The present study describes a taxonomic investigation of eight Flavobacterium strains (Supplementary Table 1) isolated in the frame of the microbiological research of the Czech Antarctic Research Program (CARP¹). Microbiological sampling is conducted yearly on the James Ross Island (near the north-eastern extremity of the Antarctic Peninsula), Antarctica. Sampling is performed during the summer period when the majority of the island becomes an ice-free area. Antarctic summer season allows sampling of upper soil layers, permafrost, lakes and proglacial streams, which are all areas of intense microbial activities (Kopalová et al., 2013; Nedbalová et al., 2013). Flavobacterium spp. analyzed in this study were isolated from environmental materials (different water, organic and soil sources) sampled during the years 2010–2019 (Supplementary Table 1). Water samples were spread on Reasoner’s 2A agar plates (R2A, Oxoid) (150 μl) and cultivated at 15°C for 5 days. Soil samples were processed by dispersing 1 g of soil and/or stone fragment material in 5 ml of sterile saline solution followed by dispersing of 100 μl of obtained solution on R2A agar plates and cultivated at 15°C for 7 days. Large numbers of well-separated yellow pigmented colonies were randomly selected, hand-picked by sterile inoculation sticks and purified by subculturing. Pure cultures were maintained on the R2A agar slants until transportation to the Czechia. After transportation, subcultures were checked for purity and maintained at −70°C for long-term storage. The type strains of the phylogenetically closest relatives Flavobacterium hercynium CCM 9054^T, Flavobacterium branchiicola CCM 9061^T, Flavobacterium chilense CCM 7940^T, Flavobacterium araucananum CCM 7939^T, Flavobacterium psychroterrae CCM 8827^T and Flavobacterium saccharophilum CCM 8770^T were obtained from the Czech Collection of Microorganisms (CCM²) for parallel testing and comparison purposes. All strains were routinely cultivated on R2A agar plates at 20°C for 48 hrs.

Genomic DNA was extracted and purified using High Pure PCR Template Preparation Kit (Roche Diagnostics) according to the manufacturer’s recommendations. The 16S rRNA genes were amplified and sequenced using universal bacterial primers pA (5′-AGAGTTTGATCCTGGCTCAG-3′) and pH (5′-AAGGAGGTGATCCAGCCGCA-3′) (Edwards et al., 1989). PCR products were purified using High Pure PCR Product Purification Kit (Roche Diagnostics). Sequencing was performed by the Eurofins MWG Operon sequencing facility³. The 16S rRNA sequences were submitted to the EzBioCloud server (Yoon et al., 2017) for initial identification and determination of the closest phylogenetic neighbors. The 16S rRNA sequences of the phylogenetically closest and validly named species were downloaded from the GenBank database⁴ and used for multiple sequence alignment using the Molecular Evolutionary Genetics Analysis (MEGA) software (v7.0) (Tamura et al., 2013). The maximum-likelihood (ML) method with the Tamura-Nei gamma distance model was used to calculate the genetic distances and to build the phylogenetic tree.

Genomic DNA was extracted using a High Pure PCR Template Preparation Kit as described above. Sequencing libraries were prepared using KAPA HyperPlus Kits (Roche, Switzerland) with 7 min of enzymatic fragmentation and final fragments length within the range of 550–650 bp. Sequencing was performed on Illumina MiSeq platform with MiSeq Reagent Kit v2 (500-cycles) (Illumina, United States). The St. Petersburg genome assembler (SPAdes v3.11.1) (Bankevich et al., 2012) was used for de novo genome assembly. The MismatchCorrector was run to reduce short indels and the number of mismatches. The read coverage cut-off value was automatically computed by the software resulting in contigs and scaffolds. Quality assessment was performed by the QUAST tool (Gurevich et al., 2013) and Bowtie2 v2.4.2 (Langmead and Salzberg, 2012).

The whole genome sequence (WGS) data were analyzed for further confirmation of the taxonomic status of the analyzed strains and for comparison with genomes of related Flavobacterium spp. The WGS data of type strains P4023^T and P7388^T were submitted to the Type Strain Genome server (TYGS) (Meier-Kolthoff and Göker, 2019) and to the Microbial Genome Atlas (MiGA) (Rodriguez-R et al., 2018). To further confirm the taxonomic status of the analyzed strains, type strain genomes along with genomes of the closest related and validly described Flavobacterium spp. were subjected to the UBCG pipeline (Na et al., 2018) with default parameters to calculate a phylogenetic tree based on 92 concatenated core genomes. The whole genome sequences of the closest Flavobacterium spp. were downloaded from the GenBank database and the genomes of F. branchiicola CCM 9061^T and F. psychroterrae CCM 8827^T were sequenced and assembled in this study as described above and deposited in the GenBank database under accession numbers GCA_018383905.1 and GCA_018380615.1, respectively. Average nucleotide identity (ANI) values were calculated using the ANI calculator on the Kostas Laboratory website⁵ using reciprocal best hits (two-way ANI) (Rodriguez-R and Konstantinidis, 2016). Protein-coding sequences were calculated using the GeneMarkS prediction tool (Besemer et al., 2001) and subjected to the amino acid average identity values (AAI) calculator on the Kostas Laboratory website⁶ (Rodriguez-R and Konstantinidis, 2016). Digital DNA-DNA hybridization (dDDH) values were generated using the Genome to Genome Distance Calculator (GGDC) version 2.1 (Meier-Kolthoff et al., 2013).

Additional genome annotation was performed using the Prokaryotic Genome Annotation Pipeline (PGAP) (Tatusova et al., 2016; Li et al., 2021) and Operon-Mapper (Taboada et al., 2018). Functional annotation was performed with EggNOG-mapper v5.0 (Huerta-Cepas et al., 2019). CRISPRs were identified by CRISPR Detect (Biswas et al., 2016). Presence of putative prophages was predicted by PHASTER (Arndt et al., 2016) and Prophage Hunter (Song et al., 2019) tools. Genome mining for potential secondary-metabolite associated biosynthetic gene clusters (BGCs) was determined using antiSMASH version 6.0.0 (Blin et al., 2021). Annotations of predicted open reading frames (ORFs) and genes/domains from candidate BGCs were visualized using the R package “gggenes” (Wilkins and Kurtz, 2019) and “ggplot2” (Wickham, 2016). ABRicate v1.0.1⁷ was used for screening of known antibiotic and metal resistance genes using the ResFinder database (Zankari et al., 2012), NCBI Bacterial Antimicrobial Resistance Reference Gene Database, CARD (Jia et al., 2017), MEGARes database (Doster et al., 2020) and ARG-ANNOT tool (Gupta et al., 2014). Further in silico prediction of resistomes and putative resistance genes was performed using Resistance Gene Identifier (RGI, v5.1.1)⁸ with CARD v3.1.1 database (Jia et al., 2017). The enzymatic potential for carbohydrate degradation was evaluated by the web-based tools dbCAN2 and dbCAN-PUL for automated carbohydrate-active enzyme annotation (Yin et al., 2012; Zhang et al., 2018; Ausland et al., 2021).

The colony morphology was determined on R2A agar after 48 h cultivation at 20°C. The cellular morphology of all strains was observed by light microscopy after Gram staining. The cellular morphology of the type strains, P4023^T and P7388^T, was also examined by transmission electron microscopy using a Morgagni 268D Philips (FEI Company) electron microscope.

The growth at different temperatures (1–40°C, in 5°C increments) and the salt tolerance (0.5, 1.0, 2.0, 3.0, 4.0 of NaCl) were determined on R2A agar plates. The pH tolerance was assessed in Trypticase Soya Broth (TSB) inoculated with two drops of cell suspension of concentration as McFarland 2.0 with pH 5–10 at intervals of 1.0 pH unit at 20°C adjusted with the following buffer systems: pH 5.0–8.0, 0.1 M KH₂PO₄/0.1 M NaOH; pH 9.0–10.0, 0.1 M NaHCO₃/0.1 M Na₂CO₃. The pH value of the TSB was confirmed after autoclaving.

All eight isolates along with reference strains F. hercynium CCM 9054^T, F. branchiicola CCM 9061^T, F. chilense CCM 7940^T, F. araucananum CCM 7939^T, F. psychroterrae CCM 8827^T and F. saccharophilum CCM 8770^T were phenotypically characterized by the most relevant tests recommended for description of new taxa within the family Flavobacteriaceae (Bernardet et al., 2002). These tests comprised of: activities of oxidase (OXItest; Erba-Lachema) and catalase (Atlas, 2010) (ID color Catalase; bioMérieux) performed according to the manufacturer’s instructions; production of urease (Christensen, 1946); production of arginine dihydrolase, ornithine and lysine decarboxylase production (Brooks and Sodeman, 1974), oxidation-fermentation test (Hugh and Leifson, 1953); production of acid from carbohydrates aerobically (Gilardi, 1985), production of H₂S on Triple Sugar Iron Agar (HiMedia), hydrolysis of aesculin, starch (Barrow and Feltham, 1993), gelatine, Tween 80 (Pacova and Kocur, 1984), casein, tyrosine (Kurup and Babcock, 1979), cellulose (R2A broth with strip of Whatman paper No.1) (Ali et al., 2009), carboxymethylcellulose (CMC) (Ibrahim et al., 2021) and O-nitrophenyl-β-D-galactopyranoside (ONPG) (Lowe, 1962), phospholipinase activity in the egg-yolk reaction (Owens, 1974), nitrate and nitrite reduction, production of indole, utilization of citrate in Simmon’s citrate agar (Barrow and Feltham, 1993), utilization of acetamide (Oberhofer and Rowen, 1974) and sodium malonate (Ewing, 1960). The presence of flexirubin-type pigments was tested using 20% (w/v) KOH solution and the presence of capsule was identified by the Congo red adsorption test (Bernardet et al., 2002).

Capability of growth on different media was tested on R2A agar (Oxoid), Nutrient agar (Oxoid), Plate Count agar (Oxoid), Tryptone Soya agar (Oxoid), marine agar (Lapage et al., 1970), MacConkey and Endo agar (HiMedia), Brain Heart Infusion agar (Oxoid), and Mueller-Hinton agar (Oxoid) under aerobic conditions. Anaerobic growth was tested using Anaerocult A system (Merck) and microaerophilic growth in atmosphere of 5% CO₂, both on R2A agar plates. All above listed tests were inoculated using cells of analyzed strains grown on R2A agar at 20°C for 48 hrs. All tests were read daily for up to 7 days with exception of the tyrosine hydrolysis test (up to 10 days). The utilization of carbon sources, enzyme activities and other additional phenotypic characteristics were further tested using identification test kits GEN III MicroPlateTN (Biolog) with the protocol A, API 20 NE and API ZYM (bioMérieux) according to the manufacturer’s instructions.

In vitro antimicrobial susceptibility was assessed using the disk diffusion method. Cell suspensions of concentration as 0.5 McFarland were prepared from cultures cultivated on R2A agar. Suspension containing each strain (100 μl) was spread on the Mueller-Hinton agar plates. Tested antibiotic disks (Antimicrobial Susceptibility Disks, Oxoid) were as follows: ampicillin (10 μg), aztreonam (30 μg), carbenicillin (100 μg), cefixime (5 μg), ceftazidime (10 μg), cephalotin (30 μg), ciprofloxacin (5 μg), gentamicin (10 μg), chloramphenicol (30 μg), imipenem (10 μg), kanamycin (30 μg), co-trimoxazole (25 μg), piperacillin (30 μg), polymyxin B (300 U), streptomycin (10 μg) and tetracycline (30 μg). CLSI (The Clinical and Laboratory Standards Institute) and EUCAST (The European Committee on Antimicrobial Susceptibility Testing) standards were strictly followed for cultivation and inhibition zone diameter reading (CLSI, 2015; EUCAST, 2017).

Cellular fatty acid methyl esters analysis was performed from biomass of all strains including the reference type strains incubated strictly under the same conditions. All cultures were grown on R2A agar at 20 ± 2°C for 48 hrs to reach the late-exponential stage of growth according to the four quadrant streak method (Sasser, 1990). Fatty acids were identified using an Agilent 7890B gas chromatograph according to the Standard Protocol of the Sherlock Identification System (MIDI Sherlock version 6.2, MIDI database RTSBA 6.21). Further chemotaxonomic analyses including polar lipids, quinones, and polyamines were performed using freeze-dried biomass prepared from cells grown in R2A broth at 20°C for 48 hrs. Extraction of all above mentioned chemotaxonomic markers was performed according to the previously described protocols (Busse and Auling, 1988; Tindall, 1990b, a; Altenburger et al., 1996; Stolz et al., 2007). Analysis of the polyamine patterns was performed using high-performance liquid chromatography (HPLC) with conditions described by Busse et al. (1997) using the HPLC apparatus as described previously (Stolz et al., 2007).

Initial identification based on the 16S rRNA gene sequences classified all Antarctic isolates to the genus Flavobacterium. Based on the 16S rRNA gene sequence similarity, Antarctic isolates formed two groups for which strains P4023^T and P7388^T were assigned as type strains. The closest related species to proposed type strain P4023^T were F. saccharophilum DSM 1811^T (98.54%), F. hercynium DSM 18292^T (98.54%) and F. chilense LMG 26360^T (98.39%). A proposed type strain for the second cluster, P7388^T, showed the highest sequence similarities to the F. branchiicola 59B-3-09^T (99.12%), F. araucananum DSM 24704^T (99.08%) and F. psychroterrae CCM 8827^T (99.04%). The 16S rRNA gene sequence similarities of strain P4023^T fell below the cut-off value 98.65%, whereas strain P7388^T reached similarities above this threshold. Nevertheless, several studies have already proved that the standard 16S rRNA gene sequence cut-off value is not sufficient for some genera, such as Streptomyces, Streptococcus or Bacillus (Fox et al., 1992; Clarridge, 2004; Guo et al., 2008). Indeed, the genus Flavobacterium and its species with validly published names also do not follow the 98.65% threshold, which therefore requires additional in-depth phylogenetic and phylogenomic analyses for designation of Flavobacterium spp. taxonomic positions and assessment of their novelty (Yi and Chun, 2006; Romanenko et al., 2015).

To evaluate phylogenetic relationships between Antarctic Flavobacterium spp. and their closest phylogenetic neighbors as identified using the EzBioCloud database, the ML phylogenetic tree was calculated and clearly showed well-supported clades comprising of strains P4023^T and P7388^T along with two and four non-type isolates, respectively (Figure 1). Strains P4023^T, P4911 and CCM 9063 formed a separate cluster with the closest phylogenetic neighbor F. hercynium DSM 18292^T supporting the 16S rRNA gene sequence similarities results. Strains P7388^T, P9670, P7381, P7475, and CCM 9065 formed a well-supported cluster close to Flavobacterium resistens BD-b365^T and F. branchiicola 59B-3-09^T.

As mentioned above, clear taxonomic delineation of Flavobacterium spp. requires additional comprehensive analyses. For this reason, draft genomes of proposed type strains P4023^T and P7388^T were initially submitted to TYGS and MiGA platforms for genomic comparisons. TYGS results assigned both strains as novel species. Genome-based phylogeny calculated by TYGS showed F. hercynium DSM 18292^T as the closest phylogenetic neighbor of strain P4023^T, and F. araucananum DSM 27404^T forming a monophyletic cluster with strain P7388^T (Figure 2). MiGA results confirmed novelty of both draft genomes and assigned Flavobacterium weaverense DSM 19727^T and Flavobacterium pectinovorum ATCC 19366^T as the closest related species to P4023^T and P7388^T, respectively. Further phylogenomic analysis through the up-to-date bacterial core gene UBCG pipeline placed strain P4023^T into a cluster shared with F. weaverense DSM 19727^T, while strain P7388^T formed a distinct phylogenetic lineage within the genus Flavobacterium (Supplementary Figure 1).

In silico comparison of genomic distances and calculation of the dDDH values between the draft genomes of strains P4023^T, P7388^T and their closest phylogenetic neighbors resulted in values below 70%, the cut-off established as a threshold value for species delineation by dDDH (Meier-Kolthoff et al., 2013). The ANI values between strain P4023^T and its closest relatives reached 77.60–82.70% with the highest nucleotide similarity to F. weaverense DSM 19727^T. Comparison between strain P7388^T and its closest phylogenetic relatives showed ANI values of 78.17–82.06% with F. araucananum DSM 24704^T with the highest genomic relatedness. Calculated ANI values were well below the 95–96% threshold for species delineation and further supported novelty of the proposed species (Ciufo et al., 2018; Jain et al., 2018). This was also consistent with the calculated AAI values ranging from 71.24–84.84% and 72.77–82.09% for strains P4023^T and P7388^T, respectively (Table 1).

Isolates from both groups represented by type strains P4023^T and P7388^T were aerobic, Gram-negative rods with rounded ends, with average cell size 0.4–0.6 μm × 1.2–2.4 μm and 0.3–0.4 μm × 1.5–3.0 μm, respectively (Supplementary Figures 4, 5). All strains formed yellowish and dark yellow to orange-colored colonies often expressing gliding motility and opalescence on the R2A medium. All strains were catalase positive and oxidase was congruently positive only for strains from the P4023^T group. Isolates from both groups can be characterized and distinguished by physiological and biochemical tests listed in Table 2 along with tests useful for their separation from their closest phylogenetic neighbors. Characteristics specific for both groups are given in the description part as well as in a comprehensive formal format in Supplementary Table 11.

The analysis of fatty acid methyl esters showed that both groups of Antarctic isolates differ from each other in their major fatty acids (Table 3). The triad comprising strain P4023^T revealed five major fatty acids, specifically C_15:1 ω6c (18.1%), Summed Feature 3 (C_16:1 ω7c/C_16:1 ω6c) (9.9%), anteiso-C_15:0 (9.8%), iso-C_15:0 (9.3%) and iso-C_16:0 3OH (9.3%). The major fatty acid of the group comprising strain P7388^T were iso-C_15:0 (23.4%), Summed Feature 3 (C_16:1 ω7c/C_16:1 ω6c) (9.9%), iso-C_17:0 3OH (8.5%) and iso-C_15:0 3OH (8.4%). The proposed species differ not only qualitatively and quantitatively in the profiles of major fatty acids, but also quantitatively in minor fatty acids, especially iso-C_14:0, iso-C_16:1 H and Summed Feature 9 (C_16:0 10-methyl/iso-C_17:1). The overall fatty acids profile of both groups is in agreement with the genus description (Bernardet and Bowman, 2015) and it is similar to those of the reference strains with mostly quantitative differences. A significant difference is the higher presence of C_15:1 ω6c detected in the P4023^T triad that enables its clear separation from the compared strains.

The quinone system of strain P4023^T was composed of 94.3% menaquinone MK-6, 5.4% MK-5 and 0.2% MK-7, and that of strain P7388^T was 90.2% MK-6, 9.6% MK-5 and 0.2% MK-7. The polar lipid profile of strain P7388^T (Figure 6A) consisted of the major lipids, phosphatidylethanolamine, an ornithine lipid, two unidentified lipids lacking a functional group (L3, L4), moderate amounts of unidentified lipid L1, minor amounts of lipids L2 and L6 and an unidentified glycolipid (GL). Strain P4023^T (Figure 6B) displayed major amounts of phosphatidylethanolamine, an ornithine lipid, and two unidentified lipids (L3, L4) lacking a functional group, moderate amounts of unidentified lipid L1 and glycolipid GL and minor amounts of an unidentified aminophospholipid (APL), and two unidentified lipids (L2, L5). Additionally, a yellow pigment spot was visible (yPigm). The polyamine pattern of strain P4023^T consisted of the major polyamine sym-homospermidine [27.3 μmol (g dry weight^–1)] and minor amounts of putrescine [0.7 μmol (g dry weight^–1)] and spermidine [0.2 μmol (g dry weight^–1)]. The polyamine pattern of strain P7388^T was similar with the major polyamine sym-homospermidine [20.5 μmol (g dry weight^–1)] and minor amounts of putrescine [0.2 μmol (g dry weight^–1)] and spermidine [0.3 μmol (g dry weight^–1)]. Quinone systems, polar lipid profile and polyamine pattern are well in agreement with the genus description (Bernardet et al., 1996; Dong et al., 2013). Both strains differed in their polar lipid profiles from their close relatives F. hercynium KACC 14934^T and F. resistens KACC 14246^T (Kim et al., 2012). Specifically, the absence of a lipid labeled PDE (explanation not provided) was noted in the related strains, whereas in the novel strains we detected an aminolipid (OL) with the same chromatographic motility as PDE and corresponded to ornithine lipid 2 (Kawai et al., 1988). The presence of an ornithine lipid or an unidentified aminolipid with the same chromatographic motility as the ornithine lipid in Figure 6 has been reported for numerous Flavobacterium species including the type species of the genus Flavobacterium aquatile (Kämpfer et al., 2012, 2015, 2020; Lee et al., 2012).

Flavobacterium flabelliforme (fla.bel.li.for’me. L. n. flabellum, fan, vane; L. suff. -formis, -e, -like, in the shape of; N.L. neut. adj. flabelliforme, fan-like shaped, referring to fan-like shaped colonies).

Cells are Gram-stain-negative rods with rounded ends, cell size in range 0.4–0.6 μm × 1.2–2.4 μm, cells occurring in irregular clusters, occasionally singly or in pairs. Endospores are not formed. Does not produce a capsule. Negative for presence of flexirubin-type pigments. Does not adhere to agar. Yellowish colonies. Motile with gliding-activity. Catalase and oxidase positive. Growth occurs on R2A, PCA, TSA, marine agar NA, blood agar with 5% sheep blood, BHI, Mueller-Hinton and Endo agar. Does not grow on MacConkey agar. Grows in pH range 6–9 and temperature range 1–30°C with optimum growth at 20°C and pH around 7. Cells grow well in presence of 1% NaCl, and 2% NaCl inhibits growth. Does grow in microaerophilic conditions but growth in anoxic conditions is limited. Does not produce fluorescein on King B medium. No utilization of Simmon’s citrate, malonate and acetamide. Negative for reduction of nitrates and nitrites. Negative for production of urease and indole. Positive for hydrolysis of gelatine, casein, and tyrosine. Does not produce brown diffusible pigment on L-tyrosine agar. Negative for hydrolysis of Tween 80, aesculin, ONPG, starch, DNA, CMC and agar. Does not produce lecithinase. Does not produce H₂S. Positive for arginine dihydrolase, and negative for ornithine and lysine decarboxylases. Does produce acid from glucose and maltose in aerobic conditions. Negative for production of acid from fructose and xylose in aerobic conditions. Positive for utilization of glucose and maltose by API 20 NE. Negative for utilization of arabinose, mannose, N-acetylglucosamine, gluconic acid, capric acid, adipic acid, malic acid, citric acid and phenylacetic acid by API 20 NE. Positive for alkaline phosphatase, leucine arylamidase, valine arylamidase and acid phosphatase by API ZYM. Negative for esterase (C 4), esterase lipase (C 8), lipase (C 14), cystine arylamidase, trypsin, α-chymotrypsin, naphtol-AS-BI-phosphohydrolase, α-galactosidase, β-galactosidase, β-glucuronidase, α-glucosidase, β-glucosidase, N-acetyl-β-glucosaminidase, α-mannosidase, and α-fucosidase by API ZYM. Carbon source utilization ability via respiration determined by Biolog GEN III MicroPlate test panels is positive for D-maltose, α-D-glucose, D-glucose-6-PO₄, gelatine, L-arginine, L-aspartic acid, L-glutamic acid, acetoacetic acid and acetic acid. Negative for D-trehalose, D-cellobiose, gentiobiose, sucrose, stachyose, D-raffinose, α-D-lactose, D-melibiose, β-methyl-D-glucoside, D-salicin, N-acetyl-β-D-mannosamine, N-acetyl neuraminic acid, D-mannose, D-fructose, D-galactose, 3-methyl glucose, D-fucose, L-fucose, L-rhamnose, inosine, D-sorbitol, D-mannitol, D-arabitol, myo-inositol, glycerol, D-fructose-6-PO₄, D-aspartic acid, D-serine, L-alanine, L-histidine, L-pyroglutamic acid, L-serine, D-galacturonic acid, D-galactonic acid lactone, D-gluconic acid, D-glucuronic acid, glucuronamide, mucic acid, quinic acid, D-saccharic acid, p-hydroxy phenylacetic acid, methyl pyruvate, D-lactic acid methyl ester, L-lactic acid, citric acid, α-keto glutaric acid, D-malic acid, L-malic acid, bromo-succinic acid, Tween 40, γ-amino-butyric acid, α-hydroxy-butyric acid, β-hydroxy-D,L-butyric acid, α-keto butyric acid, propionic acid and formic acid.

The major fatty acids are iso-C_15:1 ω6c, Summed Feature 3 (C_16:1 ω7c/ C_16:1 ω6c), anteiso-C_15:0, iso-C_15:0 and iso-C_16:0 3OH. Major respiratory quinone is MK-6 and major polyamine is sym-homospermidine. Polar lipid profile contains major amounts of phosphatidylethanolamine, an ornithine lipid, and two unidentified lipids (L3, L4) lacking a functional group, moderate amounts of unidentified lipid L1, unidentified glycolipid GL, minor amounts of an unidentified aminophospholipid (APL), and two unidentified lipids (L2, L5). The DNA G + C content of the type strain is 31.2 mol%.

Type strain P4023^T (= CCM 9062^T = LMG 31963^T) was isolated in 2011 from the organic material of an abandoned bird nest located at the Lachman Cape (GPS: −63.778333 S, −57.781666 W). All characteristics listed in the species description are shared by all strains, except for the following strain-dependent test results: motility, production of acid from mannitol, and variable results observed by Biolog GEN III MicroPlate were utilization of dextrin, D-turanose, N-acetyl-D-glucosamine, glycyl-L-proline and pectin. A formal proposal of the species “Flavobacterium flabelliforme sp. nov.” is given in Supplementary Table 11.

The GenBank/EMBL/DDBJ accession numbers for the near full length 16S rRNA gene sequences and whole genome sequences for Flavobacterium flabelliforme sp. nov. P4023^T are MW691162 and JAGFBU000000000, respectively.

Flavobacterium geliluteum (ge.li.lu’te.um. L. neut. n. gelum, cold, frost; L. adj. luteus, yellow; N.L. neut. n. geliluteum, forming yellow colonies in the cold).

Cells are Gram-stain-negative rods with rounded ends, cell size in range 0.3–0.4 μm × 1.5–3.0 μm, occurring singly and in irregular clusters, occasionally in pairs. Endospores are not formed. Does not produce a capsule. Dark yellow to orange pigmented colonies. Flexirubin-type pigment present. Does not adhere to agar. Motile with gliding-activity. Catalase positive. Oxidase negative. Growth occurs on R2A, PCA, TSA, NA, blood agar with 5% sheep blood, BHI and Mueller-Hinton agar. Does not grow on marine and MacConkey agar. Grows in pH range 6–8 and temperature range 15–30°C with optimum growth at 20°C and pH around 7. Cells grow in presence of 0.5% NaCl, and 1% NaCl inhibits growth. Does grow in microaerophilic conditions but growth in anoxic conditions is limited. Does not produce fluorescein on King B medium. No utilization of Simmon’s citrate, malonate and acetamide. Negative for reduction of nitrates and nitrites. Negative for production of urease and indole. Positive for hydrolysis of gelatine, aesculin, ONPG, starch, casein, tyrosine and CMC. Does not produce brown diffusible pigment on L-tyrosine agar. Negative for hydrolysis of DNA and agar. Does not produce lecithinase. Does not produce H₂S. Positive for arginine dihydrolase and negative for ornithine and lysine decarboxylases. Does produce acid from glucose, maltose and xylose in aerobic conditions. Negative for production of acid from mannitol in aerobic conditions. Positive for utilization of glucose, arabinose, mannose, N-acetyl-glucosamine, maltose, and hydrolysis of aesculin by API 20 NE. Negative for utilization of gluconic acid, capric acid, adipic acid, malic acid, citric acid and phenylacetic acid by API 20 NE. Positive for alkaline phosphatase, leucine arylamidase, acid phosphatase, naphtol-AS-BI-phosphohydrolase and β-glucosidase by API ZYM. Negative for esterase (C 4), esterase lipase (C 8), lipase (C 14), valine arylamidase, cystine arylamidase, trypsin, α-chymotrypsin, α-galactosidase, β-galactosidase, β-glucuronidase, N-acetyl-β-glucosaminidase, α-mannosidase and α-fucosidase by API ZYM. Carbon source utilization ability via respiration determined by Biolog GEN III MicroPlate test panels is positive for D-trehalose, D-cellobiose, gentiobiose, N-acetyl-D-glucosamine, N-acetyl-D-galactosamine, α-D-glucose, D-mannose, D-glucose-6-PO₄, glycyl-L-proline, L-aspartic acid, L-glutamic acid, D-galacturonic acid, acetoacetic acid and acetic acid. Negative for sucrose, stachyose, D-raffinose, α-D-lactose, D-melibiose, N-acetyl-β-D-mannosamine, N-acetyl neuraminic acid, 3-methyl glucose, D-fucose, L-fucose, L-rhamnose, inosine, D-sorbitol, D-mannitol, D-arabitol, myo-inositol, glycerol, D-aspartic acid, D-serine, L-alanine, D-gluconic acid, D-glucuronic acid, glucuronamide, D-saccharic acid, quinic acid, p-hydroxy phenylacetic acid, L-lactic acid, α-keto glutaric acid, D-malic acid, L-malic acid, bromo-succinic acid, γ-amino-butyric acid, α-hydroxy-butyric acid, β-hydroxy-D,L-butyric acid, α-keto butyric acid, propionic acid, and formic acid.

The major fatty acids are iso-C_15:0, Summed Feature 3 (C_16:1 ω7c/C_16:1 ω6c), iso-C_17:0 3OH and iso-C_15:0 3OH. Major respiratory quinone is menaquinone MK-6 and major polyamine is sym-homospermidine. Polar lipid profile contains major lipids phosphatidylethanolamine, an ornithine lipid, two unidentified lipids lacking a functional group (L3, L4), moderate amounts of unidentified lipid L1, minor amounts of lipids L2 and L6 and an unidentified glycolipid (GL). The DNA G + C content of the type strain is 34.5 mol%.

Type strain P7388^T (= CCM 9064^T = LMG 31962^T) was isolated from water samples taken in 2016 from a small temporary lake (GPS: −63.795894 S, −57.809928 W). All characteristics listed in the species description are shared by all strains, except for the following strain-dependent test results: growth at 5 and 10°C, growth on Endo agar, hydrolysis of Tween 80, production of acid from fructose, presence of α-glucosidase in API ZYM and variable results observed by Biolog GEN III MicroPlate were utilization of dextrin, D-maltose, pectin, β-methyl-D-glucoside, D-salicin, D-fructose, D-galactose, D-fructose-6-PO₄, gelatine, L-arginine, D-galactonic acid lactone, mucic acid, methyl pyruvate, D-lactic acid methyl ester, citric acid and Tween 40. A formal proposal of the species “Flavobacterium geliluteum sp. nov.” is given in Supplementary Table 11.

The GenBank/EMBL/DDBJ accession numbers for the near full length 16S rRNA gene sequences and whole genome sequences for Flavobacterium geliluteum sp. nov. P7388^T are MW691150 and JAGFBV000000000, respectively.