Edwardsiella piscicida sp. nov., a novel species pathogenic to fish


  • T. Abayneh,

    Corresponding author
    1. School of Veterinary Medicine and Agriculture, Addis Ababa University, Debre-zeit, Ethiopia
    • Department of Food Safety and Infection Biology, Section for Microbiology and Immunology, Norwegian School of Veterinary Science, Oslo, Norway
    Search for more papers by this author
  • D.J. Colquhoun,

    1. Norwegian Veterinary Institute, Oslo, Norway
    Search for more papers by this author
  • H. Sørum

    1. Department of Food Safety and Infection Biology, Section for Microbiology and Immunology, Norwegian School of Veterinary Science, Oslo, Norway
    Search for more papers by this author


This article is corrected by:

  1. Errata: Corrigendum Volume 115, Issue 2, 634–635, Article first published online: 17 July 2013


Takele Abayneh Tefera, Department of Food Safety and Infection Biology, Section for Microbiology and Immunology, Norwegian School of Veterinary Science, Ullevålsvein 72, PO Box 8146 Dep, Oslo 0033, Norway. E-mail: TakeleAybaneh.Tefera@nvh.no or takeletefera99@gmail.com



This study describes a novel species within the genus Edwardsiella based on phenotypic and genetic characterization of fish pathogenic Edwardsiella isolates previously identified as E. tarda.

Methods and Results

Phenotypic characterization, DNA-DNA hybridization and phylogenetic analysis of representative Edwardsiella isolates from fish previously identified as E. tarda were conducted and compared with E. tarda type strain (ATCC 15947T). Phenotypically, strains from fish grow with pin-point colonies producing slight β-haemolysis under the colony. In contrast to the E. tarda type strain, fish strains did not grow at 42°C or degrade β-methyl-d-glucoside (with the exception of NCIMB 2034), citric acid and l-proline. With the exception of strain ETK01, all fish strains were highly pathogenic to zebra fish, while ATCC 15947T and NCIMB 2034 were nonpathogenic. DNA-DNA hybridization (DDH) levels between representative fish isolates and the E. tarda type strain ranged from 15 to 43·6%, while NCIMB 2034 hybridised with the type strain at the level of 63·2%. DDH values between the various fish isolates ranged from 68·2 to 93·9% defining a new and separate DNA hybridization group differing from the E. tarda type strain consistent with the findings of phylogenetic analysis, in which the fish isolates comprised a separate clade.


Phenotypical and genetic characterizations demonstrated that Edwardsiella isolates from fish described in this study do not belong to the species E. tarda or any of the previously established taxa within the genus Edwardsiella. The fish related strains studied here (excluding NCIMB 2034) represent, therefore, a novel species within the genus Edwardsiella for which we propose the name Edwardsiella piscicida sp. nov, with strain ET883T (NCIMB 14824T = CCUG 62929) as the type strain.

Significance and Impact of the study

The current finding will improve the diagnosis, understanding of the epidemiology and in establishment of effective control measures against this serious fish pathogen.


The genus Edwardsiella, a member of the family Enterobacteriaceae, was established by Ewing and MacWhorter (Ewing et al. 1965) following the isolation of the genus type species Edwardsiella tarda from human faeces. The genus currently consists of three species, that is, E. tarda, E. ictaluri and E. hoshinae. Unlike the other two taxa within the genus, E. tarda is known to be a versatile pathogen with a wide ecological niche and host range including various species of fish (Alcaide et al. 2006; Mohanty and Sahoo 2007) and other marine fauna, reptiles and terrestrial mammals including humans (Koshi and Lalitha 1976; Kourany et al. 1977; Janda and Abbott 1993; Manchanda et al. 2006). The bacterium can be found as part of the normal micro biota in healthy fish, amphibians and reptiles as well as in humans (Koshi and Lalitha 1976; Kourany et al. 1977; Van Damme and Vandepitte 1980; Vandepitte et al. 1983). Several Edwardsiella isolates including those from fish were later identified as E. tarda based on their phenotypic similarity to either the typical biotype or the rarely occurring atypical or bio-group I variant.

The typical biotype is characterized by production of H2S and its limited ability to ferment sugars (only glucose), while the atypical variant can ferment mannitol, sucrose and arabinose but is unable to produce H2S (Grimont et al. 1980).

Recent phenotypic and genetic studies have, however, revealed differences between isolates from fish and those isolated (in the main) from humans, including the type strain. Several studies reported variation in citrate utilization among E. tarda isolates (Coles et al. 1978; Baya et al. 1997; Acharya et al. 2007; Kumar et al. 2007; Wei and Musa 2008). Molecular studies have demonstrated that strains phenotypically identified as E. tarda were genetically heterogeneous with fish isolates showing marked genetic (Nucci et al. 2002; Abayneh et al. 2012, Yang et al. 2012) and whole protein profile divergence (He et al. 2011) from E. tarda type strain 15947T. The collective evidence suggests that isolates from fish previously identified as E. tarda may have been misclassified and may, therefore, represent one or more unrecognized taxa within the genus Edwardsiella. Because routine bacterial species identification is based mainly on phenotypic characters, molecular evidence needs to be supported by phenotypic markers that enable easy and rapid species differentiation. In this study, we define a novel species within the genus Edwardsiella based on phenotypic (physiological, biochemical and virulence features) and genetic characterization (DNA-DNA hybridization and phylogenetic analysis) of Edwardsiella isolates from fish that have long been classified as E. tarda.

Materials and methods

Bacterial strains

A total of 13 bacterial isolates from diseased fish from different geographical regions previously identified as E. tarda, and the E. tarda type strain (ATCC 15947T) were studied (Table 1). One E. ictaluri strain was also included for comparison of phenotypic characters and phylogenetic analysis. Four strains (ET883, LTB4, ET080813, NCIMB 2034) were further used for DDH with E. tarda type strain (ATCC 15947T).

Table 1. Edwardsiella strains of different geographical origins used in the study
StrainName as receivedHost or biological sourceOrigin
  1. TE. tarda type strain.

ET883 Edwardsiella tarda European eel (Anguilla anguilla)Greåker, Norway (1989)
ET2640 E. tarda European eel (Anguilla anguilla)Farsund, Norway (1989)
RM 298.1 E. tarda Turbot (Scophthalmus maximus)Southern Europe (2006)
HL 9.1 E. tarda Turbot (Scophthalmus maximus)Northern Europe (2006)
ETA1 E. tarda Turbot (Scophthalmus maximus)Scotland, UK (2007)
ETB1 E. tarda Turbot (Scophthalmus maximus)Scotland, UK (2007)
LTB4 E. tarda Turbot (Scophthalmus maximus)Qingdao, China (2006)
WY 18 E. tarda Turbot (Scophthalmus maximus)Qingdao, China (2006)
ETK01 E. tarda Korean catfish (Silurus asotus)Jeollabukdo, South Korea (2008)
ET080813 E. tarda Marbled eel(Anguilla marmorata)Qingdao, China (2008)
ET080814 E. tarda Japanese eel (Anguilla japonica)Qingdao, China (2008)
NCIMB 2056 E. tarda Sea bream (Evynnis japonicas)NCIMB collection
NCIMB 2034 E. tarda Unknown fish spp,NCIMB collection
ATCC 15947T E. tarda Human faecesKentucky, USA (1959)
AL93 E. ictaluri Channel catfishUSA

All Edwardsiella isolates were grown on blood agar base no 2 (Oxoid, Adelaide, Australia) containing 5% bovine blood overnight at 30°C. Cultures in log phase of growth were used for biochemical profiling, DDH and 16S rRNA gene sequencing.

Sequence data of E. hoshinae strain JCM1679 (acc. no. NR_024768.1) and species from related genera in the family Enterobacteriaceae including E. coli strain k-12 substr. MG1655 (acc. no. NC_000913.2), Salmonella species (acc. no. HQ267226.1 and NC_003197.1), Shigella flexineri (acc. no. NC_004741) and Serratia species (acc. no. CP000826.1 and NC_009832) were used for phylogenetic analysis.

Phenotypic characterization

Colony morphology and growth characteristics were observed after 24 h of incubation at 30°C. Growth characteristics at different temperatures (12°C, 25°C, 28°C, 30°C, 37°C and 42°C), NaCl concentrations (3, 5, 6, 7 and 8%) as well as under aerobic and anaerobic conditions were investigated.

Haemolytic activity was assessed following overnight culture on blood agar containing 5% bovine blood. Catalase activity was tested using standard procedures, while cytochrome oxidase was performed using BBL dry slide (Oxoid).

Biochemical characterization was performed using the API20E (BioMérieux, Marcy l'Etoile, France) and Biolog GN2 MicroPlates (Biolog Inc., Hayward, CA, USA) (in triplicates) according to the manufacturer's instructions.

Pathogenicity study

The pathogenicity of five fish isolates identified as E. tarda including ET883, LTB4, ETK01, ET080813 and NCIMB2034 as well as the E. tarda type strain (ATCC 15947T) was determined in challenge experiments utilizing TAB line zebra fish at the experimental challenge facility of the Experimental Biomedicine unit at the Norwegian School of Veterinary Science (NVH). The fish challenge experiments presented in this study were approved by the Norwegian Board of Animal Experimentation. Pathogen-free zebra fish were supplied by the NVH Ålestrom zebra fish laboratory. A total of 210 healthy TAB line adult (≥6 months) zebra fish of equal sex ratio were randomly assigned to 14 treatment tanks of 6 l capacity. Seven treatment groups in duplicates of 15 zebra fish were used for challenge with each strain. One duplicate group was used as a control. Fish were maintained in aquaria with continuous supply of oxygen and daily water turnover of one litre in each treatment tank. Ammonia was measured daily using TetraTest (Tetra GmbH, Herrenteich, Germany), while pH was monitored by using the Test PH kit (Tetra GmbH) and were maintained below 1·5 mgl−1 and 7·5, respectively. The water temperature was maintained at 22 ± 1°C throughout the experiment.

Fish in each duplicate group were injected intramuscularly (IM) with 3 μl of saline solution containing 106 CFUml−1 of each E. tarda strain. The same amount of sterile saline solution was administered to zebra fish in the control group. Fish were anesthetized by immersion in a water tank containing 0·005 mgl−1 (benzocaine/water) before IM injection. Fish were regularly monitored for a week during which the onset of clinical signs and mortality were recorded. Moribund fish were euthanized followed by dissection and aseptic sampling of liver and spleen for re-isolation of the bacterium used in the challenge. Cumulative mortality over a 7-day period was used to compare virulence between strains.

DNA-DNA hybridization (DDH) and comparison of genome to genome distance

DNA-DNA hybridization (DDH) was conducted using six representative E. tarda isolates including ET883, LTB4, ET080813, ATCC 15947, NCIMB 2034 and NCIMB 2056. Hybridization was carried out between the E. tarda type strain (ATCC 15947T) and four of the fish isolates (ET883, LTB4, ET080813 and NCIMB 2034). Hybridization of ET883 was also carried out against fish isolates LTB4, ET080813 and NCIMB 2056. Three grams of wet biomass of each representative isolate culture were harvested and suspended in 1 : 1 isopropanol and water solution (v/v) prior to DDH by DSMZ (Braunschweig, Germany).

The preparation of DNA and DDH were performed as described previously (De Ley et al. 1970; Cashion et al. 1977). Briefly, cells were disrupted using a Constant Systems TS 0.75 KW (IUL Instruments, GmbH, Konigswinter, Germany) followed by purification of the crude lysate DNA by chromatography on hydroxyapatite as described by Cashion et al. (1977). DNA-DNA hybridization was carried out as described by De Ley et al. (1970) with modifications described by Huss et al. (1983) using a model Cary 100 Bio UV/VIS-spectrophotometer (Cary 100 Bio, San Diego, USA) equipped with a Peltier-thermostate 6 × 6 multicell changer and a temperature controller with in situ temperature probe (Varian, Cary, NC, USA).

Calculation of genome to genome distance (GGD) between representative strains included in this study and the type strain, and estimation of the corresponding DDH values was carried out using genome to genome distance calculator (GGDC), a web-based software available at http://www.gbdp.org/species/ (Auch et al. 2010a,b). Concatenated partial sequence data of eight housekeeping genes from the Edwardsiella tarda strains (acc. no. JN700526 to JN700747) and whole genome shotgun sequence data of ATCC 15947T (acc. no. AFJG00000000) were used to calculate GGD. DDH values were estimated from the GGD utilizing the regression-based approach described previously (Auch et al. 2010a,b).

16S rRNA gene sequencing and phylogenetic analysis

Amplification of 16S rRNA gene sequence fragments was performed using the universal primers (forward: 27F: AGAGTTTGATCCTGGCTCAG; reverse: 1492R: GGTTACCTTGTT ACGACTT). The PCR products were purified using QIAquick PCR purification kit (Qiagen, Hilden, Germany) following the manufacturer's instructions, followed by sequencing by GATC Biotech (Konstanz, Germany).

Phylogenetic analysis of the Edwardsiella strains and other related genera within the Enterobacteriaceae family was performed based on 16S rRNA gene sequences and concatenated sequence alignments of 8 housekeeping genes (gyrB, mdh, adk, dnaK, phoR, metG, pyrG and aroE2) (acc. no. JN700526 to JN700747). Multiple sequence alignments were carried out using ClustalW2 algorithm followed by construction of phylogenetic trees using the neighbour-joining method in Mega ver. 5 (Tamura et al. 2011). The evolutionary distances used to infer the phylogenetic trees were computed using the Kimura 2-parameter method (Kimura 1980).

Nucleotide sequence accession numbers

Nucleotide sequence data of 16S rRNA gene of the Edwardsiella isolates reported in the current study has been deposited at GenBank under accession numbers KC119628, KC202809 and KC138721 to KC138731. 16S rRNA sequence data of Edwardsiella tarda strain ATCC 15947T (acc. no. NR_024770.1) and LTB4 (acc. no. EU259315.1) were obtained from nucleotide sequence databases.


All E. tarda isolates studied, independent of isolation source, displayed similar cellular morphology after Gram staining, that is, Gram-negative straight rods.

All strains were able to grow aerobically and anaerobically at 30°C on blood agar containing 5% bovine blood. Aerobically, the isolates from fish consisted of pin-point circular, slightly convex and glistening colonies after 24 hours of incubation with slight β-haemolysis visible only under the colony. Colonies of the E. tarda type strain (ATCC 15947T) and a single fish isolate (NCIMB 2034) displayed circular, convex colonies with a distinct narrow β-haemolytic zone barely extending beyond the edge of the colony after 24 h of incubation.

All studied strains were capable of growth at 25°C, 28°C, 30°C and 37°C but not at 12°C. Isolates from fish did not grow at 42°C unlike the E. tarda type strain (ATCC 15947T) and fish isolate NCIMB 2034, which grew with colonies visible on blood agar after 24 h of incubation. All strains were capable of growth under anaerobic conditions with pin-point colonies observed after 24 h. Both isolates from fish and the type strain grew in LB broth containing 3% and 5% NaCl but not in 6% or above. All strains were negative for cytochrome oxidase and positive for catalase activity.

Biochemically, all fish isolates had the typical biochemical features of the genus Edwardsiella with respect to the commonly used biochemical tests, similar to E. tarda in particular, identifying them from related genera of the family Enterobacteriaceae (Table 2). All showed API 20E profiles similar to E. tarda type strain ATCC 15947T being unable to ferment inositol, sorbitol, rhamnose, saccharose, melibiose, amygdalin, arabinose (except strain ET080813) and mannitol (except strains ET080813 and ET080814) (Table 3). All strains were negative for production of β-galactosidase, arginine dihydrolase, urease, tryptophan deaminase (TDA), gelatinase and for the production of acetylmethylcarbinol detected in the Voges Proskauer reaction and did not degrade Simmon's citrate except for strains ET080813, ET080814 and NCIMB 2056. All were able to produce acid from glucose, produced H2S and indole and were able to produce lysine decarboxylase and ornithine decarboxylase. In biochemical characterization using the Biolog GN2 microplate substrate panel, unlike the E. tarda type strain, fish strains did not degrade β-methyl-d-glucoside, citric acid and l-proline with the exception of NCIMB 2034, which was able to degrade β-methyl-d-glucoside. The fish isolates showed variable reactions in Tween 80, l-fucose, d-galactose, maltose, d-mannose, bromosuccinic acid, glucoronamide, l-asparagine, l-aspartic acid, l-glutamic acid, glycyl-l-glutamic acid, glycyl-l-aspartic acid, l-serine, uridine, glycerol and d,l-α-glycerol phosphate all of which were degraded by the E. tarda type strain. The E. tarda type strain did not degrade acetic acid, α-keto valeric acid and quinic acid to which the fish isolates showed variable reactions. Strain NCIMB 2034 was able to assimilate l-rhamnose and degrade formic and α-keto butyric acid-, characteristics lacking in all fish isolates and the E. tarda type strain (Table 3).

Table 2. Biochemical characteristics identifying ‘ET883-like’ and other Edwardsiella species from related genera of family Enterobacteriaceae
TestEdwardsiella speciesSalmonellaa Escherichia coli a Shigella spp a
ET883-like Edwardsiella isolatesEdwardsiella tarda a Edwardsiella ictaluri a Edwardsiella hoshinae a Flexneri/dysenteriae/boydii sonnei
  1. +: 90% or more of strains are positive; −: 90% or more of strains are negative; d: 11-89% of strains are positive; [−]: negative 75–89%; d, positive 25–74% (Holt et al. 1994).

  2. a

    Holt et al. 1994.

Indole production++[−]+d
H2S prodcution+++
Methyl Red+++++++
Citrate (Simmon's)+
Lysine decarboxylase+++++
Ornithine decarboxylase++++d+
d-Glucose, gas production++d++
Acid production from
Table 3. Phenotypic (biochemical and biophysical) characteristics identifying ET883-like Edwardsiella strains (n = 12) from closely related taxa within the genus Edwardsiella. Strain: 1, ET883; 2, ET2640; 3, LTB4; 4, WY18, 5, ETA1; 6, ETB1, 7, ETK01; 8, RM298.1; 9, HL 9.1; 10, ET080813; 11, ET080814; 12, NCIMB 2056; 13, NCIMB 2034; 14, ATCC 15947T; 15, Edwardsiella ictaluri (Al93); 16, Edwardsiella hoshinaea
CharacteristicET883-like Edwardsiella strains13141516a
  1. a

    Grimont, et al., 1981; [−]: negative 75–89%; d: positive 25–74%.

  2. b

    ND, no data.

Growth at
Citrate (Simmon's)+++
Methyl Red+++++++++++++++
Indole production++++++++++++++d
H2S production++++++++++++++
Acid production from
d-mannose+++++++++++++ ND
Tween 80++++ND
Acetic acid+ND
Citric acid+ND
Formic acid+ND
d-Glucosaminic acid++++++++++++ ND
α-Ketobutyric acid+ND
α-Keto Valeric acid+ND
Quinic acid+ND
Bromosuccinic acid+++++++++++ND
l-Aspartic acid++++++++++++ND
l-Glutamic acid++++++ND
Glycyl-l-aspartic acid+++++++++++ND
Glycyl-l-glutamic acid++++++ND
d,l-α-Glycerol Phosphate+++++++++++ND

In the pathogenicity study, Edwardsiella strains ET883, LTB4, ET080813 and ETK01 showed cumulative mortality of 100%, 95%, 87·5% and 11·1%, respectively 7 days post challenge (dpc). The E. tarda type strain and NCIMB 2034 were found to be nonpathogenic to zebra fish with no mortality recorded during the 7-day period. The respective strains were re-isolated from spleen and liver specimens of all moribund fish. Clinical signs included erratic swimming, bottom-dwelling and low interest to feed 2 dpc. Ulceration on dorsal surfaces and injection sites was the prominent clinical feature observed 3 dpc. The earliest mortality was recorded 2 dpc in the group infected with strain ET080813 and at 3 and 4 days in groups challenged with ET883 and LTB4, respectively. A single fish challenged with strain NCIMB 2034 showed dorsal ulceration, but no mortality was recorded during the remaining challenge period. Neither clinical signs nor mortality was observed in zebra fish challenged with the E. tarda type strain (ATCC 15947T) or in control groups.

Based on their phenotypic features, the isolates ET883, ET2640, LTB4, WY18, ET080813, ET080814, ETA1, ETB1, ETK01, RM298.1, HL9.1 and NCIMB 2056 were designated as ‘ET883-like’, while ATCC 15947T and NCIMB 2034 were designated as ‘E. tarda type strain-like’ isolates.

DNA-DNA hybridization of ET883, LTB4 and ET080813 with the type strain ATCC 15947 resulted in hybridization values of 15, 27 and 43·6%, respectively, indicating that these strains are only distantly related to the E. tarda type strain. Strain NCIMB 2034 showed 63·2% DNA hybridization with E. tarda type strain ATCC 15947T. DNA hybridization among the fish strains (ET883, LTB4, ET080813 and NCIMB 2056) ranged from 68·2 to 93·9% forming a new hybridization group separate from the E. tarda type strain (Table 4).

Table 4. Per cent DNA-DNA similarity (in 2 × SSC + 5% formamide at 70°C) among Edwardsiella isolates from fish previously classified as Edwardsiella tarda and the Edwardsiella tarda type strain
StrainATCC 15947TLTB4ET080813NCIMB 2056
  1. Values in parentheses are results of measurements in duplicate.

ET88315 (12·8)93·9 (95·3)79·1 (81·4)69·1 (64·3)
LTB427 (22·4)
ET08081343·6 (41·7)68·2 (63·7)
NCIMB 203463·2 (64·9)

The genome to genome distances and the corresponding estimated DDH values between the strains included in the current study are presented in Table 5. The estimated DDH derived from calculated GGD values showed that the strains formed two major hybridization groups comprising ET883-like (ET883, LTB4, ETA1, ETK01, ET080813, ET080814 and NCIMB 2056) and E. tarda type strain-like isolates (ATCC 15947T and NCIMB 2034) with intragroup DDH estimates of 68·41–85·93% and 70·28%, respectively. The estimated DDH values between ET883-like strains and the E. tarda type strain-like group ranged from 33·93 to 50·76% (Table 5).

Table 5. Genome to genome distance (GGD) and the corresponding DDH estimates among Edwardsiella strains based on concatenated sequence of eight housekeeping genes. Strain or taxa: 1, ET883; 2, LTB4; 3, ETA1; 4, ETK01; 5, ET080813; 6, ET080814; 7, NCIMB 2056; 8, NCIMB 2034; 9, Edwardsiella ictaluri (Al93); 10, ATCC 15947T
  1. GGD values between the corresponding strains were shown above the diagonal, while estimated DDH values (%) below the diagonal.

1  0·0102 0·0102 0·0102 0·0502 0·0502 0·0502 0·1004 0·06180·1273
285·93  0·0102 0·0102 0·0402 0·0402 0·0502 0·0908 0·06180·1262
385·9385·93  0·0102 0·0402 0·0402 0·0502 0·0904 0·06180·1265
485·9385·9385·93  0·0402 0·0402 0·0402 0·0904 0·06350·1266
568·4172·7972·7972·79  0·0102 0·0102 0·1000 0·05500·1289
668·4172·7972·7972·7985·93  0·0102 0·1000 0·05500·1288
768·4168·4168·4172·7985·9385·93  0·1000 0·05660·1286
846·3950·6250·7650·7646·5646·5646·56  0·11000·0459
963·2963·2963·2962·5866·3166·3165·6042·19 0·1243

A phylogenetic tree obtained from 16S rRNA gene sequences showed that all Edwardsiella isolates from fish fall in a separate cluster different from the E. tarda type strain (ATCC 15947T) with the exception of strains ET080813, ET080814, NCIMB 2056 and NCIMB 2034. Strains ET080813, ET080814 and NCIMB 2056 were clustered with E. ictaluri while NCIMB 2034 was clustered with ATCC 15947T. The E. tarda type strain-like isolates, E. hoshinae and species in other enteric genera (E. coli, Salmonella, Serratia and Shigella) comprised separate clades each (Fig. 1).

Figure 1.

Phylogenetic relationships of ‘ET883-like’ isolates to Edwardsiella tarda and related genera within the family Enterobacteriaceae. The tree was inferred from sequence alignments of 16S rRNA gene sequences (706 nts) using the neighbour-joining method. The numbers at the nodes of the branches show the percentage of replicate trees in the bootstrap test (1000 replicates). The bar shows the number of base substitutions per site.

In phylogenetic analysis based on concatenated sequence alignments of housekeeping genes including E. ictaluri and other enteric genera (Salmonella, Serratia, E. coli and Shigella), the ET883-like strains were resolved into a separate clade with the exception of NCIMB 2034, which was clustered with E. tarda type strain (ATCC 15947T) (Fig. 2).

Figure 2.

Neighbour-joining tree showing phylogenetic relationships of ‘ET883-like’ isolates to Edwardsiella tarda and related genera within the family Enterobacteriaceae inferred from concatenated sequences of eight housekeeping genes (4608 nts). The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) is shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The bar shows the number of base substitutions per site.


After the first description of E. tarda isolated from human gut (Ewing et al. 1965), several isolates phenotypically identified as E. tarda have been reported from different animal species including fish and wild animals (Alcaide et al. 2006; Mohanty and Sahoo 2007). Later genotyping studies have, however, showed that these strains (mainly from fish) were genetically divergent from the E. tarda type strain (Abayneh et al. 2012; Yang et al. 2012). A genotypic study based on MLSA of seven housekeeping genes revealed that isolates from fish previously identified as E. tarda were found to be phylogenetically divergent from the E. tarda type strain and the fish isolate NCIMB 2034 (Abayneh et al. 2012). A comparative phylogenomic study using whole genome sequences, housekeeping- and LPS-related genes also showed low genetic relatedness between fish E. tarda isolates and E. tarda strains mainly from humans Yang et al. (2012). The present study describes twelve divergent representatives based on phenotypic characteristics and genetic analysis (DNA-DNA hybridization and phylogenetic analysis) with the aim of determining the correct taxonomic placement of these strains. Despite previous reports of their genetic divergence from E. tarda type strain 15947T, the fish strains were indistinguishable from the E. tarda type strain with respect to the commonly used phenotypic identification methods such as in the API20E substrate panel (Tables 2 and 3) showing their closer phenotypic relationship to E. tarda than any of the other taxa in the genus Edwardsiella.

The differences in growth characteristics (at 42°C) and biochemical profiles observed in the Biolog GN2 microplate substrate panel between the E. tarda type strain (ATCC 15947T) and twelve of the thirteen fish isolates included in this study supports previous reports of low genetic similarity (Abayneh et al. 2012; Yang et al. 2012). Furthermore, DDH values between selected isolates from the fish group (ET883, LTB4 and ET080813) with E. tarda type strain (ATCC 15947T) were rather low (15–43%) far below the threshold 70% DDH required for inclusion within the same species (Wayne et al. 1987). These findings agree with a previous report of ~82% average nucleotide identity (ANI) between two of the fish strains (LTB4 and ET080813) with the E. tarda type strain (Yang et al. 2012), which is clearly lower than that expected (94%) between isolates of the same species. The DDH estimates obtained from GGD values between the Edwardsiella isolates from fish previously identified as E. tarda (except strain NCIMB 2034), and the E. tarda type strain was also far below (33·91–34·97%) the minimum 70% DDH threshold for inclusion within the same species.

Strains ET883, LTB4 and ET080813 formed a distinct DNA hybridization group (68·2–93·9%) with the lowest value (between LTB4 and ET080813) close to the 70% species definition threshold. Because DDH is only reproducible within the range of 10%, values in this range need to be interpreted with caution taking into account phenotypic characteristics of the strains being investigated. The phenotypic characteristics of LTB4 and ET080813 separating them from the E. tarda type strain and other taxa within the genus, combined with average nucleotide identities of more than 94% (Yang et al. 2012) suggests that these strains belong to the same species. Yang and colleagues (Yang et al. 2012) also demonstrated that LTB4 and ET080813 belonged to same genotype designated as EdwGI, which differed from the type strain ATCC 15947T (EdwGII) following concatenated amino acid sequence analysis of LPS biosynthesis genes. The high derived DDH values (68·14–85·93%) observed among ‘ET883-like’ strains, which were all from fish forming a distinct genogroup further consolidates the hypothesis that these strains belong to the same novel species.

Despite the close phenotypic and genetic similarities among ‘ET883-like’ the strains, strains such as ET080813, ET080814 and NCIMB 2056 also shared some biochemical characteristics (Table 3) different from other ET883-like strains, suggesting that these strains may represent subspecies within the same novel species. The phenotypic relationship of ET080813, ET080814 and NCIMB 2056 was consistent with 16S rRNA gene sequence and a previous phylogenetic study based on MLSA of seven housekeeping genes, in which these strains clustered tightly forming a single genogroup (Abayneh et al. 2012).

The difference in virulence observed between ET883 and ATCC 15947T-like strains in the zebra fish model further consolidated the genotypic delineation observed between these two groups, indicating that virulence in fish may be an important phenotypic feature for differentiation.

The phenotypical (growth, virulence and biochemical characteristics) similarities of NCIMB 2034 (a fish strain) with the E. tarda type strain observed in this study were in agreement with previous reports of their similarity in whole protein profiles (He et al. 2011), autoagglutination, haemagglutination, siderophore production and hydrophobicity (Wang et al. 2010), as well as their closer genetic relationship (Abayneh et al. 2012). The findings of phylogenetic analysis based on 16S rRNA (Fig. 1) and housekeeping genes observed in the current study (Fig. 2) are consistent with the phenotypic similarities of NCIMB 2034 and ATCC 15947T. The obtained DDH value between NCIMB 2034 and the E. tarda type strain, however, was lower (63%) than the minimum 70% threshold for species definition. Considering the accuracy of DDH and the robust evidence of the phenotypic and genetic similarity of NCIMB 2034 with the E. tarda type strain documented in the current and previous studies, strain NCIMB 2034, unlike the remaining fish isolates belongs to the E. tarda species. The GGD and the corresponding DDH values (Table 5) and phylogenetic analysis based on 16S rRNA and the housekeeping genes (Figs 1 and 2) indicated that the ‘ET883-like’ strains including the sub group with ET080813, ET080814 and NCIMB 2056 is genetically closer to E. ictaluri than to E. tarda. The failure of 16S rRNA phylogenetic analysis in discriminating three of the ET883-like isolates (ET080813, ET080814 and NCIMB 2056) from E. ictaluri unlike the phenotypic markers and MLSA is attributed to the lower resolving power of 16S rRNA gene sequence for bacterial species definition compared with phenotypic characters, DDH and MLSA (Staley 2006).

In conclusion, the phenotypical findings relating to growth characteristics and biochemical profiling and the results of genetic characterizations (DDH and phylogenetic analysis) obtained in this study indicate that, with the exception of strain NCIMB 2034, Edwardsiella isolates from fish included in this study, previously classified as E. tarda (ET883-like strains) do not belong to E. tarda or any of the previously established taxa within the genus Edwardsiella. On the basis of these findings, we conclude that ET883-like strains (ET883T, ET2640, LTB4, WY18, ETK01, ETA1, ETB1, RM 298.1, HL9.1, ET080813, ET080814 and NCIMB 2056) represent a novel species within the genus Edwardsiella for which we propose the name of Edwardsiella piscicida sp. nov. with ET883 as the type strain.

Description of Edwardsiella piscicida sp. nov

Cells are Gram-negative straight rod-shaped and facultative anaerobes. Colony morphology on blood agar after 24 h of incubation at 30°C is pin-point, circular, slightly convex, smooth and glistening with slight β-haemolysis visible only under the colony. The cells are negative for cytochrome oxidase and positive for catalase. Optimum temperature for growth is 28–30°C although growth can occur at 25°C and 37°C but not at 12°C or 42°C. Growth occurs at NaCl concentrations of 1–5% but not at 6% or above.

Acid is produced aerobically from glucose but not from inositol, sorbitol, rhamnose, saccharose, melibiose, amygdalin, arabinose and mannitol. The cells are positive for lysine decarboxylase, ornithine decarboxylase and produce H2S and indole. The cells are negative for β galactosidase, arginine dihydrolase, urease, TDA, Voges Proskauer, and do not degrade gelatin, β-methyl-d-glucoside, citric acid and l-proline. Variable characteristics are listed in Table 3. All strains of Edwardsiella piscicida were isolated from diseased fish and were found to be pathogenic for zebra fish. The type strain of the novel species is ET883T (NCIMB 14824T = CCUG 62929).


We appreciate Jan Roger Torp Sørby for kind cooperation during the challenge experiment. We would like to thank Dr Margaret Crumlish, (University of Stirling, Stirling, UK), Professor Dawn Austin (University of Herriot Watt, Aberdeen, UK), Dr Nuria Castro and Prof. Alicia Toranzo (University of Santiago, Compostela, Spain), Dr Sung-Woo Park (Kunsan National University, Kunsan, South Korea) and Dr Timothy Welch (National Centre for Cool and Cold Water Aquaculture, USDA, Kearnesville, WV, USA) for their kind cooperation in providing Edwardsiella isolates. We thank Gaute Skogtun for his help in 16S rRNA gene sequencing. This work was supported by Norwegian School of Veterinary Science.

Competing interests

The authors declare that there are no competing interests.