Brain, liver, heart, and spleen samples from three common terns were submitted to the Athens Veterinary Diagnostic Laboratory for aerobic and anaerobic bacterial culture at the University of Georgia (UGA). Tissues were rinsed with sterile phosphate-buffered saline (PBS) and processed in 3ml of brain heart infusion (BHI) enrichment broth (BD, Franklin Lakes, New Jersey, United States) using bead beater technology (Next Advance, Troy, New York, United States). After homogenization, the specimens were cultured on 5% sheep blood (BD, Franklin Lakes, New Jersey, United States), MacConkey, and Chocolate agar (Remel, San Diego, California, United States) and incubated at 42°C for 24–48h in aerobic conditions. Brucella agar (Remel, San Diego, California, United States) was also used to evaluate growth in anaerobic conditions. Matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry analysis (VITEK MS - bioMérieux, Marcy-lʹÉtoile, France) was used following the manufacturer’s instructions to attempt bacterial genus and species identification. Additionally, specific bacterial traits were tested, including catalase, oxidase, urease, and indole production. As per the manufacturer’s instructions, carbon source utilization testing was performed using a GEN III MicroPlate (Biolog, Inc., Hayward, California, United States).

Four additional isolates were obtained by the U.S. Geological Survey, National Wildlife Health Center (NWHC; Madison, Wisconsin) during routine disease investigations across the United States. Information on these isolates is presented in Table 1. Strains from these four birds were isolated by touching the tissue directly onto 5% sheep blood agar plates and streaking the inoculated area of the plate for isolation. Plates were incubated at 37°C for 24–48h under aerobic conditions. Isolates originating from NWHC were identified by sequencing the V1-V3 regions of the 16S rRNA gene as described previously (Shearn-Bochsler et al., 2018). Carbon source utilization testing on NWHC isolates was performed with commercially available kits and assays, including GEN III MicroPlates (Biolog, Inc., Hayward, California, United States), Analytical Profile Index (API) 20E test strips (bioMérieux, Inc., Durham, North Carolina, United States), and 4-MU discs (Key Scientific Products, Stamford, Texas, United States).

At UGA, DNA was extracted using the DNeasy Extraction Kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol. The concentration of total DNA extracted was determined using the Qubit double-stranded DNA assay kit (Thermo Fisher Scientific, Waltham, Massachusetts, United States). For the short-read sequencing using Illumina technologies (Illumina Inc., San Diego, California, United States), approximately 50ng of DNA was used for library preparation using Nextera DNA Library Prep kit (Illumina). Paired-end 300bp length reads were generated using the Illumina iSeq 100 System (Illumina Inc., San Diego, California, United States). Approximately 400ng of DNA was used for library preparation using Oxford Nanopore Technologies (ONT) library preparation SQK-RBK004 Rapid Barcoding Kit for the long-read sequencing. According to the manufacturer’s instructions, sequencing was performed on MinION with the FLOW-MIN 106 (R9.4 SpotOn) flow cell (Nanopore, Oxford, United Kingdom). Barcoded ONT sequencing reads were demultiplexed using Porechop v0.2.4¹ and used as input into Flye v2.6 assembler (Lin et al., 2016; Kolmogorov et al., 2019) with the expected genome size parameters set at 2Mb. Illumina paired-end reads were first qualitatively checked using FastQC v0.11.8 (Wingett and Andrews, 2018) and then submitted to Trimmomatic v0.36 (Bolger et al., 2014) to remove adapters and select reads with Phred scores >30. Illumina reads were then aligned to the ONT-based resulting genome assembly using BWA v0.7.17 (Li and Durbin, 2009). Both draft ONT assembly and alignment results were then submitted to Pilon v1.23 (Walker et al., 2014) for error correction and assembly base call polishing. Assembly metrics were evaluated using the Quality Assessment Tool for Genome Assemblies v5.0.2 (QUAST; Gurevich et al., 2013). The polished assembly was then submitted to the Circlator v1.5.3 pipeline (Hunt et al., 2015) to attempt automatic assembly circularization. The final genome sequence was annotated using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP; Tatusova et al., 2016; Haft et al., 2018) and deposited in GenBank under the accession number CP054053.

At NWHC, DNA was extracted from overnight cultures using a phenol: chloroform extraction method (Barbier et al., 2019). This DNA was sequenced on a Pacific Biosciences RSII at the Center for Genome Sciences, U.S. Army Medical Research Institute of Infectious Diseases (USAMRIID-CGS). Following the manufacturer’s protocol, sequencing libraries were prepared using the SMRTbell^™ Template Prep Kit (Pacific Biosciences, Menlo Park, California, United States). DNA (~5μg) was fragmented using gTUBE (Covaris Inc., Woburn, Massachusetts, United States) to ~20kb. After DNA damage and ends repair, blunt hairpin adapters were ligated to the template, and failed ligation products were digested with ExoIII and ExoVII exonucleases. The resulting SMRTbell templates were size selected on a BluePippin system (Sage Science, Beverly, Massachusetts, United States) using 0.75% dye-free agarose cassette with 4–10kb Hi-Pass protocol and low-cut set on 4kb. Size selected templates were cleaned and concentrated with AMPure PB beads. P6 polymerase was used in combination with the C4 sequencing kit, and 360-min movies were collected. HGAP3 (Chin et al., 2013) was used for de novo genome assembly. Assembled genomes were visualized using Gepard v1.30 (Krumsiek et al., 2007). When evidence that a contig/chromosome was circularly complete was observed, the redundant sequence was trimmed from the ends. The breakpoint was adjusted so that the linear representation of the circular sequence began at the first base of the dnaA gene. Partial sequences were left unmodified. All assembled sequences were then error-corrected using Quiver v0.90 (Chin et al., 2013). Finally, all assemblies were submitted to Genbank for annotation with PGAP (Tatusova et al., 2016; Haft et al., 2018) and deposited in GenBank under the accession numbers CP022010–CP022013.

Average nucleotide identity (ANI) and tetranucleotide signature correlation index (TETRA) scores (Richter and Rossello-Mora, 2009) were calculated using JSpecies (Richter et al., 2016) on the Progressive Mauve (Darling et al., 2004) alignment of the five Mergibacter whole-genome sequences obtained in this study (Table 1). Additionally, visualization of the Mergibacter alignment (Figure 1) was obtained using Blast Ring Image Generator (Alikhan et al., 2011).

Three phylogenetic analyses were performed in this study based on (i) Pasteurellaceae bacterial core genomes (ii) 16S rRNA gene, and (iii) on a Pasteurellaceae housekeeping multilocus sequence (MLS). First, the complete genomes or assemblies of 58 Pasteurellaceae strains were downloaded from NCBI (Supplementary Table S1). The sequence alignment for the core genome-based phylogenetic analysis included 41 out of 58 Pasteurellaceae strains and five Mergibacter genomes, and it was performed using Progressive Mauve (Darling et al., 2004), followed by phylogenetic tree estimation with RAxML (Stamatakis, 2014). Then, the nucleotide sequences of the 16S rRNA and four additional housekeeping genes were manually extracted from whole genomes of these 58 Pasteurellaceae representatives and from Mergibacter strains (Supplementary Table S1). The 16S rRNA gene [1,261bp in length; from nucleotide 51 to 1,312 of Pasteurella multocida 20N (GenBank: CP028926)] core portion was used to construct the 16S rRNA gene tree. The Pasteurellaceae housekeeping multilocus sequence-based tree was constructed using the concatenated core sequences of four Pasteurellaceae housekeeping genes used in previous taxonomic approaches for this family (Christensen et al., 2007, 2014): atpD [1,152bp in length; from nucleotide 137 to 1,289 of Pasteurella multocida 20N (GenBank: CP028926)], infB [1,511bp in length; from nucleotide 714 to 2,225 of Pasteurella multocida 20N (GenBank: CP028926)], recN [1,347bp in length; from nucleotide 35 to 1,382 of Pasteurella multocida 20N (GenBank: CP028926)], and rpoB [3,918bp in length; from nucleotide 88 to 4,006 of Pasteurella multocida 20N (GenBank: CP028926)]. The sequence alignment for the two gene trees (16S rRNA gene, housekeeping multilocus sequence) was performed with MAFFT (Katoh and Standley, 2013) in Geneious software (version 2020.0.4), and phylogenetic trees were estimated using RAxML (Stamatakis, 2014). All trees were visualized and edited in FigTree (v1.4.4).²

Average amino acid identity (AAI) was calculated using MAFFT (Katoh and Standley, 2013) on the Progressive Mauve (Darling et al., 2004) alignment of the 5 Mergibacter sequences and 27 representatives of the Pasteurellaceae family, including the most genetically closed related species: C. oris, Pasteurella testudines, and Gallibacterium species.

The taxonomic approach known as percentage of conserved protein (POCP) was applied to the M. septicus type strain A25201^T (=DSM 112696; CP054053) and its three most closely genetically related species (C. oris, P. testudines, and Gallibacterium genomosp. 3) as previously described (Qin et al., 2014; Barco et al., 2020). Briefly, the predicted CDS (coding DNA sequences) for the four mentioned species were extracted using Artemis (Carver et al., 2012). BLASTp algorithm³ was used to find conserved proteins with an E value of less than 1e⁻⁵, sequence identity >40%, and comparable region of the query protein sequence >50%. Then, the POCP between two genomes was calculated as [(C1+ C2)/(T1+ T2)]×100%, where C1 and C2 represent the conserved number of proteins in the two genomes being compared, respectively, and T1 and T2 represent the total number of proteins in the two genomes being compared, respectively (Qin et al., 2014).

Finally, the taxonomic strategy for genus demarcation within the Pasteurellaceae family, which was previously described (Christensen and Bisgaard, 2018), was used in this study to support the reclassification of Bisgaard taxon 40 as the new genus Mergibacter. Briefly, the sequences of 31 conserved CDS selected (Christensen and Bisgaard, 2018) were manually extracted from M. septicus A25201, C. oris, Pasteurella testudines, and Gallibacterium species and concatenated using Geneious software (version 2020.0.4). Then, multisequence amino acid alignments were performed, and inter genus pairwise similarities were estimated with MUSCLE (Edgar, 2004).

BLASTN v 2.2.9 (Altschul et al., 1997) was used to align the whole-genome sequences of the five Mergibacter isolates presented in this study against the Virulence Factors Database (VFDB; Chen et al., 2005). The parameters used for the BLAST search were ≥70% gene identity and ≥0% sequence length. Additionally, virulence factors previously described in other Pasteurellaceae members (Chen et al., 2014; Christensen et al., 2014; Khamesipour et al., 2014; Furian et al., 2016), including those related to capsule, outer membrane proteins (ompA, ompH, oma87, plpB, psl), adhesins (ptfA, pfhA, tadD, hsf-1, hsf-1, hsf-2, ptfA, fimA, pfhA, tadD) iron metabolism (fur, exbD, tonB, hgbA, hgbB), superoxide dismutase (sodA, sodC, tbpA), sialidase (nanH, nanB), toxin (toxA, cdt), and hyaluronic acid synthetase (pmHAS), were manually searched in all the annotated genomes using Geneious software (version 2020.0.4). The corresponding DNA sequences of virulence factors of interest were then manually extracted, and comparative analyses with those of the reference strains most closely related to Margibacter and/or associated with avian disease insurgence in the Pasterellaceae family were performed. A multisequence alignment for each virulence factor was performed with MAFFT (Katoh and Standley, 2013), and phylogenetic relatedness was estimated with RAxML (Stamatakis, 2014). Alignments and phylogenetic trees were visualized in Gingr (v 1.1.1) from the Harvest Package (Treangen et al., 2014). As a result of the comparison, heatmaps containing the percentage similarity of each virulence factor were generated using GraphPad Prism version 8.4.3 (GraphPad Software San Diego, California United States).

Antimicrobial susceptibility testing was performed on Mergibacter type strain M. septicus A25201^T (DSM 112696) by broth microdilution using the Trek Sensititre system (Trek Diagnostic Systems, Cleveland, OH), followed by strip testing using Etest (bioMérieux). For the Trek Sensititre testing, individual colonies were selected from a blood agar plate and used to prepare a 0.5 McFarland suspension for inoculation of AVIAN1F plates (Thermo Fisher Scientific, Waltham, MA, United States), according to the manufacturer’s instructions. The minimum inhibitory concentration (MIC) values were visually confirmed using an inverted mirror after 24 and 48h of incubation at 42°C. The following antimicrobial compounds are included in the AVIAN1F plates: amoxicillin, ceftiofur, clindamycin, enrofloxacin, erythromycin, florfenicol, gentamicin, neomycin, novobiocin, oxytetracycline, penicillin G, spectinomycin, streptomycin, sulfadimethoxine, sulphathiazole, tetracycline, trimethoprim/sulfamethoxazole, and tylosin. Additionally, we used E-strips on Mueller-Hinton agar plates (bioMérieux) to evaluate the susceptibility for amoxicillin, amoxicillin-clavulanate, imipenem, and doxycycline, according to the manufacturer’s instructions.

We then searched for antimicrobial resistance genes (AMR) in all five whole-genome sequences analyzed, using three different databases: (i) Comprehensive Antibiotic Resistance Database (CARD) with criteria selected as perfect and strict, excluding nudge of loose hits to strict and high quality/coverage sequences⁴; (ii) ResFinder for acquired antimicrobial resistance genes, with default settings of 90% nucleotide similarity and a 60% of minimum length⁵ (Zankari et al., 2012); and (iii) AMRFinderPlus for acquired antimicrobial resistance genes, with default settings of −1.0 as the minimum identity for a blast-based hit and 0.5 as minimum coverage of the reference protein.⁶ Additionally, Plasmid Finder⁷ and COPLA plasmid taxonomic classifier⁸ databases were used to identify plasmids in all the five whole-genome sequences. Further, for the A25201 strain, three different assembly approaches were used as databases input: The Nanopore-based assembly, the Illumina reads-based assembly, and the polished assembly to investigate to which extent the assembly approach could influence the detection of plasmids.

Growth and phenotypic analysis were performed in parallel at the University of Georgia (UGA, n=1) and at the National Wildlife Health Center (NWHC, n=4) on five bacterial isolates. In addition, a preliminary taxonomic identification of the isolates was performed by amplification and sequencing of a portion of bacterial housekeeping gene the 16S rRNA, indicating an exact match of these five isolates with sequences of the Bisgaard taxon 40 from the Pasteurellaceae family [i.e., AY172732.1 (Christensen et al., 2003b); MG735704.1 (Knowles et al., 2019)].

Isolate A25201 was obtained at UGA from heart and liver samples of one common tern after 48h of aerobic incubation at 42°C. Isolate A25201 grew on blood agar presenting β-hemolysis and a shiny, circular, whitish-cream colony morphology (Supplementary Figure S1). Additionally, isolate A25201 was catalase and oxidase-positive, while urease and indole tests were negative. No bacterial growth was obtained on MacConkey and Brucella agar, and Gram-negative staining rods were identified when observing A25201 isolate under the microscope (data not shown). No bacterial identification was obtained from MALDI-TOF analysis. In contrast, GEN III Microbial ID analysis revealed a mixed identification of the isolates with a low percentage of probability, including Pasteurella canis/stomatis (51.6%), Histophilus somni (15.5%), Nicoletella semolina (9%), and Pasteurella pneumotropica (9%). Acid was formed from D-glucose, D-mannose, D-maltose, D-fructose, D-galactose, D-cellobiose, methyl pyruvate, inosine, and acetoacetic acid.

Similar growth characteristics and phenotypes were observed in the four isolates (27643, 27576, 16309, and 275176) processed at NWHC (β-hemolytic, shiny, circular, whitish colonies after 48h incubation, catalase-positive, oxidase-positive, urease negative, indole negative, and no growth of on MacConkey). Surprisingly, three of the four NWHC isolates differ from isolate A25201 in their ability to utilize various carbon sources. All the phenotypic characteristics comparable between the five strains are reported in Table 2, but overall, within the characters that separate the Bisgaard taxon 40 strains from the other Pasteurellaceae genera are the presence of β-hemolysis and acid formation from D-mannitol and sucrose (Table 2).

For UGA strain A25201, 229 ONT reads were assembled, resulting in one single contig of 1.900Mb length with no extra leftover contigs. A total of 1.760,502 Illumina paired-end reads were aligned to the ONT resultant draft genome, and the polished assembly consisted of one single contig with a total length of 1. 887Mb. No signs of circularization and overlapping ends were identified in silico. For the four NWHC strains, an estimated 590-731x coverage of the genome was obtained with single-molecule real-time (PacBio) reads, and assembly resulted in a single contig. In silico evidence for these contigs being circularly complete was observed in strains 27517 and 27576, presenting a genome size of ~1.887Mb in length. The de novo assembled genomes for the other two strains, 16309 and 26643, were 1.908Mb and 1.861Mb in length, respectively. The genomic features for the five strains are summarized in Table 3.

The genomes of the five Bisgaard taxon 40 strains identified and studied in this project showed high genome similarity, despite missing regions, as depicted in Figure 1. The ANI scores between the five genomes were 98.29–100%, well above the species boundary (Richter et al., 2016). Similarly, the TETRA analysis yielded correlation coefficients of 0.9985, confirming they were all isolates of the same species (Richter et al., 2016). The genomic alignment of the five isolates produced a 1,647,356bp consensus sequence with 1630; 1680; 1678; 1697; and 1677 annotated regions for strains A25201, 27643, 27576, 16309, and 27517, respectively. Of the annotated regions, 1468; 1462; 1460; 1479 and 1459 were, respectively, confirmed as protein-coding sequences (CDS).

The genetic relatedness of the five strains of this study with 41 representative members of the Pasteurellaceae family was assessed by whole-genome sequence alignments. The highest inter-genus nucleotide similarity percentages were 69.2–72.9% with Pasteurella, 70% with Frederiksenia, 70–70.4% with Mannheimia, 70.4–70.5% with Bibersteinia, 70.6–70.7% with Actinobacillus, 70.6–72.4% with Avibacterium, 70.9–71.1% with Caviibacterium, 70.9–71.1% with Chelonobacter, 71% with Glaesserella, 72.4–72.6% with Haemophilus, 72.7–73.0% with Histophilus, 72.8–72.9% with Conservatibacter, 73.3–73.4% with Volucribacter, and 73.6–75.4% with members belonging to Gallibacterium genus. A core genome-based phylogenetic analysis divided Pasteurellaceae into five main clades and consistently placed the five Bisgaard taxon 40 strains as a distinct, well-defined terminal monophyletic group (Figure 2). Although closely related, Bisgaard taxon 40 and genera Gallibacterium, Pasteurella, and Chelonobacter form differentiated clades under a common node indicating the identity of Bisgaard taxon 40 as a new genus.

Three different taxonomic tests were performed to determine the integrity of this finding. First, an AAI analysis using 764 CDS conserved across Pasteurellaceae showed that the five Bisgaard taxon 40 isolates shared the highest amino acid similarity percentages with Gallibacterium, ranging from 67.78–68% with G. genomosp. 2 to 69.8–70% with G. genomosp. 3. The AAI scores observed here are far below the AAI thresholds for separating potential genera in other bacterial families (Konstantinidis and Tiedje, 2007; Sangal et al., 2016), further supporting the belonging of the five subject strains to a new genus. Second, a POCP taxonomic analysis was performed on Bisgaard taxon 40 and its three most closely genetically related species (according to core genome phylogeny). Surprisingly, the results obtained by testing the taxonomy of the Bisgaard taxon 40 strains with this popular approach were inconclusive. The current POCP value proposed as a genus boundary for most prokaryotic groups is <50% (Qin et al., 2014). However, all the POCP values obtained in this study were above the 50% threshold even between members of supposedly genetically distant genera, such as Chelonobacter and Pasteurella, that shared the highest percentage (~79%) of conserved proteins in this study. Indeed, POCP values of 68.34, 63.28, and 60.81% were obtained between Bisgaard taxon 40 strain A25201 and G. genomosp. 3, C. oris, and P. testudinis, respectively (Supplementary Table S3). In addition, strain A25201 had a lower shared POCPs with the representative strains of its closest genera (Gallibacterium, Chelonobacter, and Pasteurella) than those genera members between each other. Finally, we performed a Pasteurellaceae-specific taxonomic approach based on the AAI of 31 randomly selected conserved protein sequences (Christensen and Bisgaard, 2018). Our results showed that the similarity percentages calculated between Bisgaard taxon 40 A25201 and the closest Pasteurellaceae members ranged from 68.2% with P. testudinis to 69.59% with G. genomosp. 3, well below the highest inter genus pairwise similarity (88%) identified between Pasteurellaceae members (Christensen and Bisgaard, 2018; Supplementary Table S4). These results recognize once again Bisgaard taxon 40 as a new genus within the Pasteurellaceae family.

Hence, based on this taxonomic evidence and the core genome phylogeny described above, we renamed Bisgaard taxon 40 as genus Mergibacter, with M. septicus type strain A25201 (DSM 112696; CP054053) as a type strain.

We tested the utility of 16S-based taxonomy to differentiate Mergibacter from other Pasteurellaceae genera. New sequence similarity and phylogenetic analyses were performed by using the 16S sequences extracted from the 58 Pasteurellaceae species and the five Mergibacter strains (Supplementary Table S1). In addition, 19 16S rRNA sequences downloaded from NCBI were included, to represent also some Pasteurellaceae species for which a whole-genome sequence was not publicly available (Supplementary Table S2). The five M. septicus strains showed high sequence similarities among each other, with nucleotide identities of 99.5–100% for the 16S rRNA gene. The highest nucleotide identities between Mergibacter and other Pasteurellaceae species were found with P. testudinis (92.2%), Spirabiliibacter pneumoniae (91.3–91.4%), Spirabiliibacter mucosae (91.8–92%), and members belonging to the genus Gallibacterium (91–92.3%). In our 16S rRNA-based phylogenetic tree, the five Mergibacter strains appeared on a distinct terminal group separated from its closest relatives by good bootstrap support values (Figure 3). This finding further supports Mergibacter as a new genus and the capacity of 16S-based taxonomy to identify Mergibacter representatives among other Pasteurellaceae. Still, some differences were observed between the core genome- and the 16S-based phylogeny estimated in this study. First, we observed that the support values for most of the nodes of the 16S were very low (in many cases <50%), even when 19 additional sequences were included in the study. In fact, the only terminal nodes that showed bootstrap values of 100% were found in the clade that separates Mergibacter from the other Pasteurellaceae genera (Figure 3). Second, the core genome-based phylogeny classified the Pasteurellaceae into five main clades. This topology is not as clear in the 16S tree. Additionally, the core genome phylogenetic analysis placed Gallibacterium as Mergibacter closest genus (Figure 2), while 16S-based phylogeny placed Gallibacterium in a very distant clade (Figure 3). This indicates that although the amplification and subsequent sequencing of the 16S rRNA gene are adequate to identify Mergibacter, this gene may not be the best representative to estimate phylogenetic relatedness among the members of the Pasteurellaceae family.

Considering the findings mentioned above, we decided to investigate a cost-effective alternative method for accurately identifying Mergibacter and classifying Pasteurellaceae. Housekeeping genes infB, recN, atpD, and rpoB have been used for this purpose in a previous study (Christensen et al., 2004). We extracted these four genes from the same 63 genomes (Supplementary Table S1) used in the 16S rRNA-based phylogeny, and for each of the genomes, we concatenated the cores of the four genes to create a representative multilocus sequence (MLS). Once again, the five M. septicus showed high similarities among each other, with nucleotide identities of 99.2–99.9% for the MLS, and the closest relatedness was found with members of the Gallibacterium genus, with nucleotide identity values of 80.2–83.3%, 76.6–82.7%, 62.8–64.8%, 82.2–83.5%, and 79.5–80.7%, respectively. The MLS-based phylogenetic tree showed a similar topology to the core genome-based tree with five distinct clades, and Mergibacer as a terminal monophyletic group in a cluster containing closest genera Gallibacterium, Chelonobacer, Spirilibacterium, and Testudinibacter (Figure 4). Although higher values are appreciated in the MLS tree, these are still below those observed in the core genome tree and genera Mesocribacter and Basfia appear misplaced (Figure 4). This suggests that our MLS approach is better than conventional 16S rRNA-based taxonomic methods but still inferior to core genome-based phylogeny.

Twelve putative virulence factors were identified in the genome of M. septicus type strain A25201 (Table 4), including genes encoding: a putative capsule (capsule), outer membrane proteins (ompA, ompH), a cytolethal distending toxin (cdt), iron metabolic proteins (exbD, fur), a superoxide dismutases (sodA), and proteins involved in lipooligosaccharide (LOS) synthesis (galU, galE, lpxA, lpxC, and kdsA). Because virulence factors are essential for establishing an infected host and provide insight into the evolutionary characteristic of bacteria (Parkhill et al., 2003), we performed a comparative analysis of each virulence factor identified in M. septicus A25201 with those of the other Mergibacer and other Pasteurellaceae avian pathogens. M. septicus A25201 presented 1 virulence factor (LOS synthesis lpxA gene) missing in the other isolates of this new genus but present in other Pasteurellaceae (Figures 5, 6). Nine virulence genes (galU, galE, lpxC, lpxA, kdsA, ompA, sodA, kdsA, and capsule) were shared by all Mergibacter members and most of the other Pasteurellaceae genera, and two putative virulence genes rarely found in common Pasteurellaceae avian pathogens, cdt and ompH, were found in all Mergibacer. For most of the shared virulence genes, Mergibacter showed the highest sequence similarity with members of the genus Gallibacterium (Figure 5), and in all resultant phylogenetic trees, it clustered on a different branch, which descended from the same node of Gallibacterium members (Figure 6).

The antimicrobial resistance phenotype of M. septicus type strain A25201 (DSM 112696; CP054053) was tested in this study. Isolate A25201 presented considerably low MICs for all the drugs tested, including penicillin, amoxicillin, and tetracycline (Table 5). Furthermore, no antimicrobial resistance genes were found in all the five Mergibacter genomes after searching on the ResFinder and ARMFinder databases. Of note, CARD database analysis identified sequence variants (92.6% of nucleotide identity) described in Escherichia coli elongation factor Tu (EF-Tu), conferring resistance to pulvomycin (Zeef et al., 1994). Comparison with phenotype for pulvomycin was not possible as it is not included in commercially available antimicrobial susceptibility panels. No plasmids were identified using Plasmid Finder and COPLA plasmid taxonomic classifier databases.

Mergibacter (Mer.gi.bac’ter. L. masc. n. mergus a seabird; N.L. masc. n. bacter a rod; N.L. masc. n. Mergibacter a rod of a seabird). Mergibacter is a new genus within the Pasteurellaceae family. The cells are Gram-negative staining rods, and no flagella or pili have been observed. Mergibacter was isolated on 5% sheep blood agar after 48h of aerobic incubation at 42°C, while no growth was observed on MacConkey. The colonies were β-hemolytic, shiny, circular, and whitish-cream. Positive reactions were observed for oxidase, catalase; negative tests were observed for urease and indole. Acid was formed from maltose and dextrin but not from D-mannitol, L-fucose, sucrose, and trehalose. A comparison of phenotypic characters separating the genus Mergibacter from the other Pasteurellaceae is given in Table 2. Mergibacter septicus is the type species of the genus.

Mergibacter septicus (sep’ti.cus. L. masc. Adj. septicus septic). In addition to characteristics included in the genus description, acid is formed from D-sorbitol and D-galactose. Most of the strains, including the type strain, are able to produce acid from D-mannose. Isolates have mainly been obtained from internal organs, such as the heart, lung, spleen, and brain.

The type strain is Mergibacter septicus A25201^T (DSM 112696), isolated from the heart and liver of one common tern (Sterna hirundo) from a mortality event in Florida, United States. In addition to the characteristics described above, using GEN III Microbial ID acid was formed from D-maltose, D-cellobiose, N-acetyl-D-Glucosamine, N-acetyl-β-D-Mannosamine, N-Acetyl-Neuraminic acid, D-Glucose, D-Mannose, D-Sorbitol, D-Glucose6-PO4, D-Fructose6-PO4, and acetoacetic acid. The genome of the type strain is 1,887,770bp long with a G+C content of 36.4%. The genome sequence is deposited in GenBank under the accession number CP054053.1.

The type strain A25201^T (Mergibacter septicus sp.; nov.) has been deposited in the DSMZ (German collection of microorganisms and cell cultures) under the number DSM 112696. In addition deposit at ATCC is in progress, AcqID-003043.