Genomes of nine biofilm-forming filamentous strains of Cyanobacteria (genera Jaaginema, Scytonema, and Karukerafilum gen. nov.) isolated from mangrove habitats of Guadeloupe (Lesser Antilles)

Abstract Biofilm-forming cyanobacteria are abundant in mangrove ecosystems, colonizing various niches including sediment surface and periphyton where they can cover large areas, yet have received limited attention. Several filamentous isolates were recently isolated from Guadeloupe, illustrating the diversity and novelty present in these biofilms. In this study, nine strains belonging to three novel lineages found abundantly in Guadeloupe biofilms are characterized by genome sequencing, morphological and ultrastructural examination, metabolome fingerprinting and searched for secondary metabolites biosynthesis pathways. Assignation of two lineages to known genera is confirmed, namely Scytonema and Jaaginema. The third lineage corresponds to a new Coleofasciculales genus herein described as Karukerafilum gen. nov. The four strains belonging to this genus group into two subclades, one of which displays genes necessary for nitrogen fixation as well as the complete pathway for geosmin production. This study gives new insights into the diversity of mangrove biofilm-forming cyanobacteria, including genome-based description of a new genus and the first genome sequence available for the genus Jaaginema.


Introduction
Biofilm-forming c y anobacteria (phylum Cy anobacteriota) can contribute significantly to ecosystems primary production, in particular in the tr opics wher e they can cov er v ery lar ge ar eas of sediment, r oc ks or a variety of alive or dead biological surfaces .T hey ar e mainl y filamentous in biofilms, and can either colonize freshwater or shallow-water marine environments such as mangrove or rivers (Guidi-Rontani et al. 2014, Alvarenga et al. 2015, Shah et al. 2017 ).These biofilms also harbor many other prokaryotic and eukaryotic lineages that interact closely, competing or cooperating in nutrient cycling, and production of protective compounds (Rigonato et al. 2013, Basak et al. 2016, Allard et al. 2020 ).Aside from carbon fixation, c y anobacteria in biofilms can contribute to nutrient cycling through nitrogen fixation, accumulation of calcium, ma gnesium and phosphor ous (Lov eloc k et al. 2010 ).Inter estingl y, gr azing does not seem to cause massiv e dama ge on biofilms, suggesting the existence of defense molecules to which c y anobacteria, as producers of various bioactive compounds , ma y contribute activ el y (Demay et al. 2019 ).Indeed, these phototrophs are known to produce a wide diversity of bioactive compounds (Dema y et al. 2019 ).T hese r ange fr om c y anotoxins that are of major significance to ecosystem, animal and human health, to molecules of pharmacological inter est suc h as Br entuxymab v e-dotin, based on dolastatin 10 from Symploca , which reached the market for the treatment of Hodgkin's lymphoma (Mi et al. 2017, Shah et al. 2017, Demay et al. 2019 ).
Aquatic c y anobacteria fr om tr opical ar eas ar e r egarded as a r elativ el y accessible yet unta pped source of ne w taxa and biomolecules (Alivisatos et al. 2015, Allard et al. 2020 ), and incr easing efforts ar e being deplo y ed to c har acterize their div ersity.Curr entl y, 184 r efer ence genomes ar e av ailable fr om c y anobacteria species, a subset of the 5700 described c y anobacterial species, itself certainly a small subset of the group's true diversity, whic h r emains v astl y under estimated (estimates up to 8000 species have been proposed) (Nabout et al. 2013, Komarek et al. 2014, Strunecký et al. 2023 ).The tropical regions and the marine benthic compartment ar e particularl y under-explor ed compared to the potential diversity their harbor (Alvarenga et al. 2015 ).In a recent study, a high diversity of novel benthic biofilmforming c y anobacterial lineages w as evidenced in coastal habitats of Guadeloupe (Lesser Antilles) using strain isolation and 16S rRNA compar ativ e gene sequence anal ysis (Duperr on et al. 2020 ).Among these, se v er al potential species or gener a wer e r epr esented by m ultiple distinct isolates.Further exploring the diversity (e.g.taxonomical, inter-or intraspecific), characteristics and potential of these strains requires whole length genome se-quencing, mor phological, ultr astructur al and metabolomic analysis.Concerning genome sequencing, more and more studies point to w ar ds high heterogeneity in genome contents among closel y r elated str ains, emphasizing the need to sequence beyond individual strains genomes, but rather the genomes of multiple closel y r elated str ains to get a glimpse into the so-called pangenome of a species-le v el taxon (Meyer et al. 2017, Pérez-Carrascal et al. 2019, Willis and Woodhouse 2020 ).Evaluating this micr o-scale div ersity is k e y to understanding the actual breadth of a species ecological niche (Dvo řák et al. 2023, Halary et al. 2023 ).
In this study, three novel lineages of filamentous c y anobacteria isolated from coastal habitats in Guadeloupe were explored by sequencing the genomes of two to four strains per lineage and by studying their mor phology, ultr astructur e and metabolomic profiles .T hese lineages were provisionally assigned to genera Jaaginema , Oscillatoria and Scytonema and were selected because the y re present a significant fraction of the bacteria occurring in the biofilms from which they were isolated (Duperron et al. 2020 ), se v er al str ains ar e av ailable, and because they r epr esent linea ges for which little genomic information is currently available in the liter atur e. Genome of eac h str ain is sequenced, and a comparativ e anal ysis using a pol yphasic a ppr oac h is conducted to ascertain their taxonomic affiliation, identify their metabolic capabilities, and document inter-str ain v ariability, with a focus on secondary metabolites biosynthesis pathwa ys .T his study pro vides the first genomes from Guadeloupe strains, and the first Jaaginema genomes .T he strains previously assigned to Oscillatoria appear to belong to a new genus, with Karukerafilum mangrovensis herein described as r epr esentativ e species.Intr a-clade genomic v ariability r e v eals major differ ences between closel y r elated str ains that may explain the intra-and inter-taxa diversity and the ecological success of these c y anobacteria.

Origin of strains
The nine str ains anal yzed in this study were isolated from three distinct locations in Guadeloupe (Duperron et al. 2020 ).Dense gr een-to-br own benthic or periphytic biofilms were sampled in July 2018 in the station of Manche-à-Eau lagoon close to Rhizophora mangle roots, and epiphytic biofilms were collected from two stations located alongside the Canal des Rotours, a 6 kmlong canal connecting the marine lagoon to the city of Morneà-l'eau thr ough v arious mangr ov e ( Supplementary Fig. 1 ).In accordance with Article 17, par a gr a ph 2, of the Nago y a Protocol on Access and Benefit-sharing, a sampling permit was issued and published ( https:// absch.cbd.int/en/ database/ CNA/ ABSCH-CNA-FR-240495 ).Back to the laboratory, a fraction of each biofilm was fr ozen for comm unity c har acterization (see below).Another fr action was examined under a binocular and individual c y anobacterial morphotypes were isolated manually to plates containing solid Z8 medium (Rippka 1988 ) containing 0, 20, and 35 g.L −1 salt (Instant Ocean, Aquarium Systems, Fr ance).Cultur es wer e stabilized and maintained in the Paris Museum Collection (PMC) (Duperron et al. 2020 ).Among the recovered culturable strains (described in Duperron et al. 2020 ), nine were selected for subsequent genome sequencing.For simplification purposes, these strains will be referred in the rest of the manuscript as PMC 1050, PMC 1051, PMC 1068, PMC 1076, PMC 1069, PMC 1070, PMC 1078, PMC 1079, and PMC 1080 (strains origin are summarized in Table 1 ).

Metabarcoding and biofilm composition
DN A w as extr acted fr om the 3 biofilms using the ZymoBIOMICS Fecal/Soil Kit (Zymo Resear ch, CA) accor ding to the manufacturer's instructions, including a 3 min disruption of cells using ceramic beads.PCR using universal primers to amplify the V4-V5 region of the 16S rRNA-encoding gene was performed as described in (Duperron et al. 2020), using primers 515F and 926 R (P ar ada et al. 2016, Newman and Cragg 2017 ) and sequenced on an Illumina MiSeq platform (2 ×300 bp, paired-end sequencing, Genoscr een, Fr ance).Compan y-pr ovided moc k comm unities of known composition were used as an internal control for the whole sequencing process .Ra w r eads wer e deposited into the GENBANK Sequence Read Arc hiv e (SRA, Biopr oject PRJNA994497, Biosamples SAMN36430094-6).
Sequence analysis was performed using QIIME2 (Ca por aso et al. 2012, Callahan et al. 2017 ).Amplicon Sequence Variants (ASVs) were identified using DEBLUR (Amir et al. 2017 ) using default par ameters, i.e. a maximal pr obability for indels of 0.01 and mean r ead err or r ate of 0.5% for normalization.Chimeric sequences were identified and discarded using UCHIME ( de novo c himer a detection) (Edgar et al. 2011 ), and the taxonomic affiliations were obtained by using the sklearn-based classifier (GreenGenes 13-8-99 r elease).Sequences matc hing "Eukaryota", "Chlor oplast" and "Mitoc hondria" wer e discarded.AVSs corr esponding or highl y similar to the nine strains sequenced in this study were searched for, and their abundance among total reads was computed.

Cultiv a tion and strains genome sequencing
Biomass was pr oduced fr om the nine selected isolates for two months in increasing volumes of liquid Z8 or Z8X media (25 ± 1 • C; 15 μmol.m −2 .s−1 white light; 16 h light: 8 h dark) with 20 g.L −1 Instant Ocean salt (Aquarium Systems, France) (Rippka 1988 ).DNA was then extracted from culture-derived biomass using the Zymo-BIOMICS Fecal/Soil Kit (Zymo Resear ch, CA) follo wing manufacturer's instructions including a 3 min disruption of cells using cer amic beads.Concentr ations wer e measur ed using Nanodr op and Qubit (Thermo Fisher).DNA sequencing was performed using an Illumina MiSeq platform on 2 ×250 bp reads libraries (Genoscreen, F rance).Assemblies w ere achieved using SPAdes v3.15.4 (with 'meta' option and default parameters) (Bankevich et al. 2012 ) and the resulting scaffolds were then taxonomically annotated using CAT (with default parameters) (von Meijenfeldt et al. 2019 ).All scaffolds affiliated to Cyanobacteria constituted the present genomes, for which completeness and contamination were assessed using CheckM (Parks et al. 2015 ).

Gene annotation and comparati v e genome analysis
Coding DNA sequences were predicted using Prodigal (Hyatt et al. 2010 ) and functionally annotated using eggnog-mapper v2 (Huerta-Cepas et al. 2019 ).Clusters of orthologous genes among str ains wer e determined with Orthofinder (Emms and K ell y 2019 ).Secondary metabolite biosynthesis gene clusters were identified using antiSMASH (Blin et al. 2021 ).Metagenome sequences from the cultur es, and Meta genome-Assembled-Genomes (MAGs) wer e deposited in the SRA database (Bioprojet: PRJNA994497, Table 1 ).

Phylogenomic inference
All av ailable r efer ence genomes of Cyanobacteria with a complete assembl y le v el in December 2021 were downloaded from NCBI (173 genomes in total), as well as a set of 3 genomes for an outgroup constitution, namely Anthocerotibacter panamen-Table 1. Str ains isolated fr om Guadeloupe habitats and associated genome c har acteristics.Str ain ID corr esponds to the r efer ence number in the P aris Museum Collection (PMC) of c y anobacteria from which strains are available upon request.Affiliation is according to the 16SrRNA-based phylogenetic analysis displayed on Fig. S2 and to the distance matrix in Table S1 .sis C109, Gloeobacter kilaueensis JS1 and Gloeobacter violaceus PCC 7421.FetchMG was then used to r etrie v ed single copy marker genes (MG) present in this dataset and our 9 c y anobacterial strain genomes (Kultima et al. 2012 ).In order to maximize the number of marker genes, genomes displaying a low number of MG were discarded.In total, 144 genomes sharing 30 MG altogether were k e pt for the phylogenomic anal ysis.First, a m ultiple alignment was ac hie v ed for eac h MG using MAFFT with local alignment iter ativ e r efinement option (Katoh et al. 2019 ).All alignments were then concatenated and refined using BMGE with default options (Criscuolo and Gribaldo 2010 ).Finally, the resulting alignment of length 6457 aa was used to build a phylogenomic tree using RaxML v8.2.12 (WAGGAMMA model, 100 bootstr a ps) (Stamatakis et al. 2012 ).

Metabolomic profiling
For each c y anobacterial strain, cellular biomasses w ere obtained fr om 500-mL cultur es in 2-L Erlenmeyer flasks with a photon flux density of 6 μmol.m −2 .s−1 and a 13:11 h light: dark cycle, with the aim to produce enough biomass in standardized conditions for the different strains.Cyanobacterial cells were centrifuged at 4,000 rpm for 10 min.The supernatants w ere discar ded and the pellets were freeze-dried and ly ophilized (F reezone 2.5 L, Labconco, Kansas City, USA).Then, the lyophilized cells w ere w eighted then sonicated 2 min in 80% methanol with a constant ratio of 100 μL of solvent for 1 mg of dried biomass and centrifuged at 4 • C (12,000 g; 5 min).Two microliters of the supernatant were analysed in triplicate with an ultra-high-performance liquid chromatogr a ph (UHPLC Ultimate 3000, Thermo, Waltham, MA, USA) using a Polar Advances II 2.5 pore C18 column (Thermo, Waltham, MA, USA) at a 300 μL •min −1 flow rate with a linear gradient of acetonitrile in 0.1% formic acid (5%-90% of 21 min) coupled with a high-resolution mass spectrometer.The eluted metabolite contents wer e anal ysed using an electr ospr ay ionization hybrid quadrupole time-of-flight (ESI-QqTOF) high resolution mass spectrometer (Maxis II ETD, Bruk er).Positi ve and negative autoMSMS mode was used with information dependent acquisition (IDA), on the 50-1500 m/z range at 2 Hz or between 2-8 Hz speed, for MS and MS/MS r espectiv el y, according to the r elativ e intensity of the parent ions, in consecutive cycle times of 2.5 s, with active exclusion of previously analysed parents .T he data were analysed with the MetaboScape 4.0 software (Bruker) in order to automatically perform internal recalibration ( < 0.5 ppm), search and group all together classical adduct forms (M + H + , M + 2H + , M + 3H + , M + Na + , M + K + and M + NH 4 + ) using a thr eshold v alue of 0.8 value for the co-elution coefficient factor.Metabolite annotation w as attempted accor ding to the precise mass of the molecules and their r espectiv e MS/MS fr a gmentation patterns with r egards to MS/MS libraries (NIH, GNPS, EMBL, MassBank and ReSpect) and the CyanoMet database (Jones et al. 2020(Jones et al. , 2021 ) )  The molecular network was created using the MetGem v1.2.2 softw are (Oliv on et al. 2018 ) from the whole MS/MS data (in mgf format) for the two extracted out of the nine c y anobacteria.The netw ork w as cr eated wher e edges wer e filter ed to hav e a cosine scor e abov e 0.65 and mor e than four matc hed peaks.Further edges between two nodes were k e pt in the network only if each of the nodes a ppear ed in eac h other's r espectiv e top 10 most similar nodes .T he MetGem database search function was used to screen each spectrum with GNPS spectral libraries.

Light and electron microscopy analyses
Light micr oscopy photogr a phs of the specimens were taken with an AxioCam MRc digital camera coupled to an Axio ImagerM2 Zeiss microscope.Cell and filament width, length, morphology, color and motility were determined.Strains were identified morphologically using the updated taxonomic literature (Komarek et al. 2014 , Komárek andJohansen 2015 ).For electron microscopy, str ains fr om a gr owing cultur e wer e fixed with 2.5% glutar aldeh yde, 2% paraformaldeh yde, 0.18 M sucrose and 0.1% picric acid in 0.1 M Sorensen phosphate buffer (pH 7.4) for 1 hour at RT. Cells were post-fixed with 1% osmium tetroxide during 1 hour in the same buffer (A hour, RT), then rinsed with distilled water and dehydrated in a graded ethanol series (30%, 50%, 70, 85%, 95% and 100%, 15 min each).Cy anobacteria w ere then embedded in Epon resin in an increasing gradient of resin in ethanol.Samples were sectioned (60 nm, thick) with an ultramicrotome (RMC Ultramicrotome Po w erTome XL) and tr ansferr ed onto 150 mesh copper formv ar grids.Grids wer e stained with ur an yl acetate satur ated in 50% ethanol and examined under a tr ansmission electr on micr oscope (Hitac hi 7700, Ja pan) under an acceler ation volta ge of 80 kV.
Amplicon Sequence Variants identical to 16S rRNA sequences of O/Coleofasciculales and O/Leptolyngb y ales strains from this stud y re presented 2% and 5.7% of the reads in the biofilm from station 7 where PMC strains were isolated, respectively.O/Nostocales strains represented 50% of 16S rRNA reads in the biofilm from station 5 where the two strains PMC 1069 and 1070 were isolated.

Polyphasic approaches and taxonomic affiliation
The strains PMC 1069 and 1070 gr e w as emer ald-gr een isopolar filaments including barr el-sha ped heter ocytes (11.5 ± 0.7 ×18.6 ± 3.4 μm, Fig. 1 A-B), displaying a ∼2 μm-thick sheath (Fig. 1 C), and a false br anc hing pattern, with no akinetes.Vegetative cells measured 13.2 ±1.0 ×14.3 ±1.0 μm ( Table S1 ).Terminal cells appear ed lar ger than long (10.6 ±0.8 ×8.8 ±0.8 μm, Fig. 1 B-C; Table S1 ).Filaments were more or less constricted (Fig. 1  Strains PMC 1078, 1079, and 1080 appeared as green tuft-like colonies of thin and uniseriate filaments, without sheaths (Fig. 2 A-B).Vegetativ e cells ar e cylindrical and measur ed 2.9 ± 0.2 × 4.1 ± 0.8 μm ( Table S1 ).Apical cells were elongated and hooked (Fig. 2 A-B).Hormogonia wer e observ ed.A thin m ucila ge is visible ar ound the filaments under the electron microscope, along with parietal localization of the thylakoids, and slight constriction at the cross w alls betw een consecutive cells (Fig. 2 C-D).Other components such as c y anophycin granules (nitrogen reserves) were observed within the cells (Fig. 2 D).The strains PMC 1078, 1079 and 1080 appear ed closel y r elated together in the phylogenomic tr ee (Fig. 4 , ANI > 0.999).Their closest r elativ e was a clade that contained two Spirulina ( S. salsa and S. major ).Ho w e v er, this clade was dis-tant, ANI v alues wer e not high (0.720 to 0.723), and these strains did clearly not display the typical Spirulina -like coiled morphology (see Fig. 2 ).Morphological and ultrastructural features and the 16S rRNA-based phylogenetic clustering ( Fig. S2 ) are congruent with the described genus Jaaginema, belonging to the Leptolyngb y ales or der (Anagnostidis and Komár ek 1988, Mar eš et al. 2019, Strunecký et al. 2023 ).Due to the lac k of av ailable sequences for Jaaginema , the phylogenomic tree alone is not conclusive regarding this affiliation (Fig. 4 ).
Strains PMC 1050, 1051, 1068 and 1076 displa yed o v er all common features .T hey a ppear ed as blue-gr een filaments , motile , with hormogonia (Fig. 3 A).Cells wer e lar ger than long, or as large as long (5.84 ± 0.43 × 5.77 ± 1.28 μm for PMC 1050 and 4.72 ± 0.51 × 4.21 ± 0.97 μm for PMC 1068, Table S1 ).Cell content, and thylakoids with fascicular arrangement in the c ytoplasm w ere visible (Fig. 3 B-D).Mor phological observ ations and 16S rRNA-based phylogenetic clustering ( Fig. S2 ) are overall consistent with the Coleofasciculales (Komarek et al. 2014, Strunecký et al. 2023 ).The phylogenomic tree also clearly displayed a highly supported clade consisting of these four strains (bootstrap, BS: 100, Fig. 4 ).They clustered together, with no close relative, suggesting novelty and not allowing confident assignation to any of the existing genera (highest ANI < 0.724).Their sister group consisted of representativ e of v arious gener a belonging to se v er al orders, and the grouping was not robust as evidenced by low bootstr a p v alues (Fig. 4 ).Their pr e vious affiliation to genus Oscillatoria (Duperron et al. 2020 ) is thus not supported based on the various criteria used in this study.Based on these results and the criteria required to c har acterize a new genus, that include 16S rRNA sequence similarity below 95% with existing genera combined with at least one auta pomor phic c har acter, habitat specificity and low av er a ge nucleotide identity (ANI), the classification of PMC 1050, 1051, 1068, and 1076 is proposed as a new genus and species .T hus , they were classified into the new genus Karukerafilum with the type species K. mangrovensis , described here as a benthic filamentous c y anobacterium associated to benthic mats from mangro ves .

Formal description of the genus and species:
Karukerafilum mang ro vensis Halary, Duval, Marie, Bernard et Duperron, gen. nov., sp. nov.The new genus Karukerafilum contains a single species ( K. mangrovensis ), thus a single description is provided for both genus and species.

Inter-strain genomic heterogeneity
The Scytonema -r elated str ains PMC 1069 and 1070 shar ed an ANI v alue abov e 0.999, and 7288 ORFs (Fig. 5 , Table S2 ).They shared 8652 ORFs while 350 and 96 were unique to each strain, r espectiv el y.The Jaaginema -r elated str ains PMC 1078, 1079, and 1080 had v ery fe w str ain-specific ORFs (16 to 198 vs. 4 771 shar ed by all strains), and ANI values above 0.999.Interestingly, 20 to 25 ORFs wer e shar ed betw een tw o strains and absent from the third, showing a limited le v el of inter-strain differentiation.It must be noted that all three strains were isolated from the same biofilm sampled at station 7 (Table 1 ).The four Karukerafilum mangrovensis strains displayed a different picture.Overall, 3741 ORFs were shared among all four strains, and very few (19 to 136) were unique to a single str ain.Inter estingl y, str ains PMC 1050 and PMC 1051 shared additional 1289 ORFs unique to this clade, while PMC 1068 and PMC 1076 shared 982 (Fig. 5 ).Within each of these two subgr oups, ANI v alues wer e abov e 0.999, while v alues between the two subgroups were 0.862, advocating for two distinct potential genotypes within a single new species (see below).Only a limited number of the ORFs that ar e differ entiall y pr esent among str ains were annotated (33 out of 446 (7.4%) in Scytonema , 34 out of 304 (11.2%) in Jaaginema , and 161 out of 2591 (6.2%) in K. mangrovensis , Supplementary Table 3 ), the rest encoding for yet unknown functions.

Genes involved in nitrogen and phosphorous metabolism
Scytonema -assigned strains PMC 1069 and 1070 possessed nine nif genes (Fig. 6 ), including four as dual copies ( nifD, E, H, and K ) and Mo.Nit in 8 copies.Jaaginema -related strains PMC1078, 1079 and 1080 had none of these.On the other hand, gene-content heterogeneity was observed among the four Karukerafilum mangrovensis strains (Fig. 6 , Table S3 ).Indeed, PMC 1050 and 1051, possessed ten nif genes as well as one gene involved in dinitrogenase ironmolybdenum cofactor (FeMo-co) synthesis.A contrario , none of these genes was found in strains PMC 1068 and 1076.
As for nitr ogen-r elated genes, the two Scytonema strains displayed highly similar gene composition for genes involved in phosphorous metabolism, and the three Jaaginema strains displayed identical profiles (Fig. 6 ).Three genes differed among the two subgroups of K. mangrovensis strains, but only in terms of copy numbers ( phoD-like, phn C , pstA , the former an alkaline phosphatase, the other two elements of the transport system).

Secondary metabolites biosynthesis pathways
Biosynthesis pathways for secondary metabolites w ere sear ched for in all strains.Scytonema -affiliated genomes were those displaying the greatest number of putative pathways ( Supplementary Table 4 ).Nonribosomal peptide synthetases (NRPS) were identified in Scytonema PMC 1069 and 1070.These included homologues with 100% similarity for the complete biosynthesis pathway of Anabaenopeptin NZ857/nostamide A found in Nostoc punctiforme PCC 73102 (three and one gene clusters, respectively).Jaaginemaassigned strains did not yield many matches with high similarity, the only complete and highly similar biosynthesis pathway was the one encoding for 1-heptadecene production.The Karukerafilum mangrovensis PMC1050 and PMC 1051 strains yielded a complete pathway for geosmin biosynthesis with 100% similarity, and two complete pathways for nostopeptolide A2 biosynthesis but with only 50% similarity (i.e.50% of the genes had a significant BLAST hit to the genes).None of these was found in the two other K. mangrovensis strains PMC 1068 and 1076.

Metabolomic profiling
Results from LC-MS positive and negative modes were assembled and compared among the nine strains .T hey yielded a total of 3836 metabolites, and strain clustering according to metabolome similarity was ov er all congruent with their r espectiv e affiliations, with closel y r elated str ains clustering together and sharing a higher proportion of their metabolites compared with other strains (Fig. 7 ).Although strains and cultures were non-axenic, we assume that most-to-all metabolites observed originated from c y anobacterial cells themselves, as most of the biomass in cultur es corr esponds to c y anobacteria.As for genomes and aforementioned gene contents, the four Karukerafilum mangrovensis str ains wer e split in two gr oups, with PMC1050 and 1051 being distinct from PMC 1076 and 1068.Metabolite annotation r e v ealed mostl y primary metabolism, and v ery fe w secondary metabolites were successfully annotated (Fig. 8 ).For example, Anabaenopeptins, of which biosynthetic pathway genes were identified in strains PMC 1069 and 1070 were not detected despite the use of standards in our databases that should have permitted their identification, suggesting a lack of expression under culture conditions used.No production of other c y anotoxins w as detected either under the culture conditions used in the laboratory.

No vel str ains representing genomes of poorl y documented cyanobacterial taxa
Strains PMC 1069 and 1070 displayed the typical ultr astructur e and morphology described for other Scytonema .Phylogenomics confirmed that they were related to Tolypothrix bouteillei , a filamentous nitrogen-fixing species isolated from building stones in India with a 11.5 Mb genome (Chandrababunaidu et al. 2015 ), and to Scytonema hofmanni PCC7110, also a nitr ogen-fixer.As observ ed for these two close r elativ es, both str ains also possess the nif genes and are thus able to fix nitr ogen, a featur e congruent with the occurrence of heterocytes in cultures.
The three strains PMC 1078, 1079 and 1080 can be classified as a single species based on ph ylogenetic, ph ylogenomic trees and the very high ANI similarity.They display the typical morphological features of the genus Jaaginema .Due to the lack of pr e viousl y sequenced Jaaginema r efer ence genome, their closest r elativ es ar e quite distant in the phylogenomic tree, consisting of Spirulina salsa and S. major .This distant relationship is likely due to a lack of a vailable genomes , y et it w as also observed in a r ecentl y published 16S rRNA-based phylogeny of several Jaaginema strains in which the same two Spirulina species were closest relatives of the various strains, yet with limited similarity (Brito et al. 2017 ).Jaaginema strains do not display heterocyts and do not possess the genes necessary for nitrogen fixation from N 2 (Brito et al. 2017 ).
The strains PMC 1050, 1051, 1068 and 1076 were difficult to affiliate based on the 16S rRNA encoding gene, due to a limited number of closely related sequences, all from uncultured c y anobacteria (Duperron et al. 2020 ).The whole clade in this former tree was onl y distantl y r elated to other Oscillatoriales , suggesting no velty.Whole genome sequencing confirms this novelty, with the four str ains r epr esenting a ne w clade that is onl y distantl y r elated to a group that contains genomes from numerous genera representing various c y anobacterial or ders, with lo w ANI values, and does F igure 8. Netw ork based on GNPS analysis displaying successfully annotated nodes, colors corresponding to the different strains in which nodes were identified.
not support affiliation to any of the genomes available assigned to well-described (or known) genera.On the basis of 16S rRNA gene, str ains wer e pr e viousl y assigned to the Oscillatoriales, most likel y famil y Oscillatoriaceae (Komar ek et al. 2014, Duperr on et al. 2020 ).Some featur es ar e compatible with featur es described in the genus Oscillatoria but this genus is reported to be polyphyletic and thus this genus-le v el classification is of limited r ele v ance (e.g.(Hauerová et al. 2021 )).Based on r esults fr om the pr esent study, we propose that these four strains represent a new genus and new species with the type species Karukerafilum mangrovensis as a benthic filamentous mangr ov es c y anobacterium within the Order Coleofasciculales.

Closel y rela ted str ains display differences in gene content and potential capabilities
Analyzing two to three strains per clade revealed that closely related strains may harbor numerous unique ORFs despite very high ANI values .T he four strains affiliated to Karukerafilum mangroven-sis split into two sub-species-le v el subclades, eac h sharing its own additional set of ORFs beyond those shared by all four strains .T his supports that the differentiation among str ains involv es differences in gene content, as r ecentl y described in large-scale comparisons of Microc ystis , Aphanizomenon or Limnospira strains (Meyer et al. 2017, Pér ez-Carr ascal et al. 2019, Halary et al. 2023, Roussel et al. 2023 ).The 'pan-genome' concept has emerged as relevant to describe the metagenome of very closely related strains that cannot always be distinguished based on their 16S rRNAencoding genes .T he concept is particularl y r ele v ant to bloomforming planktonic c y anobacteria in which distinct genotypes may bloom successiv el y in what looks at first glance like a singlespecies bloom e v ent (Bec k et al. 2018 ).Based on our results, we suggest that its r ele v ance should also be e v aluated in the case of benthic biofilm-forming c y anobacteria, in particular because some of the non-shared genes are associated with important functions.Among the ORFs that are not common to all strains within each of the three clades, few were successfully annotated, suggesting that most of the heterogeneity is associated with unkno wn functions.Ho w e v er, some w ell-kno wn functions did sho w inter-str ain heter ogeneity.An example is the occurr ence of genes involv ed in nitr ogen fixation in one of the two K. mangrovensis sub-species clades (strains PMC 1050 and 1051), but not the other (strains PMC 1076 and 1068).This suggests that only the former clade ma y ha ve the ability to fix nitrogen, an important featur e particularl y in nitr ogen limited habitats suc h as mangr ov es (Gonthar et et al. 2017, Zilius et al. 2020 ).Various strains of nitrogen-fixing "Oscillatoria"-like bacteria are reported, despite that these c y anobacteria do not de v elop heter ocytes.Instead, certain cells usually localized in particular regions of the filament specialize into this metabolism (Bryceson and Fay 1981 ).Withingenus and e v en species differences in abilities to fix nitrogen were r ecentl y documented among strains of Tolypothrix , of which strain PCC 7712 can for example fix nitrogen while strain PCC 7601 does not (Bozan et al. 2022 ).Inter estingl y in this published report, both Tol ypothrix str ains possessed all necessary nitr ogen fixation genes, the difference being hypothesized to result from inability of the latter to differentiate heterocytes.
Another type of heterogeneity between the two groups of K. mangrovensis strains is observed in the genes involved in phosphorous metabolism, with three genes displaying different copy numbers between the two groups.According to ANI values, these two sub-species clades belong to separate genotypes within K. mangrovensis , only one (PMC1050 and 1051) displaying genes for nitr ogen fixation.Inter estingl y, str ains form this subclade wer e isolated from the Manche-à-Eau lagoon, a seawater mangro ve , while strains PMC 1068 and 1076 were isolated from periphytic biofilms at stations located in the Canal des Rotours, which water is slightly less salty (25 vs 35 g.l −1 ) in the upper layer of the water column where the biofilm samples were collected (Laverman et al. 2023 ).Mangr ov e habitats ar e known to be nitrogen-depleted (Fernandes et al. 2012 ), which may favor nitrogen-fixing strains.Nitr ogen is r eportedl y av ailable in the Manc he-à-Eau la goon, while NO 3 − and NH 4 + ar e av ailable at low concentr ations in the water column of the sites collected in the Canal des Rotours (Gontharet et al. 2017 ) r espectiv el y ar ound 3 and 11 μM.

Secondary metabolites potentially relevant to biofilm resistance to predation
Anabaenopeptin NZ857 biosynthesis gene clusters were identified in both strains of Scytonema .It was also identified in its r elativ e Tolypothrix bouteillei VB521301.These hexapeptides inhibit phosphatases and pr oteases, whic h can induce toxicity against zooplankton ( Lenz et al. 2019, Monteiro et al. 2021 ).They ar e r eported fr om v arious c y anobacterial genera including Anabaena , Nostoc , Microcystis , Planktothrix , Lyngb y a , and Brasilonema , the latter a close r elativ e of Scytonema (Sanz et al. 2015 ).The variant NZ957 was specificall y r eported in Synec hococcus sp.PCC 7502 and Anabaena sp .TA U NZ-3-1 (Monteiro et al. 2021 ).Induced toxicity to zooplankton could possibly play a major role in grazing limitation.
Ho w e v er, anabaenopeptin was not detected in the metabolome of cultur ed str ains, suggesting that its expression is not constitutive, and possibly occurs in conditions differ ent fr om those used in our labor atory cultur es.
Strains PMC 1050 and 1051 ( K. mangrovensis ) have the potential to produce geosmin, the volatile compound responsible for earthly odor.Recent work on Caenorhabditis elegans indicates that despite geosmin itself is not harmful to the worm, it is detected by this bacterivorous nematode and acts as an efficient repellent (Zaroubi et al. 2022 ).By possibly reducing grazing, geosmin production could be relevant to biofilm forming c y anobacteria.Zaroubi and co-workers suggested that geosmin could r epr esent a chemical warning cue emphasizing the unpalatability of the producing bacteria, in their case a Streptomyces , in a way comparable to color patterns that induce learned avoidance responses in animals.Inter estingl y, str ains PMC1050 and PMC 1051 are the two K. mangrovensis strains that have genes for nitrogen fixation, larger genomes, and these possibilities not shared with PMC 1076 and 1068, confirming that they belong to two different genotypes.

Conclusion
Genome sequencing and polyphasic ( i.e .ph ylogenetics, ph ylogenomics, ANI, mor phological & ultr astructur al c har acteristics, metabolites composition) analysis of nine selected strains abundant in benthic and epiphytic biofilms from Guadeloupe provides information r egarding poorl y explor ed c y anobacterial linea ges, namel y Jaaginema , Scytonema and a new genus Karukerafilum gen.no v.T hese three biofilm-forming clades display distinct abilities with regards to nitrogen fixation, phosphorous utilization as well as secondary metabolites biosynthesis potential, illustrating some possible successful strategies in the mangrove-tofr eshwater tr ansition zone.Besides their ability for primary production, an interesting feature is the genomic potential to produce to xins ( i.e. anabaenope ptin) or re pellent molecules ( i.e. geosmin) that may pr e v ent biofilms gr azing, emphasizing biofilm-forming c y anobacteria as producers of potentially interesting bioactive molecules that warrant further exploration.
D), constriction F igure 1.Light microscop y (A and B) and TEM (C and D) micr ogr a phs of Sc ytonema sp.PMC 1069.18 (Or der Nostocales) fr om Guadeloupe mangr ov es.Abbr e viations: cc: cross-wall constriction, ca: carboxysome, cm: cytoplasmic membrane, f: false br anc hing, h: heter ocyte, s: sheath, t: thylakoids, tc: terminal cell and tz: tr anspar ent zone.Scale bars A and B = 20 μm, C = 5 μm and D = 500 nm.le v el being higher to w ar ds the terminal end.The two phylogenetic trees based on 16S rRNA sequences and the genomes ( Fig. S2 and Fig. 4 , r espectiv el y), cluster ed str ains PMC 1069 and 1070 together.The phylogenomic tree based on concatenated sequences fr om 30 shar ed genes among 144 av ailable r efer ence genomes sho w ed a bootstr a p-supported sister clade to one clade containing Tolypothrix bouteillei and Scytonema hofmanni (Fig. 4 ) with average nucleotide identity (ANI) of 0.774 and 0.773, respectively.The pol yphasic a ppr oac hes with all the criteria used in this study are congruent with affiliation of PMC1069 and 1070 to the order Nostocales, family Scytonemataceae and genus Scytonema (Komarek et al. 2014 ).
content with radial or fasciculated thylakoid arrangement in the cytoplasm.Very thin-to-absent sheath, attached to the trichome.Dia gnosis: Ne w genus differs from other genera by the substantial differences in the nucleotide sequence of the 16S rRNAencoding gene and of 30 concatenated coding sequences.Type species (PMC 1050.18),here designated: Karukerafilum mang ro vensis Halary, Duval, Marie, Bernard et Duperron Holotype : a cryopr eserv ed and formaldehyde-fixed sample of the strain PMC 1050.18 was deposited at Paris Museum Collection (PMC), P aris, Fr ance.Corr esponding r efer ence 16S rRNA encoding sequence was deposited in GenBank under accession number MN823169 and an assembled MAG was deposited under accession SAMN36471718.Type locality: Isolated from a dense benthic filamentous brown mat collected in the Manche-à-Eau lagoon, Guadeloupe (N 16.276 • , W 61.555 • ).

Figure 4 .
Figure 4. Phylogenomic tree based on the comparison of 30 concatenated genes from 144 available reference genomes (see text for methodology).

Figure 5 .
Figure 5. Venn dia gr ams showing the number of shar ed and unique ORFs for eac h of the thr ee c y anobacterial lineages: Sc ytonema (left), Jaaginema (middle), Karukerafilum mangrovensis (right).

Figure 6 .
Figure 6.Occurrence and copy numbers of genes involved in nitrogen fixation and metabolism (left) and phosphorous metabolism (right).

Figure 7 .
Figure 7. Metabolome profiling of the 9 strains.Left: Principal Component Analysis plot based on the analysis of 3836 analytes.Right: occurrence and intensity of MS-MS peaks corresponding to analytes (lines) in each strain (columns).