Characterization of Edwardsiella Strains Isolated from Fish in The Netherlands (1995–2020) Shows a Relationship Between Edwardsiella Species and Source Authors: Janneke Allaart, Andries Kampfraath, Frank Harders, Michal Voorbergen-Laarman, Betty van Gelderen, Belén Fouz, Marc Engelsma Published: 2026 ABSTRACT To date, four species of the genus Edwardsiella—E. tarda, E. ictaluri, E. piscicida, and E. anguillarum—have been identified as the cause of disease and mortality in various fish species. Thirty-six Edwardsiella strains isolated from different fish species in the Netherlands were identified using multilocus sequence analysis (MLSA) based on sequence analysis of the genes gyrB, pgi and phoU. In parallel, an Edwardsiella matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) database was created by analyzing peptide spectra of a subset of these strains belonging to the four fish pathogenic species. MALDI-TOF MS analysis of the Edwardsiella isolates using the newly created database revealed 100% agreement with results of MLSA. Moreover, the presence of each Edwardsiella species was associated with specific fish host and source, with E. piscicida being the dominant species in Dutch aquaculture. In addition, a novel strain was isolated from electric eel, genetically diverging from the four established species. Whole-genome comparisons of the novel strain with those species showed DNA-DNA hybridization and average nucleotide identity scores suggestive of a new species. In conclusion, both MLSA and MALDI-TOF analysis can be used to identify Edwardsiella species. Moreover, it has been observed that different Edwardsiella species show a preference for certain hosts or sources. 1. Introduction Bacteria belonging to the genus Edwardsiella are considered to be among the most pathogenic bacteria affecting fish. Edwardsiella species are Gram-negative, motile, short, rod-shaped, facultative aerobic bacteria responsible for the disease commonly known as ‘fish gangrene’, ‘emphysematous putrefactive disease’ or ‘red disease of eels’, leading to mass mortality in various populations and age groups of fish [1]. In 1965, Edwardsiella tarda was the first Edwardsiella species identified as a distinct taxon within the family Enterobacteriaceae [2]. Edwardsiella ictaluri, causing enteric septicemia in catfish (ESC), was the next species identified in 1976 [3], and in 1980, a third species was identified from birds and reptiles: Edwardsiella hoshinae [4]. More recently, sequencing of the genome of six different E. tarda strains from different origins demonstrated two genetically distinct, polyphyletic groups [5]. Also, multilocus sequence analysis (MLSA) from twenty-seven E. tarda isolates supported the demonstration of two genetically distinct groups [6]. Subsequent genotypic and phenotypic analyses of fifteen of these isolates revealed a novel species within the genus Edwardsiella, for which the name Edwardsiella psicicida was proposed [7]. Simultaneously, a third variant was isolated from eel and sea bream characterized by phylogenetic analysis [7], as well as in another study from tilapia using repetitive sequence-mediated PCR [8]. A phylogenetic analysis based on the gyrB gene confirmed that these isolates belonged to the same subgroup, and the genetic variant was called E. piscicida-like [9]. In a subsequent study, more strains from the same variant were identified from eel and sea bream, and in addition, the variant was proposed as a new species: Edwardsiella anguillarum [10]. Consequently, the Edwardsiella genus now consists of five species. A study by Reichley and co-workers [11] evaluated phenotypic and genotypic analyses, microbial identification systems and molecular confirmatory methods for differentiating between the five Edwardsiella species. Phylogenetic analysis revealed that partial 16S rRNA had low discriminatory power. However, single-copy genes gyrB (encoding DNA gyrase subunit B) and sodB (encoding superoxide dismutase B) displayed high discriminatory power, while in the meantime maintaining high intraspecific similarity between Edwardsiella species. Also, the genes pgi (encoding glucose-6-phosphate isomerase) and phoU (encoding phosphate-specific transport system accessory protein) have a high discriminatory power [12,13]. Therefore, MLSA analyses based on these genes can be used for differentiation between the Edwardsiella species. Matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) is an emerging technology for rapid identification of microorganisms. With this method, bacteria can be identified based on their proteomic mass fingerprint after ionization of a sample and determining the mass-to-charge ratio (m/z) based on the time of flight. Compared to MLSA, MALDI-TOF strongly reduces the time for identification of isolated strains. At present, only representatives of E. tarda, E. ictaluri and E. hoshinae are present in the Bruker spectrum database. The Bruker MALDI-TOF method correctly identified all E. tarda, E. ictaluri, and E. hoshinae isolates, while other phenotypical tests failed to 100% discriminate between the three Edwardsiella species [11]. In the study, E. piscicida and E. anguillarum isolates were also identified as E. tarda; however, unique species-specific peptide mass peaks were observed in the spectral profiles for E. anguillarum, E. piscicida, and E. tarda, respectively, suggesting that MALDI-TOF analysis can offer rapid, reliable identification of the five Edwardsiella species [11]. Over the past few decades, Edwardsiella species have been regularly isolated from diseased fish in the Netherlands (unpublished data), making it a reoccurring problem in aquaculture as well as in wild and ornamental fish. Edwardsiella bacteria were isolated mainly from European eel, which is an important fish species in Dutch aquaculture. Furthermore, the bacterium was detected in turbot, common sole, catfish and several ornamental fish species. In the diagnostic process, all strains are identified at genus or species level by biochemical tests or MALDI-TOF analysis. In order to explore the presence of the different Edwardsiella species in Dutch fish and to validate reliable detection of the four fish-pathogenic Edwardsiella species by MALDI-TOF, a collection of thirty-six Edwardsiella strains isolated from diseased fish (different sources) and one environmental strain (from water) were identified to species level based on an MLSA of the gyrB, pgi and phoU genes. An Edwardsiella species main spectrum projection (MSP) library was created using MALDI-TOF and a selection of the typed strains. Subsequently, the new custom MALDI-TOF database was validated using strains belonging to the four different Edwardsiella species. 2. Materials and Methods 2.1. Edwardsiella spp. Isolates and Culture Conditions Thirty-six strains from diseased fish and one environmental strain (from water) collected from diagnostic samples from the Netherlands over the period 1995–2020 were used in this study (Table 1, further background information on clinical signs is provided in Supplemental Table S1). The strains were previously identified as Edwardsiella spp. by biochemical properties or MALDI-TOF analysis. Isolates were stored at −80 °C using glycerol (30%) stocks. Isolates were thawed and grown on Heart Infusion supplemented with 5% sheep blood (HIS) agar plates (in house WBVR) under aerobic conditions for 48 h at 22 °C. Subsequently, the strains were subcultured for another 48 h at 22 °C before using the strains for further analysis. 2.2. Lysate Preparation, PCR and Sequencing A full 10 µL inoculation loop of bacterial colonies from the agar plate (pure culture) was suspended in 500 µL distilled water. The suspension was heated for 10 min at 98 °C and subsequently 1:10 diluted and frozen at −20 °C until further processing. Amplifications were performed on a ProFlex PCR System (Applied Biosystems, Waltham, MA, USA). Primers used for PCR amplification and sequencing of the pgi, phoU and gyrB genes, originally described by Griffin and co-workers (2013; 2014), are listed in Table 2. For pgi, phoU, GyrB-1 and GyrB-2, amplification reactions (50 µL) were performed using 31.5 µL Super-Q water, 5 µL PCR buffer (10×), 5 µL MgCl2, 1 µL deoxynucleoside triphosphate mix (2.5 mM each), 0.2 µM of each primer, 0.5 µL AmpilTaq Gold polymerase (5 U/µL, Applied Biosystems) and 5 µL template DNA with the following cycling conditions: 3 min of denaturation at 94 °C; 45 cycles of 30 s at 94 °C, 30 s at 52 °C, and 2 m at 68 °C; and 7 min of extension at 68 °C. For GyrB-3 and GyrB-4, amplification reactions (50 µL) were performed using 18.5 µL Super-Q water, 10 µL SuperFi-Buffer (5×), 10 µL SuperFi-GC-enhancer (5×), 1 µL deoxynucleoside triphosphate mix (2.5 mM each), 0.5 µM of each primer, 0.5 µL Platina SuperFi DNA polymerase (2 U/µL, Invitrogen, Waltham, MA, USA) and 5 µL template DNA with the following cycling conditions: 30 s of denaturation at 98 °C; 35 cycles of 10 s at 98 °C, 10 s at 60 °C, and 1 m at 72 °C; and 5 min of extension at 72 °C. Amplicons were visualized using the E-gel system (Invitrogen) with a 2% agarose E-gel. The remaining PCR product was purified using QIAquick columns (Qiagen, Hilden, Germany). Purified PCR products were amplified in a sequence PCR using 2 µL Big Dye Reactionmix (BigDye v 3.1, Applied Biosystems), 1 µL Sequence-buffer (5×) (BigDye v 1.1, Applied Biosystems), 1 µL Super-Q water, 0.4 µM of each primer (Table 2) and 4 µL purified PCR product. Sequence PCR products were subsequently purified using Sephadex™ G-50 M DNA Grade (Cytiva, Marlborough, MA, USA) in 96-well AcroPrep™ filter plates (Cytiva). Sequencing was performed on the SeqStudio Genetic Analyzer (Applied Biosystems). 2.3. Phylogenetic Analysis of Isolates The raw sequence data were analyzed using Sequencher 5.4.6 (Gene Codes Corporation, Ann Arbor, MI, USA). The processed consensus sequences were aligned using MAFFT v7.475 [14] and the phylogeny of the different genes was reconstructed using maximum likelihood (ML) analysis with IQ-TREE software v2.0.3 and 1000 bootstrap replicates [15] and subsequently visualized using the R package ggtree [16]. For the concatenation of the gyrB, pgi and phoU gene sequences, the R package concatipede was used [17]. As a backbone for the phylogenetic tree, additional nucleotide sequences from Edwardsiella species were downloaded from NCBI GenBank [18]. The sequences obtained in this study are available under GenBank accession numbers from PZ797225 to PZ797369. 2.4. MALDI-TOF Database Creation and Validation A commercial MALDI-TOF mass spectrometer (Bruker MALDI-TOF Biotyper Sirius GP system, Bruker, Bremen, Germany) was used for generation of peptide mass spectral profiles (MSP). Based on the results of the MLSA, bacterial colonies from 1–5 strains per species plus an additional E. anguillarum reference strain (Ea BF1) were picked from the agar plate and suspended in 300 µL distilled water. Then, 900 µL ethanol (100%) was added and the suspension was mixed well. The suspension was centrifuged at 18.407 rcf and the supernatant was discarded. The centrifugation step was repeated once more and the pellet was dried for 5 min. Then 25 µL of 70% formic acid was added, mixed well and left for 3 min. After this, 25 µL acetonitrile was added and mixed well. The suspension was centrifuged at 18.407 rcf and the supernatant was used for further investigation. The supernatant was applied to eight spots on the MALDI-TOF target plate (1 µL per spot) and overlaid with 1 µL α-cyano-4-hydroxycinnamic acid (HCCA) matrix solution according to the manufacturer’s recommended protocol. Each spot was measured three times. Mass accuracy was calibrated using the Bruker Bacterial Test Standard. The spectra were captured in positive linear mode in a mass range of 2 to 20 kDa with a laser frequency of 200 Hz (IS1, 20 kV; IS2, 18.3 kV; lens, 6 kV; extraction delay time, 390 ns). Spectra were acquired in automatic mode by accumulating a maximum of 240 profiles (6 × 40 laser shots from different positions of the target spot). The peptide spectra were collected and analyzed using flexControl 3.4 software (Bruker Daltonics). Flatline spectra and spectra with other anomalies were removed. Furthermore, the spectra were analyzed for the presence or absence of the unique species-specific peptide mass peaks of E. anguillarum, E. piscicida, E. tarda and E. ictaluri [11]. A minimum of 20 spectra were used for MSP creation. The controlled spectra were loaded to the Maldi Biotyper Compass HT software v1.4 (Industry version, Bruker Daltonics) and MSPs were created. The new MSPs were added to a separated Edwardsiella database. After creating the database, bacterial colonies from the remaining strains were picked from the agar plate and applied to one spot on the MALDI-TOF target plate and overlaid with 1 µL α-cyano-4-hydroxycinnamic acid (HCCA) matrix solution. The strains were identified by MALDI-TOF and the results were compared with the obtained genetic data. A score value > 2.0 indicates a high probability of species-level identification; a score value > 1.7 suggests genus-level identification, while score values < 1.7 are generally considered unreliable for identification. Besides a score value, a consistency value is generated. Category A indicates high confidence, category B indicates moderate confidence, and category C means that the result is unreliable. 2.5. Whole-Genome Sequencing, Characterization and Comparison of Edwardsiella spp. The initial results from the Sanger sequencing showed a divergence of strain WBVR15015646 from the known Edwardsiella species. To explore this further, it was decided to fully sequence the strain. For whole-genome sequencing of strain WBVR15015646, genomic DNA was extracted with the Qiagen blood & tissue kit following the manufacturer’s instructions, and both long- and short-read sequencing was done. For the short-read library preparation, the Illumina DNA Prep kit (Illumina, Inc., San Diego, CA, USA) was used with 50 ng DNA input according to the manufacturer’s instructions. The recommended number of PCR amplification cycles was applied. Library quality was assessed using the Quant-iT™ dsDNA High Sensitivity Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) and the Agilent TapeStation 2200 (Agilent Technologies, Inc., Santa Clara, CA, USA) with the High Sensitivity D1000 ScreenTape assay. The library met the required quality criteria for DNA concentration and average fragment size. Short-read sequencing was performed on an Illumina NextSeq 2000 platform, with output targeted to achieve an average genome coverage of approximately 125×, assuming a genome size of 4 Mb. For long-read sequencing, libraries were prepared from 400 ng genomic DNA using the Oxford Nanopore Native Barcoding Kit 96 V14 (SQK-NBD114.96, Oxford Nanopore Technologies, Oxford, United Kingdom) according to the manufacturer’s instructions, and sequenced on an Oxford Nanopore PromethION using an R10.4.1 flow cell. Basecalling was performed with Dorado v 0.9.0 (Oxford Nanopore Technologies, Oxford, United Kingdom) with a highly accurate basecalling model (dna_r10.4.1_e8.2_400bps_hac@v5.0.0). Sequencing data were processed using an in-house bioinformatics pipeline. Read quality was assessed with FastQC v0.12.1. Hybrid assembly was performed with Hybracter v0.7.3 (hybracter hybrid) with a specified chromosome size of 4 Mb. Within this workflow, long reads were filtered and subsampled with Filtlong (default target depth 100×) and adapter-trimmed with Porechop_ABI, while short reads were quality-controlled with fastp. Long reads were assembled with Flye, followed by targeted plasmid recovery with Plassembler, long-read polishing with Medaka, reorientation of circularized chromosomes to begin with the dnaA gene using Dnaapler (v 1.3.0.) [19], and short-read polishing with Polypolish and pypolca. Each polishing round was scored with ALE, and the final assembly was selected according to Hybracter’s default selection logic. Sequences shorter than 200 bp (n = 10) were excluded, consistent with the standard contig length cutoff used for submissions to EMBL-EBI (project accession number PRJEB121486). Genome annotation was performed with Bakta (v1.11.4; database v6.0, full) [20], and assembly quality was evaluated using CheckM2 v1.1.0 with the corresponding reference database. To compare the Edwardsiella species, the following five Edwardsiella reference genomes were collected from GenBank: E. anguillarum ET080813 (GenBank: GCA_000264765.2), E. hoshinae (GenBank: GCA_016026395.1), E. piscicida (GenBank: GCA_021733145.1), E. ictaluri (GenBank: GCA_003074995.2), and E. tarda (GenBank: GCA_019933175.1). Assemblies were reoriented with Dnaapler (v1.3.0.). Alignments of the main chromosome were made with progressiveMauve (build date 25 Feb 2015) [21] and displayed using pyGenomeViz (v1.6.1) [22]. DNA-DNA hybridization (DDH) values were calculated in silico using the genome-to-genome distance calculator webtool (GGDC 3.0) [23]. Average nucleotide identity (ANI) values were calculated using FastANI (version 1.34) [24]. 2.6. Biochemical Analysis of Edwardsiella spp. To compare the biochemical properties of strain WBVR15015646 with the biochemical properties of the four different Edwardsiella species pathogenic to fish, a selection of Dutch isolates containing all different species and the E. anguillarum reference strain (Ea BF1) were used for biochemical analysis. The commercial API 20E system (bioMérieux, Marcy-l’Étoile, France) was used in accordance with the manufacturer’s instructions. Briefly, bacterial colonies were picked from HIS agar, and API 20E strips were inoculated and incubated for 24–48 h at 37 °C. All reagents were added, a seven-digit profile number was generated, and the results of the biochemical tests were used for comparison between the species. 3. Results 3.1. Genetic Identification Based on the genetic information from the MLSA, all strains could be identified as belonging to one of the described Edwardsiella species (Figure 1), with the exception of one strain isolated from an electric eel which was divergent from the other strains. The Edwardsiella species largely corresponded to the source: E. piscicida was mainly isolated from fish from aquaculture systems including eel, turbot and common sole. Furthermore, E. piscicida was isolated from wild eel and koi carp. E. tarda was only isolated from ornamental fish from zoos and traders of (sub)tropical fish and from a water sample from a catfish farm. E. ictaluri was recovered from both catfish and electric eel. E. anguillarum was not isolated from fish in the Netherlands during the study period. 3.2. MALDI-TOF: MSP-Profiles For E. tarda and E. piscicida, the MSP profiles created from one and five strains respectively were added to the custom Edwardsiella database. For E. ictaluri there were three strains isolated from the Netherlands added to the custom Edwardsiella database. For E. anguillarum one reference strain was available and added to the database. Strain WBVR15015646 isolated from electric eel diverged from the known Edwardsiella species in both the MLSA analyses and the MALDI-TOF MS analysis. It was therefore added with its own MSP profile in the custom database. Five E. piscicida strains, five E. tarda strains and three E. ictulari strains were used to validate the custom database. If the strain was present in the custom database, the result of the corresponding strain was removed from the validation results. For E. anguillarum and the novel Edwardsiella sp., only one strain was available, limiting the validation. Table 3 provides an overview of the strains used to create a custom MALDI-TOF database and to validate this custom database. Strains were tested twice, first using the Bruker MALDI-TOF database and subsequently using the Bruker MALDI-TOF database combined with the newly created Edwardsiella MALDI-TOF MS database. Using the standard Bruker database, the strains identified by MLSA as E. tarda and E. ictaluri were correctly identified (Table 4). In general, strains identified by MLSA as E. piscicida, E. anguillarum and the strain not belonging to any of the well-described Edwardsiella species were identified as Edwardsiella (score > 1.7) using the standard Bruker database, but could not be identified to species level (score > 2.0). Only one E. piscicida strain had a score value of 2.0 with a high consistency score (A), but the strain was misidentified as E. tarda. When testing the strains using the combined databases all strains were identified in accordance with their MLSA typing. Only the E. anguillarum strain still had a medium consistency score (B), probably because the entry is limited to the single E. anguillarum strain present in the database. The custom Edwardsiella spectrum database is available on request from the authors. Spectra of the strains selected for the MALDI-TOF database were also analyzed for the presence of the earlier-identified unique species-specific peptide mass peaks of E. anguillarum, E. piscicida, E. tarda and E. ictaluri [11]. However, these mass peaks were detected only in a subset of the analyzed strains. 3.3. Charaterisation of Edwardsiella sp. The genome of Edwardsiella sp. was assembled and compared to reference strains of the five known Edwardsiella species. The first Edwardsiella sp. assembly contained 18 contigs; these were annotated with Bakta. Examination of the DNApler output identified origins of replication in four contigs, three of which were circular; these three circular contigs were treated as the draft genome. Among the remaining five non-circular contigs, one encoded a repA gene and three contained an RNAI locus, suggesting the possible presence of an additional plasmid that was not sequenced to sufficient depth to allow complete reconstruction. As plasmid characterization was not the primary focus of this study, these contigs were not analyzed further. The three remaining contigs, consisting of one main chromosome and two plasmids, were re-annotated with Bakta individually (Figure 2A). With a size of 3.5 Mbp, this is the smallest Edwardsiella genome, containing the fewest coding sequences (Supplementary Table S2). While Edwardsiella sp. is most closely related to E. anguillarum and E. piscicida, the in silico DDH values below 70% and ANI scores below 95%, the generally accepted thresholds for species-level delineation [23,24], support its distinction from these species (Table 5). In addition, structurally the genome is also closest to E. piscicida (Figure 2B). While large genomic rearrangements can be seen between E. tarda and E. hoshinae versus the other Edwardsiella species, the most notable difference between Edwardsiella sp. and E. piscicida is the reduction in size, most likely due to multiple smaller deletions. 3.4. Biochemical Analysis The biochemical properties of strain WBVR15015646 were most similar to E. ictaluri and very different from E. piscicida, E. anguillarum and E. tarda (Supplementary Table S3). The API 20E system could not distinguish between the latter three species, which is in line with previous findings [11]. 4. Discussion To investigate the occurrence and severity of fish diseases caused by Edwardsiella species, accurate identification of the causative agent is of prime importance. Using genotypic analysis such as MLSA or whole-genome analysis, the five different Edwardsiella species can be readily distinguished [5,6,9,11]. However, both techniques are still relatively expensive and labor-intensive. Matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) offers a rapid and cost-effective method for the identification of bacteria. However, the accuracy of the method depends on the completeness of the database, and the identification of the Edwardsiella species is hampered by the absence of several species from this genus in the standard database provided by the manufacturer. In an earlier study, discriminatory peaks for each Edwardsiella species were described [11]. However, analysis of the mass spectra in our study revealed that these unique, species-specific peptide mass peaks were not consistently present in all spectra. Nevertheless, using a newly created MALDI-TOF database with spectra of the four described fish-pathogenic Edwardsiella species in addition to the standard Brüker database, all strains could be identified to species level. The consistency value was high for all species except for E. anguillarum (medium). Including additional E. anguillarum strains will most likely increase the consistency value. Furthermore, for full validation of the database, testing of additional E. anguillarum strains is necessary. With regard to MLSA of the strains from Dutch aquaculture, ornamental fish, wild fish and the environment, 27 were identified as E. piscicida, six as E. tarda, three as E. ictaluri and one as a potential novel Edwardsiella species. The species largely corresponds to the source: E. piscicida was isolated from cultured fish, wild eel and koi carp while E. tarda was isolated from ornamental fish from zoos and traders of (sub)tropical fish. These results may indicate a relationship between species identity and the source of isolation. The fact that E. piscicida was the dominant species in Dutch aquaculture in this study is consistent with previous research, in which MLSA-typed strains—isolated from diseased eel, turbot and whitefish from aquaculture in Norway, the UK and Finland—that had been identified as E. tarda were reclassified as belonging to the species E. piscicida [6,9,10,25]. Publications prior to 2012 on E. tarda should be regarded with some reservation as this could well be E. piscicida. Therefore, E. piscicida seems to be the major Edwardsiella species in aquaculture in North-Western Europe. In our study E. tarda was only isolated from ornamental fish and from a water sample from a catfish farm where no clinical signs were observed. There are very few studies that describe the isolation of E. tarda from ornamental fish species. E. tarda strains have been recovered from Asian swamp eel [26] as well as from Oscar fish [27]. Besides from ornamental fish species, E. tarda has been isolated from (Nile) tilapia [9,28,29], hybrid striped bass [9] and catfish [9,27,30], farmed in regions where relatively high seasonal water temperatures can occur (25–30 °C) compared with Dutch aquaculture in indoor recirculation systems (18–25 °C). Therefore, a preference of E. piscicida for low water temperatures and of E. tarda for high water temperatures—rather than a preference for a specific host—could therefore explain why, in our study, E. piscicida was isolated from diseased fish from aquaculture sources, whereas E. tarda was isolated exclusively from ornamental fish. However, in vitro data do not fully support this hypothesis; although E. piscicida may grow at a lower temperatures than E. tarda, E. piscicida has also been shown to grow at an in vitro culture temperature of 37 °C, which is much higher than the water temperature in fish husbandry systems [6,10,25,26,31]. Therefore, environmental factors other than water temperature may contribute to the susceptibility of fish to certain Edwardsiella species. E. tarda has zoonotic potential [9,10], whereas E. piscicida has thus far not been isolated from humans. The fact that only E. piscicida was isolated from fish from Dutch aquaculture and from wild eel from the Netherlands may explain why human cases of edwardsiellosis (which are generally attributed to E. tarda) are rare in the northern hemisphere and mainly occur in (tropical) regions [32]. E. tarda-infected ornamental fish in the northern hemisphere, imported from these tropical regions, potentially represent a zoonotic risk. However, case studies from ornamental fish are rare and only one case study on human edwardsiellosis traced to ornamental fish has been published so far [33]. E. anguillarum has not been isolated from fish in the Netherlands. The bacterium has not been isolated from the North of Europe so far. E. ictaluri was only isolated from catfish, the most frequent host, and electric eel [9,10]. The fact that the three strains isolated from the electric eels were genetically very different from each other shows that the electric eel is susceptible to several Edwardsiella species and may therefore serve as a “mixing vessel” for the emergence of new Edwardsiella species. One specific strain isolated from an electric eel (WBVR15015646) diverged from the previously described species, both genetically and in its protein spectra. The in silico DDH and ANI scores showed a close relationship to E. anguillarum and E. piscicida, also reflected in the phylogenetic tree based on the concatenated gyrB, pgi and phoU gene sequences. Structurally, based on the whole genome, the strain is closest to E. piscicida. However, the genome was smaller than the genome of the other species and differed sufficiently to be considered a separate species. Moreover, the biochemical properties were very different from those of E. piscicida and E. anguillarum strains. The strain originated from the electric organs of an electric eel at a Dutch zoo, where a high mortality rate among the recently imported electric eels was reported. It would be interesting to screen more Edwardsiella isolates from imported ornamental fish with MALDI-TOF to determine whether comparable strains are present. Subsequent genetic analysis could then confirm the existence of a sixth Edwardsiella species. 5. Conclusions In conclusion, in the recent Dutch aquaculture the Edwardsiella species present seems to be limited to E. piscicida, while in (sub)tropical ornamental fish E. tarda was primarily observed. MALDI-TOF analysis can be used to rapidly differentiate species within the genus for further research on host and environmental preferences as well as zoonotic risk of species. Adding additional E. anguillarum strains to the newly developed database will increase the consistency value of the measurements. Furthermore, strain WBVR15015646 isolated from electric eel diverged from the currently known Edwardsiella species. Collecting similar strains by typing isolates from (sub)tropical fish could support the description of this possible new Edwardsiella species.