Johansen, J.R. (2014). Taxonomic classification of cyanoprokaryotes (cyanobacterial genera) 2014, using a polyphasic approach. Preslia 86: 295–335.

    Komárek, J., Nedbalová, L. and Hauer, T. (2011). Phylogenetic position and taxonomy of three heterocytous cyanobacteria dominating the littoral of deglaciated lakes, James Ross Island, Antarctica. Polar Biology 35: 759-774.

    Komárek, J., Zapomělová, E., Šmarda, J., Kopeckỳ, J., Rejmánková, E., Woodhouse, J., Neilan, B.A. and Komárková, J. (2013). Polyphasic evaluation of Limnoraphis robusta, a water-bloom forming cyanobacterium from Lake Atitlán, Guatemala, with a description of Limnoraphis gen. nov. Fottea, Olomouc 13: 39-52.

    Kunisawa, R., and Cohen-Bazire, G. (1970). Mutations of Anacystis nidulans that affect cell division. Arch. Mikrobiol. 71: 49-59.

    Lamprinou, V., Christodoulou, M., Hernandez-Mariné, M., Parmakelis, A. and Economou-Amilli, A. (2016). Spelaeonaias gen. nov., a new true-branched cyanobacterium from Cave Vlychada (Diros, Peloponnese, Greece). Phytotaxa 282: 171-185.

    Lamprinou, V., Hernandez-Mariné, M., Canals, T., Kormas, K., Economou-Amilli, A. and Pantazidou, A. (2011). Morphology and molecular evaluation of Iphinoe spelaeobios gen. nov., sp. nov. and Loriellopsis cavernicola gen. nov., sp. nov., two stigonematalean cyanobacteria from Greek and Spanish caves. International Journal of Systematic and Evolutionary Microbiology 61: 2907-2915.

    Lamprinou, V., Hernandez-Mariné, M., Pachiadaki, M. G., Kormas, K. A., Economou-Amilli, A. and Pantazidou, A. (2013). New findings on the true-branched monotypic genus Iphinoe (Cyanobacteria) from geographically isolated caves (Greece). Fottea 13: 15-23.

    Lamprinou, V., Skaraki, K., Kotoulas, G., Economou-Amilli, A. and Pantazidou, A. (2012). Toxopsis calypsus gen. nov., sp. nov. (Cyanobacteria, Nostocales) from cave ‘Francthi’, Peloponnese, Greece: a morphological and molecular evaluation. International Journal of Systematic and Evolutionary Microbiology 62: 2870-2877.

    Lang, N.J. (1977). Starria zimbabweënsis (cyanophyceae) gen. nov. et sp. nov.: A filament triradiate in tranverse section. Journal of Phycology 13: 288–296.

    Lee, N.-J., Bang, S.-D., Kim, T., Ki, J.-S. and Lee, O.-M. (2021). Pseudoaliinostoc sejongens gen. & sp. nov. (Nostocales, Cyanobacteria) from floodplain soil of the Geum River in Korea based on polyphasic approach. Phytotaxa 479: 55–70.

    Lefler, F.W., Berthold, D.E. and Laughinghouse H.D. (2021). The occurrence of Affixifilum gen. nov. and Neolyngbya (Oscillatoriaceae) in South Florida (USA), with the description of A. floridanum sp. nov. and N. biscaynensis sp. nov. Journal of Phycology 57: 92–110.

    Lehner, J., Zhang, Y. Berendt, S., Rasse, T.M., Forchhammer, K. and Maldener, I. (2011). The morphogene AmiC2 Is pivotal for multicellular development in the Cyanobacterium Nostoc punctiforme. Molecular Microbiology 79: 1655-1669.

    León-Tejera, H.P., González-Resendiz, L., Johansen, J.R., Segal-Kischinevzky, C., Escobar-Sánchez, V. and Alba-Lois, L. (2016). Phylogenetic position reevaluation of Kyrtuthrix and description of a new species K. huatulcensis from Mexico's Pacific coast. Phytotaxa 278: 1-18.

    Lewin, R. A. (1989). Group II. Order Prochlorales. In Bergey’s Manual of Systematic Bacteriology, pp. 1799–1805. Edited by J. T. Staley, M. P. Bryant, N. Pfennig and J. G. Holt. Williams & Wilkins, Baltimore.

    Lopes, V. R., Ramos, V., Martins, A., Sousa, M., Welker, M., Antunes, A. and Vasconcelos, V. M. (2012). Phylogenetic, chemical and morphological diversity of cyanobacteria from Portuguese temperate estuaries. Marine Environmental Research 73: 7-16.

    Ludwig, W., Strunk, O., Westram, R., Richter, L., Meier, H., Yadhukumar, A., Buchner, A. and 25 other authors (2004). ARB: A Software Environment for Sequence Data. Nucleic Acids Research 32: 1363-1371.

    Lundgren, P., Janson, S., Jonasson, S., Singer, A. and Bergman, B. (2005). Unveiling of novel radiations within Trichodesmium cluster by hetR gene sequence analysis. Applied and Environmental Microbiology 71: 190–196.

    Mai, T., Johansen, J.R., Pietrasiak, N., Bohunická, M. and Martin, M.P. (2018). Revision of the Synechococcales (Cyanobacteria) through recognition of four families including Oculatellaceae fam. nov. and Trichocoleaceae fam. nov. and six new genera containing 14 species. Phytotaxa 365: 1-59.

    Mareš, J. (2018). Multilocus and SSU rRNA gene phylogenetic analyses of available cyanobacterial genomes, and their relation to the current taxonomic system. Hydrobiologia 811: 19–34.

    Mareš, J., Johansen, J.R., Hauer, T., Zima, J., Ventura, S., Cuzman, O., Tiribilli, R., Kaštovský, J. and Collier, J. (2019). Taxonomic resolution of the genus Cyanothece (Chroococcales, Cyanobacteria), with a treatment on Gloeothece and three new genera, Crocosphaera, Rippkaea and Zehria. Journal of Phycology 55: 578-610.

    Mareš, J., Lara, Y., Dadáková, I., Hauer, T., Uher, B., Wilmotte, A. and Kaštovský, J. (2015). Phylogenetic analysis of cultivation-resistant terrestrial cyanobacteria with massive sheaths (Stigonema spp. and Petalonema alatum, Nostocales, Cyanobacteria) using single-cell and filament sequencing of environmental samples. Journal of Phycology 51: 288–297.

    Martins, M.D., Rigonato, J., Branco, L.H.Z. and Taboga, S.R. (2016). Proposal of Ancylothrix gen. nov., a new genus of Phormidiaceae (Cyanobacteria, Oscillatoriales) based on a polyphasic approach. International Journal of Systematic and Evolutionary Microbiology 66: 2396–2405.

    McGregor, G.B. and Sendall, B.C. (2015). Phylogeny and toxicology of Lyngyba wollei (Cyanobacteria/Oscillatoriales) from north-eastern Australia, with a description of Microseira gen. nov. Journal of Phycology 51: 109-119.

    McGregor, G.B. and Sendall, B.C. (2019). Potamosiphon australiensis gen. nov., sp. nov. (Oscillatoriales), a new filamentous cyanobacterium from subtropical north-eastern Australia. Phytotaxa 387: 77-93.

    Meffert, M.-E. (1988). Limnothrix Meffert nov. gen. The unsheathed planktic cyanophycean filaments with polar and central gas vacuoles. Algological Studies 50–53: 269–276.

    Micallef, M.L., Sharma, D., Bunn, B.M., Gerwick, L., Viswanathan R. and Moffitt, M.C. (2014). Comparative analysis of hapalindole, ambiguine and welwitindolinone gene clusters and reconstitution of indole-isonitrile biosynthesis from cyanobacteria. BMC microbiology 14: 213.

    Mikhailyuk, T., Vinogradova, O., Holzinger, A., Glaser, K., Samolov, E., and Karsten, U. (2019). New record of the rare genus Crinalium Crow (Oscillatoriales, Cyanobacteria) from sand dunes of the Baltic Sea, Germany: epitypification and emendation of Crinalium magnum Fritsch et John based on an integrative approach. Phytotaxa 400: 165-179.

    Miscoe, L.H., Johansen, J.R., Kociolek, J.P., Pietrasiak, N., Sherwood, A.R. and Vaccarino, M.A. (2016). Diatom flora and cyanobacteria from caves on Kauai, Hawaii. Bibliotheca Phycologica 120: 3-152.

    Moore, K.R., Magnabosco, C., Momper, L., Gold, D.A., Bosak, T. and Fournier, G.P. (2019). An Expanded Ribosomal Phylogeny of Cyanobacteria Supports a Deep Placement of Plastids. Frontiers in Microbiology 10: 1612.

    Moreira, D., Tavera, R., Benzerara, K., Skouri-Panet, F., Couradeau, E. and 5 other authors (2017). Description of Gloeomargarita lithophora gen. nov., sp. nov., a thylakoid-bearing, basal-branching cyanobacterium with intracellular carbonates, and proposal for Gloeomargaritales ord. nov. International Journal of Systematic and Evolutionary Microbiology 67: 653–658.

    Moro, I., Rascio, N., La Rocca, N., Di Bella, M. and Andreoli, C. (2007). Cyanobacterium aponinum, a new Cyanoprokaryote from the microbial mat of  Euganean Thermal Springs (Padua, Italy). Algological Studies 123: 1–15.

    Nakayama, T., Kamikawa, R., Tanifuji, G., Kashiyama, Y., Ohkouchi, N., Archibald, J.M. and Inagaki, Y. (2014). Complete genome of a nonphotosynthetic cyanobacterium in a diatom reveals recent adaptations to an intracellular lifestyle. Proceedings of the National Academy of Sciences 111: 11407–11412.

    Nakayama, T. and Inagaki, Y. (2017). Genomic divergence within non-photosynthetic cyanobacterial endosymbionts in rhopalodiacean diatoms. Scientific Reports 7: 1–8.

    Nazifi, E., Wada, N., Asano, T., Nishiuchi, T., Iwamuro, Y., Chinaka, S., Matsugo,S. and Sakamoto,T. (2015). Characterization of the chemical diversity of glycosylated mycosporine-like amino acids in the terrestrial cyanobacterium Nostoc commune. J. Photochem. Photobiol. B, Biol. 142: 154-168.

    Novis, P.M. and Visnovsky, G. (2011). Novel alpine algae from New Zealand: Cyanobacteria. Phytotaxa 22: 1-24.

    Nowicka-Krawczyk, P., Mühlsteinová, R., and Hauer, T. (2019). Detailed characterization of the Arthrospira type species separating commercially grown taxa into the new genus Limnospira (Cyanobacteria). Scientific Reports 9: 694.

    Nowruzi, B. and Shalygin, S. (2021). Multiple phylogenies reveal a true taxonomic position of Dulcicalothrix alborzica sp. nov. (Nostocales, Cyanobacteria). Fottea 21: 235–246.

    Nowruzi, B. and Soares, F. (2021). Alborzia kermanshahica gen. nov., sp. nov. (Chroococcales, Cyanobacteria), isolated from paddy fields in Iran. International Journal of Systematic and Evolutionary Microbiology 71: 004828.

    Ondov, B.D., Treangen, T.J., Melsted, P., Mallonee, A.B., Bergman, N.H., Koren, S. and Phillippy, A.M. (2016). Mash: fast genome and metagenome distance estimation using MinHash. Genome Biology 17: 132.

    Osorio-Santos, K., Pietrasiak, N., Bohunická, M., Miscoe, L.H., Kováčik, L., Martin, M.P. and Johansen, J.R. (2014). Seven new species of Oculatella (Pseudanabaenales, Cyanobacteria): taxonomically recognizing cryptic diversification. European Journal of Phycology 49: 450–70.

    Palińska, K.A., Abed, R.M.M., Charpy, L., Langlade, M.-J., Beltrán-Magos, Y. and Golubic, S. (2015). Morphological, genetic and physiological characterization of Hydrocoleum, the most common benthic cyanobacterium in tropical oceans. European Journal of Phycology 50: 139-154.

    Palińska, K.A., Thomasius, C.F., Marquardt, J. and Golubic, S. (2006). Phylogenetic evaluation of cyanobacteria preserved as historic herbarium exsiccata. International Journal of Systematic and Evolutionary Microbiology 56: 2253-2263.

    Parks, D.H., Imelfort, M., Skennerton, C.T., Hugenholtz, P. and Tyson, G.W. (2015). CheckM: assessing the quality of microbial genomes recovered from Isolates, single cells, and metagenomes. Genome Research 25: 1043-1055.

    Perkerson III, R.B., Johansen, J.R., Kováčik, L., Brand, J., Kaštovský, J. and Casamatta, D.A. (2011). A unique Pseudanabaenalean (cyanobacteria) genus Nodosilinea gen. nov. based on morphological and molecular data. Journal of Phycology 47: 1397-1412.

    Pietrasiak, N., Osorio-Santos, K., Shalygin, S., Martin, M.P. and Johansen, J.R. (2019). When is a lineage a species? A case study in Myxacorys gen. nov. (Synechococcales: Cyanobacteria) with the description of two new species from the Americas. Journal of Phycology 55: 976-996.

    Pietrasiak, N., Reeve, S., Osorio‐Santos, K., Lipson, D. and Johansen, J.R. (2021). Trichotorquatus gen. nov. - a new genus of soil cyanobacteria discovered from American drylands. Journal of Phycology (in press).

    Podell, S., Blanton, J.M., Oliver, A., Schorn, M.A., Agarwal, V., Biggs, J.S., Moore, B.S. and Allen, E.E. (2020). A genomic view of trophic and metabolic diversity in clade-specific Lamellodysidea sponge microbiomes. Microbiome 8: 97.

    Potts, M. (1997). Etymology of the genus name Nostoc (cyanobacteria). International Journal of Systematic and Evolutionary Microbiology 47: 584.

    Price, M.N., Dehal, P.S., and Arkin, A.P. (2009). FastTree: computing large minimum-evolution trees with profiles instead of a distance matrix. Molecular Biology and Evolution 26: 1641-1650.

    Pushkareva, E., Pessi, I.S., Wilmotte, A., Elster, J. and Häggblom, M. (2015). Cyanobacterial community composition in Arctic soil crusts at different stages of development. FEMS Microbiology Ecology 91: fiv143.

    Raabová, L., Kovacik, L., Elster, J. and Strunecký, O. (2019). Review of the genus Phormidesmis (Cyanobacteria) based on environmental, morphological, and molecular data with description of a new genus Leptodesmis. Phytotaxa 395: 1–16.

    Ragon, M., Fontaine, M.C., Moreira, D. and López-García, P. (2012). Different biogeographic patterns of prokaryotes and microbial eukaryotes in epilithic biofilms. Molecular Ecology 21: 3852–3868.

    Rajaniemi, P., Hrouzek, P., Kastovska, K., Willame, R., Rantala, A., Hoffmann, L., Komárek, J. and Sivonen, K. (2005). Phylogenetic and morphological evaluation of the genera Anabaena, Aphanizomenon, Trichormus and Nostoc (Nostocales, Cyanobacteria). International Journal of Systematic and Evolutionary Microbiology 55: 11-26.

    Řeháková, K., Johansen, J.R., Casamatta, D.A. Jr., Li, X.S. and Vincent, J. (2007). Morphological and molecular characterization of selected desert soil cyanobacteria; three species new to science including Mojavia pulchra gen. et sp. nov. Phycologia 46: 481-502.

    Rigonato, J., Sant’Anna, C.L., Giani, A., Azevedo, M.T.P., Gama, W.A., Viana, V.F.L., Fiore, M.F. and Werner, V.R. (2018). Sphaerocavum: a coccoid morphogenus identical to Microcystis in terms of 16S rDNA and ITS sequence phylogenies. Hydrobiologia 811: 35-48.

    Rippka, R. and Cohen-Bazire, G. (1983). The cyanobacteriales: A legitimate order based on the type strain Cyanobacterium stanieri? Annales de l'Institut Pasteur. Microbiology 134B: 21–36.

    Rippka, R., Deruelles, J., Waterbury, J.B., Herdman, M. and Stanier, R.Y. (1979). Generic assignments, strain histories and properties of pure cultures of cyanobacteria. Journal of General Microbiology 111: 1-61.

    Roldan, M., Ramirez, M., del Campo, J., Hernandez-Mariné, M. and Komárek, J. (2013). Chalicogloea cavernicola gen. nov., sp. nov. (Chroococcales, Cyanobacteria), from low-light aerophytic environments: combined molecular, phenotypic and ecological criteria. International Journal of Systematic and Evolutionary Microbiology 63: 2326-2333.

    Salazar, V.W., Tschoeke, D.A., Swings, J., Cosenza, C.A., Mattoso, M., Thompson, C.C. and Thompson, F.L. (2020). A new genomic taxonomy system for the Synechococcus collective. Environmental Microbiology 22: 4557–4570.

    Samylina, O.S., Sinetova, M.A., Kupriyanova, E.V., Starikov, A.Y., Sukhacheva, M.V., Dziuba, M.V. and Tourova, T.P. (2021). Ecology and biogeography of the ‘marine Geitlerinema’ cluster and a description of Sodalinema orleanskyi sp. nov., Sodalinema gerasimenkoae sp. nov., Sodalinema stali sp. nov. and Baaleninema simplex gen. et sp. nov. (Oscillatoriales, Cyanobacteria). FEMS Microbiology Ecology 97: fiab104.

    Sant’Anna, C.L., Gama, W.A., Rigonato, J., Correa, G., Mesquita, M.C.B. and Marinho, M.M. (2019). Phylogenetic connection among close genera of Aphanizomenonaceae (Cyanobacteria): Amphiheterocytum gen. nov., Cylindrospermopsis and Sphaerospermopsis. Algal Research 37: 205–214.

    Saraf, A., Dawda, H.G., Suradkar, A., Behere, I., Kotulkar, M., Shaikh, Z.M., Kumat, A., Batule, P., Mishra, D. and Singh, P. (2018). Description of two new species of Aliinostoc and one new species of Desmonostoc from India based on the Polyphasic Approach and reclassification of Nostoc punensis to Desmonostoc punense comb. nov. FEMS Microbiology Letters 365.

    Saraf, A.G., Dawda, H.G. and Singh, P. (2019a). Desikacharya gen. nov., a phylogenetically distinct genus of Cyanobacteria along with the description of two new species, Desikacharya nostocoides sp. nov. and Desikacharya soli sp. nov., and reclassification of Nostoc thermotolerans to Desikacharya thermotolerans comb. nov. International Journal of Systematic and Evolutionary Microbiology 69: 307-315.

    Saraf, A., Dawda, H.G. and Singh, P. (2019b). Validation of the genus Desikacharya gen. nov. (Nostocaceae, Cyanobacteria) and three included species. Notulae algarum 107: 3 pp.

    Saraf, A., Suradkar, A., Dawda, H.G., Gaysina, L.A., Gabidullin, Y. and 5 other authors (2019c). Phylogenetic complexities of the members of Rivulariaceae with the re-creation of the family Calotrichaceae and description of Dulcicalothrix necridiiformans gen nov., sp nov., and reclassification of Calothrix desertica. FEMS Microbiology Letters 366: Epub ahead of print.

    Schirrmeister, B.E., Antonelli, A. and Bagheri, H.C. (2011). The origin of multicellularity in cyanobacteria. BMC Evolutionary Biology 11: 45.

    Schorn, M.A., Jordan, P.A., Podell, S., Blanton, J.M., Agarwal, V., Biggs, J.S., Allen, E.E. and Moore, B.S. (2019). Comparative Genomics of Cyanobacterial Symbionts Reveals Distinct, Specialized Metabolism in Tropical Dysideidae Sponges. mBio 10: e00821-19.

    Scotta Hentschke, G.S., Johansen, J.R., Pietrasiak, N., Fiore, M.D.F., Rigonato, J., Sant'Anna, C. and Komárek, J. (2016). Phylogenetic placement of Dapisostemon gen. nov. and Streptostemon, two tropical heterocytous genera (Cyanobacteria). Phytotaxa 245: 129.

    Scotta Hentschke, G., Johansen, J.R., Pietrasiak, N., Rigonato, J., Fiore, M.F. and Sant’Anna, C.L. (2017). Komarekiella atlantica gen. et sp. nov. (Nostocaceae, Cyanobacteria): a new subaerial taxon from the Atlantic Rainforest and Kauai, Hawaii. Fottea 17: 178-190.

    Sendall, B.C. and McGregor, G.B. (2018). Cryptic diversity within the Scytonema complex: Characterization of the paralytic shellfish toxin producer Heteroscytonema crispum, and the establishment of the family Heteroscytonemataceae (Cyanobacteria/Nostocales). Harmful Algae 80: 158-170.

    Shafik, H.M., Vörös, L., Sprőber., P, Présing, M. and Kovács, A.W. (2003). Some special morphological features of Cylindrospermopsis raciborskii in batch and continuous cultures. Hydrobiologia 506–509: 163–167.

    Shalygin, S., Huang, I.-S., Elle, A., Burkholder, J. and Zimba, P.V. (2019a). Odorella benthonica gen. & sp. nov. (Pleurocapsales, Cyanobacteria): an odor and prolific toxin producer isolated from a California aqueduct. Journal of Phycology. Epub ahead of print. DOI: 10.1111/jpy.12834.

    Shalygin, S., Kavulic, K.J., Pietrasiak, N., Bohunická, M., Vaccarino, M.A., Chesarino, N.M. and Johansen, J.R. (2019b). Neotypification of Pleurocapsa fuliginosa and epitypification of P. minor (Pleurocapsales): resolving a polyphyletic cyanobacterial genus. Phytotaxa 392: 245-263.

    Shalygin, S., Shalygina, R.R., Redkina, V.V., Gargas, C.B. and Johansen, J.R. (2020). Description of Stenomitos kolaenensis and S. hiloensis sp. nov. (Leptolyngbyaceae, Cyanobacteria) with an emendation of the genus. Phytotaxa 440: 108–128.

    Shih, P.M., Wu, D., Latifi, A., Axen, S.D., Fewer, D.P. and 18 other authors (2013). Improving the coverage of the cyanobacterial phylum using diversity-driven genome sequencing. Proceedings of the National Academy of Sciences 110: 1053–1058.

    Silva, C.S.P., Genuário, D.B., Vaz, M.G.M.V. and Fiore, M.F. (2014). Phylogeny of culturable cyanobacteria from Brazilian mangroves. Systematic and Applied Microbiology 37: 100-112.

    Smith, H.D., Baqué, M., Duncan, A.G., Lloyd, C.R., McKay, C.P. and Billi, D. (2014). Comparative analysis of cyanobacteria inhabiting rocks with different light transmittance in the Mojave Desert: a Mars terrestrial analogue. International Journal of Astrobiology 13: 271–277.

    Soares, F., Ramos, V., Trovão, J., Cardoso, S.M., Tiago, I. and Portugal, A. (2020). Parakomarekiella sesnandensis gen. et sp. nov. (Nostocales, Cyanobacteria) isolated from the Old Cathedral of Coimbra, Portugal (UNESCO World Heritage Site). European Journal of Phycology 2020: in press.

    Soo, R.M., Skennerton, C.T., Sekiguchi, Y., Imelfort, M., Paech, S.J., Dennis, P.G., Sheen, J.A., Parks, D.H., Tyson, G.W. and Hugenholtz, P. (2014). An expanded genomic representation of the phylum Cyanobacteria. Genome Biol Evol 6: 1031–1045.

    Stackebrandt, E., Frederiksen, W., Garrity, G.M., Grimont, P.A.D., Kampfer, P. and 8 other authors (2002). Report of the ad hoc committee for the re-evaluation of the species definition in bacteriology. International Journal of Systematic and Evolutionary Microbiology 52: 1043–1047.

    Stackebrandt, E. and Goebel, B.M. (1994). Taxonomic note: A place for DNA-DNA reassociation and 16S rRNA sequence analysis in the present species definition in bacteriology. International Journal of Systematic Bacteriology 44: 846-849.

    Stackebrandt, E. and Ebers, J. (2006). Taxonomic parameters revisited: tarnished gold standards. Microbiology Today 06: 153-154.

    Stoyanov, P., Moten, D., Mladenov, R., Dzhambazov, B. and Teneva, I. (2014). Phylogenetic Relationships of Some Filamentous Cyanoprokaryotic Species. Evolutionary Bioinformatics 10: 39–49.

    Strunecký, O., Bohunická, M., Johansen, J. R., Čapková, K., Raabová, L., Dvořák, P. and Komárek, J. (2017). A revision of the genus Geitlerinema and a description of the genus Anagnostidinema gen. nov. (Oscillatoriophycidae, Cyanobacteria). Fottea 17: 114-126.

    Sukenik, A., Kaplan-Levy, R.N., Viner-Mozzini, Y., Quesada, A. and Hadas, O. (2013). Potassium deficiency triggers the development of dormant cells (akinetes) in Aphanizomenon ovalisporum (Nostocales, Cyanoprokaryota). Journal of Phycology 49: 580–587.

    Taton, A., Grubisic, S., Ertz, D., Hodgson, D.A., Piccardi, R. and 6 other authors (2006). Polyphasic study of Antarctic cyanobacterial strains. Journal of Phycology 42: 1257–1270.

    Tronholm, A. and Engene, N. (2019). Moorena gen. nov., a valid name for "Moorea Engene & al." nom. inval. (Oscillatoriaceae, Cyanobacteria). Notulae Algarum 122: 1-2

    Turicchia, S., Ventura, S., Komárková, J. and Komárek, J. (2009). Taxonomic evaluation of cyanobacterial microflora from alkaline marshes of northern Belize. 2. Diversity of oscillatorialean genera. Nova Hedwigia 89: 165–200.

    Uyeda, J.C., Harmon, L.J. and Blank, C.E. (2016). A Comprehensive Study of Cyanobacterial Morphological and Ecological Evolutionary Dynamics through Deep Geologic Time. PLOS ONE 11: e0162539.

    Vera, I.C.D.P., Lownsbery, B.F. and Maggenti, A.R. (1983). Rhizonema sequoiae n. gen. n. sp. from Coast Redwood Sequoia sempervirens (D. Don) Endl. Journal of nematology 15: 460-466.

    Wacklin, P., Hoffmann, L., and Komárek, J. (2009). Nomenclatural validation of the genetically revised cyanobacterial genus Dolichospermum (Ralfs ex Bornet et Flahault) comb. nova. Fottea 9: 59–64.

    Walter, J.M., Coutinho, F.H., Dutilh, B.E., Swings J., Thompson, F.L. and Thompson, C.C. (2017). Ecogenomics and Taxonomy of Cyanobacteria Phylum. Frontiers in Microbiology 8: 2132.

    Walter, J.M., Coutinho, F.H., Leomil, L., Hargreaves, P.I., Campeão, M.E. and 13 other authors (2020). Ecogenomics of the Marine Benthic Filamentous Cyanobacterium Adonisia. Microbial Ecology: 80: 249-265.

    Wang, Y., Cai, F., Jia, N. and Li, R. (2019). Description of a novel coccoid cyanobacterial genus and species Sinocapsa zengkensis gen. nov. sp. nov. (Sinocapsaceae, incertae sedis), with taxonomic notes on genera in Chroococcidiopsidales. Phytotaxa 409: 146-160.

    Wang, Y., Jia, N., Geng, R., Yu, G. and Li, R. (2021). Phylogenetic insights into chroococcus‐ like taxa (Chroococcales, Cyanobacteria), describing Cryptochroococcus tibeticus gen. nov. sp. nov. and Limnococcus fonticola sp. nov. from Qinghai‐Tibet plateau. Journal of Phycology In press.

    Wayne, L.G., Brenner, D.J., Colwell, R.R., Grimont, P.A.D., Kandler, P. and 6 other authors (1987). Report of the ad hoc committee on reconciliation of approaches to bacterial systematics. International Journal of Systematic Bacteriology 37: 463–464.

    Yarza, P., Yilmaz, P., Pruesse, E., Glöckner, F.O., Ludwig, W., Schleifer, K.-H., Whitman, W.B., Uzébey, J., Amann, R. and Rosselló-Móra, R. (2014). Uniting the classification of cultured and uncultured bacteria and archaea using 16S rRNA gene sequences. Nature Reviews Microbiology 12: 635-645.

    Yooseph, S., Nealson, K.H., Rusch, D.B., McCrow, J.P., Dupont, C.L. and 15 further authors (2010). Genomic and functional adaptation in surface ocean planktonic prokaryotes. Nature 468: 60-66.

    Zapomělová, E., Hrouzek, P., Řeháková, K., Šabacká, M., Stibal, M., Caisová, L., Komárková, J. and Lukešová, A. (2008). Morphological variability in selected heterocystous cyanobacterial strains as a response to varied temperature, light Intensity and medium composition. Folia Microbiologica 53: 333-341.

    Zapomělová, E., Jezberová, J., Hrouzek, P., Hisem, D., Reháková, K. and Komárková, J. (2009). Polyphasic characterization of three strains of Anabaena reniformis and Aphanizomenon aphanizomenoides (Cyanobacteria) and their reclassification to Sphaerospermum gen. nov. (incl. Anabaena kisseleviana). Journal of Phycology 45: 1363–1373.

    Zapomělová, E., Jezberová, J., Hrouzek, P., Hisem, D., Reháková, K. and Komárková, J. (2010). (Correction) Polyphasic characterization of three strains of Anabaena reniformis and Aphanizomenon aphanizomenoides (Cyanobacteria) and their reclassification to Sphaerospermum gen. nov. (incl. Anabaena kisseleviana). (45:1363-73). Journal of Phycology 46: 415.

    Zimba, P.V., Huang, I.-S., Foley, J.E. and Linton, E.W. (2017). Identification of a new-to-science cyanobacterium, Toxifilum mysidocida gen. nov. and sp. nov. (Cyanobacteria, Cyanophyceae). Journal of Phycology 53: 188-197.

    Zwickl, D.J. and Hillis, D.M. (2002). Increased taxon sampling greatly reduces phylogenetic error. Syst. Biol. 51: 588-598.