Taxonomic relatedness between Pectobacterium carotovorum subsp. carotovorum, Pectobacterium carotovorum subsp. odoriferum and Pectobacterium carotovorum subsp. brasiliense subsp. nov



Shaza Nabhan, Institute for plant disease and plant protection, Hannover University, D-30419 Hannover, Germany; German Collection of Microorganisms and Cell Cultures (DSMZ) Inhoffenstrasse 7B, 38124 Braunschweig, Germany. E-mail:;



Pectobacterium carotovorum is a heterogeneous species consisting of two named subspecies, P. carotovorum subsp. carotovorum and Pcarotovorum subsp. odoriferum. A third subspecies, Pcarotovorum subsp. brasiliense, was previously proposed. The study aimed to confirm the subspecies status and validate the proposed name of Pcarotovorum subsp. brasiliense using a novel and standard microbial taxonomy.

Methods and Results

DNA-DNA hybridization confirmed that P. carotovorum subsp. brasiliense is a different species from Pwasabiae, Pbetavasculorum and P. atrosepticum, with 28, 35 and 55% similarity values, respectively, but is a member of the P. carotovorum species with 73–77% similarity values. Sequencing the entire 16S rRNA gene of two polymorphic copies from strains of each of the P. carotovorum subspecies demonstrated that the average 16S rRNA gene sequence diversity between Pcarotovorum subsp. brasiliense and Pcarotovorum subsp. carotovorum was lower than the maximum genetic distances between two sequence types obtained from the same strain. Multilocus sequence analysis based on eight housekeeping genes (mtlD, acnA, icdA, mdh, pgi, gabA, proA and rpoS) differentiated the subspecies and delineated two Pcarotovorum subsp. brasiliense clades.


Pectobacterium carotovorum subsp. brasiliense clade I was comprised of strains isolated from Brazil and Peru, while clade II included strains from Asia, North America and Europe. Strains in clade I but not clade II were phenotypically consistent with the original description of Pcarotovorum subsp. brasiliense in that they produced reducing substances from sucrose and acid from α-methyl glucoside. The type strain for Pcarotovorum subsp. brasiliense 212T (= LMG2137T = IBSBF1692T = CFBP6617T) was previously designated. The GC mol content of the type strain is 51·7%.

Significant and Impact of the Study

the study introduces a full description for the strains belonging to the two different clades assigned to P. carotovorum subsp. brasiliense.


Pectobacterium carotovorum, in the family Enterobacteriaceae, is a highly diverse species consisting of at least two valid names, Pcarotovorum subsp. carotovorum (Gardan et al. 2003) and Pcarotovorum subsp. odoriferum (Gallois et al. 1992), and a suggested third taxon, Pcarotovorum subsp. brasiliense (Duarte et al. 2004). Despite the lack of valid publication, the Pcarotovorum subsp. brasiliense name has been used in more than ten publications since first published in 2004 as Erwinia carotovora subsp. brasiliense (Duarte et al. 2004; El Tassa and Duarte 2004; Ma et al. 2007; Glasner et al. 2008; Naum et al. 2008; Czajkowski et al. 2009; Kim et al. 2009; van der Merwe et al. 2010; Williamson et al. 2010; Marquez-Villavicencio et al. 2011; Nabhan et al. 2011). Assigning strains to this taxon was based mainly on genetic information of the 16S-23S intergenic spacer region of the rRNA operon, partial sequence of 16S rRNA gene and multilocus sequence analysis (MLSA) of housekeeping genes.

Pectobacterium carotovorum subsp. brasiliense was first described as causing blackleg disease on potatoes (Solanum tuberosum L.) in Brazil and has since been described as also causing soft rot in Capsicum annum L., Ornithogalum spp., and Daucus carota subsp. sativus. Strains of this taxon were isolated in USA, Canada, South Africa, Peru, Germany, Japan, Israel and Syria (Duarte et al. 2004; Ma et al. 2007; Nabhan et al. 2011). About 20% of the P. carotovorum strains collected in Syria were identified as Pcarotovorum subsp. brasiliense (Nabhan et al. 2011).

Glasner et al. (2008) showed that Pcarotovorum subsp. brasiliense has a conserved core genome of 3·9 Mb in common with Pcarotovorum subsp. carotovorum (WPP14) and P. atrosepticum (strain SCRI 1043). However, 13% of genes in the chromosome of strain P. carotovorum subsp. brasiliense 212T were found neither in P. atrosepticum nor in Pcarotovorum subsp. carotovorum. Glasner et al. (2008) also suggested that Pcarotovorum subsp. brasiliense should have species status because of its unique genomic organization, an observation that was confirmed by the amplified fragment length polymorphism (AFLP) analyses of Nabhan et al. (2011).

The objective of this study was to determine the validity of the suggested taxon Pcarotovorum subsp. brasiliense using several different types of analyses. We considered both the classical approach to taxonomy to authoritatively propose the Pcarotovorum subsp. brasiliense nomenclature (Lapage et al. 1992).

Materials and methods

Strain collection

Strains used in this study (Table 1) represented the three subspecies of P. carotovorum, Pcarotovorum subsp. carotovorum, Pcarotovorum subsp. brasiliense and Pcarotovorum subsp. odoriferum. Pectobacterium carotovorum subsp. carotovorum strains were represented by isolates in two clusters from the previously reported MLSA study of Nabhan et al. (2011) and the type strain CFBP 2046T. Pectobacterium carotovorum subsp. brasiliense and Pcarotovorum subsp. odoriferum strains from several geographical origins were obtained from various sources and also included the type strains of P. carotovorum subsp. brasiliense (strain 212T) and P. carotovorum subsp. odoriferum (strain CFBP 1878T).

Table 1. Strains of Pectobacterium carotovorum included in the study
Bacterial strainHost PlantLocationReference or source
  1. CFBP, the French collection of plant pathogenic bacteria; JKI, Julius Kühn-Institut; SCRI, Scottish Crop Research Institute.

Pectobacterium carotovorum subsp. carotovorum
1Type strain (CFBP 2046T)CFBP
2SCRI2 Solanum tuberosum USASCRI
3A10·1 S. tuberosum SyriaNabhan et al. (2011)
4A18 S. tuberosum SyriaNabhan et al. (2011)
5C3 S. tuberosum SyriaNabhan et al. (2011)
6M30 S. tuberosum SyriaNabhan et al. (2011)
7C338 S. tuberosum SyriaNabhan et al. (2011)
8C144 S. tuberosum SyriaNabhan et al. (2011)
9C150 S. tuberosum SyriaNabhan et al. (2011)
10C267 S. tuberosum SyriaNabhan et al. (2011)
11C412·4 S. tuberosum SyriaNabhan et al. (2011)
12C137 S. tuberosum SyriaNabhan et al. (2011)
13C142·2 S. tuberosum SyriaNabhan et al. (2011)
14JKI4·3·8 Brassica oleracea GermanyNabhan et al. (2011)
Pectobacterium carotovorum subsp. odoriferum
15Type strain (CFBP 1878T)CFBP
16NB1892  Nabhan et al. (2011)
17JKI582Cichorium endivia/1985SwitzerlandNabhan et al. (2011)
Pectobacterium carotovorum subsp. brasiliense
18Type strain 212T (= LMG 2137T = IBSBF 1692T = CFBP 6617T)Duarte et al. (2004)
198 S. tuberosum BrazilDuarte et al. (2004)
20317 S. tuberosum BrazilDuarte et al. (2004)
21C18 S. tuberosum SyriaNabhan et al. (2011)
22A17 S. tuberosum SyriaNabhan et al. (2011)
23SCRI1073 S. tuberosum PeruNabhan et al. (2011)
241001S. tuberosum/stem 2007CanadaDe Boer
251033S. tuberosum/stem 2009CanadaDe Boer


Sequence data for 18 P. carotovorum strains were used from the previously obtained sequence files (Nabhan et al. 2011). For an additional five Pcarotovorum subsp. brasiliense strains, Pcarotovorum subsp. carotovorum CFBP 2046T and Pcarotovorum subsp. odoriferum CFBP 1878T, fragments of eight conserved housekeeping genes (mtlD, acnA, icdA, mdh, pgi, gabA, proA and rpoS) were amplified using PCR as described previously (Nabhan et al. 2011). The fragments were purified from agarose gels and sent for sequencing without cloning. Sequencing was carried out by SeqLab (Göttingen, Germany), and sequences were submitted to GenBank under accession numbers JF926762 to JF926841.

The housekeeping gene sequences were tested for accuracy by Chromas Lite 2.01 ( software and aligned using Clustal_W in mega version 5·05 (Tamura et al. 2011). The phylogenetic analysis was conducted using all nucleotide sites of each partial gene sequence [mtlD (461 bp), acnA (332 bp), icdA (552 bp), mdh (505 bp), pgi (543 bp), gabA (490 bp), proA (701 bp) and rpoS (852 bp)]. Based on the general time reversible model (GTR+G+I) (Nei and Kumar 2000), the informative maximum-likelihood tree was constructed in MEGA with a bootstrapping test of 1000 replications. As well, the bootstrapping phylogenetically derived tree for the amino acid sequence of the concatenated eight genes was computed based on the Jones–Taylor–Thornton (JTT+G+I) (Jones et al. 1992) (Fig. S1).

16S rRNA gene sequence analysis

The 25 strains (Table 1) were subjected to 16S rRNA gene analysis. The available genome sequence of strain Pcarotovorum Pc1 (NCBI GenBank CP001657) was used to obtain the seven copies of the 16S rRNA gene sequences. The sequences were aligned and used to design new primers for sequencing the entire gene. Primer pair Lpf/Rpr (Table S1) was selected from the flanking regions of the aligned consensus sequence. PCR was performed using a 30-μl reaction mix containing 2 μg μl−1 of each primer, 15 μl master mix (Phusion Flash F-548s, Biozym, Hameln, Germany), 12 μl H2O and 3·5 μl DNA template (50 μg μl−1). The PCR amplification programme consisted of an initial denaturation at 98°C for 10 s, 30 cycles of 98°C for 5 s, 62°C for 5 s and 72°C for 50 s, and a final 10-min elongation step at 72°C in a T-gradient thermocycler (Biometra, Göttingen, Germany). Amplified fragments were detected by electrophoresis on a 1·5% agarose gel and purified from the gel using a gel extraction kit (SeqLab, Fermentas, St Leon-Rot, Germany). The extracted DNA was cloned using CloneJET PCR Cloning kit in the vector pGET1·2/blunt. Taking into account the presence of multiple polymorphic copies of the 16S rRNA operon in species of Enterobacteriacae and knowing that P. carotovorum encodes seven rRNA operons (Glasner et al. 2008), we sequenced up to four transformed clones for each of the 25 bacterial strains in order to represent more than one 16S rRNA gene sequence type per strain. The clones were grown at 37°C for 4–5 h and the plasmids were extracted. The extracted plasmids were used for sequencing the inserted fragments using different primers (Table S1). Sequencing was performed by Seqlab (BigDye Terminator ready reaction mix v3·1; Applied Biosystems, Foster City, CA). Sequences were submitted to GenBank under accession numbers JF926716 to JF926761.

The 16S rRNA gene sequences were analysed by PubMLST ( to check for identical sequences. The DNA sequences were aligned in MEGA 5·05, and the phylogenetic relationships were inferred by using the maximum-likelihood (ML) method based on the Hasegawa–Kishino–Yano model (Hasegawa et al. 1985). ML trees were constructed in mega 5·05 representing each strain by the more representative 16S rRNA gene sequence type, available as Fig. S2, and representing all retrieved sequences from all strains, available as Fig. S3.

DNA-DNA hybridization (DDH) and GC% mol content

To obtain genomic DNA, bacterial cells were disrupted using a French pressure cell (Thermo Spectronic, New York). The DNA was purified from the crude lysate by chromatography on hydroxyapatite as described by Cashion et al. (1977). DDH was carried out as described by De Ley et al. (1970), with the modifications of Huss et al. (1983) using a model Cary 100 Bio UV/VIS spectrophotometer equipped with a Peltier-thermostat 6 × 6 multicell changer and temperature control with an in situ temperature probe (Varian). The extracted DNA from strain P. carotovorum subsp. brasiliense 212T was also used to determinate the GC% mol content from the ratio of deoxyguanosine (dG) to thymidine (dT) according to the method of Mesbah et al. (1989). The experiments were conducted at the facilities of the German Collection of Microorganisms and Cell Culture (DSMZ).

Polar lipid, lipoquinone and fatty acid composition

Chemotaxonomic characteristics were tested for five strains representing the different clades revealed in the phylogenetic analysis of P. carotovorum subsp. brasiliense 212T and A17, P. carotovorum subsp. carotovorum CFBP 2046T and C150, and P. carotovorum subsp. odoriferum CFBP 1878T. The strains were grown overnight at 28°C for 16 h and harvested by centrifugation. The collected bacterial cells were used to perform the analyses (polar lipid and respiratory quinone analyses were carried out by the Identification Service and Dr B.J. Tindall, DSMZ, Braunschweig, Germany). Fatty acid methyl ester analysis using the Microbial Identification System (MIDI Inc., Sherlock, DE, USA) was also performed using the same service for the strains provided on TSA medium.

Phenotypic assays

Carbon source utilization studies were carried out using the Microlog system (Biolog Inc., Hayward, CA). Twenty-one strains were assayed for the oxidation of the 95 carbon sources in the GN2 microplates as suggested by the manufacturer. Microplates were inoculated with 150 μl of suspension per well at 27°C. Additionally, all strains were assayed by tests commonly used for differentiating Pectobacterium subsp.: acid production from maltose and α-methyl-d-glucoside, reducing substances from sucrose, growth at 37°C, erythromycin sensitivity, tolerance to 5% NaCl, lactose fermentation and gas production from d-glucose (Schaad 1988).



The MLSA analysis (Fig. 1) revealed grouping of the Pcarotovorum subsp. carotovorum strains into the two clusters (clades III, IV) identified in the previous MLSA study (Nabhan et al. 2011). The type strain CFBP 2046T fell in clade IV with a high level of similarity to strain SCRI 2 (NCPPB275), which is still misidentified as P. atrosepticum in different international collections. The Pcarotovorum subsp. brasiliense strains also grouped into two clades, I and II. Strains in clade I were from Brazil and Peru, while clade II was comprised of strains from Europe, Asia, and North America, the latter having been identified previously as P. carotovorum subsp. brasiliense by MLSA. The two strains from Canada (including strain 1001 that had a low DDH similarity value) were clearly assigned to clade II. Strains of Pcarotovorum subsp. odoriferum, assigned to clade V, formed a separate homogeneous cluster. The topology of the MLSA tree based on the amino acid sequences of the eight genes (Fig. S1) reconstructed and supported a clade representing the P. carotovorum subsp. brasiliense and resolved clade I as a distinctly separate cluster supported with an 87% bootstrapping value.

Figure 1.

Phylogenetic reconstruction showing evolutionary relationships among Pectobacterium carotovorum subspecies based on multilocus sequence analysis of eight housekeeping gene fragments by the maximum–likelihood method. The bootstrap consensus tree is inferred from 1000 replicates. There were a total of 4354 positions in the final data set. Bootstrapping values <70% were cut off.

16S rRNA gene analysis

The 16S rRNA gene analyses were based on exactly 1530 bp representing the complete sequence of the gene. The 16S rRNA gene sequences were confirmed either by sequencing the same sequence type two times or by sequencing the reverse strand. The informative parsimony nucleotide sites were limited to 35 sites distributed mainly in four regions of the gene (positions 65–91, 210, 445–489 and 1001–1039, Escherichia coli 16S rRNA gene sequence numbering is indicated) provided online as Table S2, while there were 12 sites at which base substitution only occurred once which is not shown in Table S2. The sequence divergences among the three subspecies were estimated at 1·5% mean distance between Pcarotovorum subsp. odoriferum and Pcarotovorum subsp. brasiliense, 1·2% between Pcarotovorum subsp. odoriferum and Pcarotovorum subsp. carotovorum and 0·9% between Pcarotovorum subsp. brasiliense and Pcarotovorum subsp. carotovorum. The overall diversity among the 35 sequence types obtained from the 25 strains of the three subspecies was 0·825%, indicating that the strains all belong to the same species. Only two 16S rRNA gene sequence types were obtained for the three Pcarotovorum subsp. odoriferum strains, indicating 0·065% diversity. While maximum diversity in 16S rRNA gene sequences within subspecies was 1·39 and 1·125% for Pcarotovorum subsp. carotovorum and Pcarotovorum subsp. brasiliense, respectively, maximum diversity in the different copies of the gene within the same strain was 1·057% for Pcarotovorum subsp. brasiliense (strain 317). In the phylogenetic analysis of the 16S rRNA gene sequence, Pcarotovorum subsp. odoriferum strains clustered separately from Pcarotovorum subsp. carotovorum and Pcarotovorum subsp. brasiliense, but the latter two subspecies were not clearly differentiated from one another (Fig. S2).

DNA-DNA hybridization

DNA of strain P. carotovorum subsp. brasiliense 212T was used as the template for DDH with P. carotovorum CFBP 2046T, P. wasabiae CFBP 3304T, P. atrosepticum CFBP 1526T and P. betavasculorum CFBP 2122T, as well as for DDH analysis of an additional five Pcarotovorum subsp. brasiliense strains. The results confirmed that Pcarotovorum subsp. brasiliense was a different species from P. wasabiae, P. atrosepticum and P. betavasculorum with homologies of <70%, but with a high level of similarity (>70%) to Pcarotovorum subsp. carotovorum, suggesting assignment to this species (Table 2). The five Pcarotovorum subsp. brasiliense strains 212T, A17, 8, 317 and C18 were confirmed as belonging to the same taxon using the threshold value of 70% DNA-DNA similarity for defining bacterial species as per the recommendation of the ad hoc committee of bacterial systematics (Wayne et al. 1987) (Table 2). The identity of strain Pcarotovorum subsp. brasiliense 1001 with low DDH values is uncertain and needs to be confirmed using a different kind of analysis.

Table 2. Percentage DNA-DNA hybridization resulting from pairwise hybridization of DNA from strain Pectobacterium carotovorum subsp. brasiliense 212T with DNA from the type strains of the four species of Pectobacterium (Gardan et al. 2003) and from six additional strains identified as P. carotovorum subsp. brasiliense by MLSA
Hybridizing bacterial species and subspeciesPer cent DNA-DNA hybridization (%)
  1. Values in parentheses are results of duplicate measurements.

Pectobacterium wasabiae CFBP 3304T28·8 (35·0)
Pectobacterium betavascularum CFBP 2122T35·0 (41·0)
Pectobacterium atrosepticum CFBP 1526T54·9 (58·8)
Pectobacterium carotovorum CFBP 2046T73·6 (77·5)
P. carotovorum subsp. brasiliense clade II A1780·6 (84·7)
P. carotovorum subsp. brasiliense clade II 100173·4 (67·1)
P. carotovorum subsp. brasiliense clade I 31777·0 (78·7)
P. carotovorum subsp. brasiliense clade I 890·0 (91·6)
P. carotovorum subsp. brasiliense clade II C1883·0 (88·4)

Polar lipid, lipoquinone and fatty acid composition

The polar lipid pattern on thin-layer chromatography (TLC) showed that the phosphatidylethanolamine (PE) was a major component of the polar lipids in all strains. Additional polar lipids including phosphatidylglycerol (PG), phospholipids (PL2 and PL3) and aminophospholipid (PN) were also reported to be present in the five strains but at lower levels (Fig. S4). The analysis of respiratory lipoquinones revealed large amounts of ubiquinone-8 (Q-8) and ubiquinone-MK-8 (MK-8) in all strains. The whole-cell fatty acids of all the five strains were dominated by straight-chain acids, particularly C16:0, and considerable amounts of unsaturated fatty acids were detected mainly of the C18:1 ω7c type (Table S3).

Phenotypic assays

Only 13 of the 95 carbon sources tested were not metabolized by any of the P. carotovorum strains. Twenty-nine carbon sources were utilized by all strains in all subspecies. P. carotovorum strains demonstrated large differences in their ability to utilize different number of carbon sources ranged between 31 (strain A17) to 63 (strain SCRI2). The carbohydrate utilization patterns did not differentiate between Pcarotovorum subsp. carotovorum strains in clades III and IV with one exception that strains in clade III utilized glucose-1-phosphate along with a limited number of strains in all other clusters including Pcarotovorum subsp. carotovorum clade IV.

Strain Pcarotovorum subsp. brasiliense SCRI1073 (clade I) utilized 75 carbon sources and its phenotype was consistent with the original description of the subspecies based on Brazilian strains, in that it produced reducing substances from sucrose and acid from maltose and α-methyl glycoside (Table 3). Pcarotovorum subsp. brasiliense strains in clade II differed from clade I strains, in that they did not produce reducing substances from sucrose or utilized α-methyl glycoside. Furthermore, unlike clade I strains, clade II strains also neither produced acid from maltose nor utilized acetic and lactic acids. All Pcarotovorum subsp. odoriferum strains utilized Tween 80, d-psicose and d-sorbitol, while only a limited number of strains in all other clades utilized these compounds.

Table 3. Selected phenotypic characteristics of Pectobacterium carotovorum subsp. brasiliense (clades I and II), P. carotovorum subsp. carotovorum (clades III and IV) and P. carotovorum subsp. odoriferum (clade V)
 Biochemical testNumber of strains positive
Duarte et al. 2004 n = 16Clade I n = 1Clade II n = 5Clade III n = 7Clade IV n = 6Clade V n = 3
  1. n, is the number of tested strains; –, not tested.

  2. a

    Strain JKI4·3·22 produced a black reaction.

1Reducing substances from sucrose1610a003
2Growth at 37°C1615553
3α-Methyl glucoside1610003
4Tolerance of 5% NaCl15763
5Acid production from maltose10003
6Acetic acid810130
11Succinamic acid1613642
15Malonic acid011000
16L-Glutamic acid401000
24d,l-Lactic acid1110340
25d-Glucuronic acid011310
26d-Glucosaminic acid011220
27d-Gluconic acid313652
28d-Galacturonic acid1215762
29d-Galactonic acid lactone214541


Establishment of Pcarotovorum subsp. brasiliense as a separate taxon was initially based on differences in 16S rRNA gene sequence, amplification of the intergenic spacer region in PCR and analysis of biochemical reactions (Duarte et al. 2004). Later, it was validated by the MLSA analysis of Ma et al. (2007). The study of Ma et al. (2007) extended the diversity within the taxon by including strains from North America. A follow-up study by Glasner et al. (2008) using a pan-genomic approach suggested that Pcarotovorum subsp. brasiliense is phylogenetically distinct from the other pectobacteria and referred to it as a separate species. The notion of a separate species status for Pcarotovorum subsp. brasiliense, however, was not sustained by our DDH analysis which confirmed the subspecies status of the taxon. It is well known that DDH has limited capacity for differentiating among strains belonging to the same species and was therefore not expected to differentiate among the subspecies (Vandamme et al. 1996).

Our analysis of 16S rRNA gene sequences of Pcarotovorum subsp. brasiliense strains from both Brazilian and non-Brazilian sources failed to consistently substantiate their differentiation from other P. carotovorum strains on this basis (Fig. S2). The relative diversity in the 16S rRNA gene sequence types obtained from Pcarotovorum subsp. brasiliense and Pcarotovorum subsp. carotovorum strains and the low genetic distances prevented clustering of strains into individual groups representing the subspecies (Fig. S2). In the study of Gardan et al. (2003) on Pectobacterium species, the 16S rRNA gene similarity values were not analysed, and now, it is known that the analysis of partial 16S rRNA gene sequences fails to differentiate strains of Pcarotovorum subsp. carotovorum and Pcarotovorum subsp. brasiliense (Naum et al. 2008). With seven copies of the ribosomal gene in the chromosome, it is unknown how many different homologs of the 16S rRNA gene are present in a single strain. Both P. carotovorum subsp. carotovorum and Pcarotovorum subsp. brasiliense displayed 20 and 13 different 16S rRNA gene sequence types, respectively, in the 14 and 8 strains that we studied. This indicates that the apparent phylogenetic relationships will differ depending on which 16S rRNA gene sequence type of each strain is used in the analysis, and brings into question the accuracy of resulting evolutionary trees.

Informative nucleotides among the 16S rRNA gene sequence types, although few in number, are generally at the same positions for each of the three subspecies (Table S2). With the possibility of any of the four bases being present, there is a high probability of finding the same base in the same position for each of the sequence types representing strains of different subspecies. Base substitutions at discriminating nucleotide sites are exemplified at position 445 where all Pcarotovorum subsp. carotovorum sequence types have A but all Pcarotovorum subsp. brasiliense sequence types have G, and at position 489 where all Pcarotovorum subsp. carotovorum sequence types have T but all P. carotovorum subsp. brasiliense sequence types have C. However, taxon-related base substitutions such as those at positions 445 and 489 are not sufficient within the 16S rRNA gene to separate the two subspecies within the phylogenetic tree (Fig. S2). Such consistent base substitutions, however, can explain why strains of both Pcarotovorum subsp. carotovorum and Pcarotovorum subsp. brasiliense occur within the same cluster, and as in one case, Pcarotovorum subsp. brasiliense sequence types clustered clearly within a predominantly Pcarotovorum subsp. carotovorum 16S rRNA group (Fig. S2).

The 16S rRNA gene could not be used to discriminate among the subspecies, although it provided information on their relationships. A threshold of 97% similarity is generally accepted as the cut-off value required for discriminating species. However, the low level of 16S rRNA gene sequence diversity we observed among the three P. carotovorum subspecies was in agreement with what was previously reported among taxa of the family Enterobacteriaceae, and different studies reported very low discriminatory values which cannot guarantee species identity (Ibrahim et al. 1983; Kwon et al. 1997; Hauben et al. 1998; Spröer et al. 1999; Tindall et al. 2010). Staley (2006) confirmed that 16S rRNA gene lacks resolution below the genus level and that MLSA better resolves bacterial speciation. We observed good sequence stability among the three strains of P. carotovorum subsp. odoriferum (Fig. 1 and Figs S1 and S2). The uniformity among P. carotovorum subsp. odoriferum strains is consistent with its narrow host range on celery, chicory and hyacinth and restricted distribution in Europe (Gallois et al. 1992).

On the basis of our MLSA results, Pcarotovorum subsp. brasiliense was differentiated as a separate phylogenetic group from Pcarotovorum subsp. carotovorum in contrast to the DNA hybridization data and the 16S rRNA gene phylogeny. Of the 4354 nucleotides analysed per strain by MLSA, 465 were polymorphic informative sites and were used to define the three subspecies, Pcarotovorum subsp. odoriferum, Pcarotovorum subsp. carotovorum and P. carotovorum subsp. brasiliense. The 15% polymorphism in the eight concatenated MLSA sequences was significantly higher than the 3% polymorphism in the 16S rRNA gene sequences, reflecting a higher rate of divergence among the metabolic genes compared to the ribosomal operon and probably represents adaptation of various strains to specific environmental niches (Achtman 2008). The better resolution obtained by MLSA based on different regions within the genome interestingly revealed two clades, designated Pcarotovorum subsp. brasiliense clades I and II, within the subspecies but belonging to the same monophyletic group based on DDH. In other words, MLSA distinguished among strains with different adaptations at the infrasubspecies level.

The housekeeping genes analysed in this study are protein encoding. This means it is likely that MLSA measures a degree of phenotypic relatedness that cannot be resolved by the analysis of 16S rRNA gene. Technically, sequencing a sufficient number of housekeeping genes shared by taxa under investigation can provide a higher level of resolution than the nonprotein-coding genes. The nucleotide substitutions occurring in the protein-coding genes investigated in the study were informative at the protein level (Fig. S1). The advantage of using amino acid sequences is the lower substitution rate of amino acid sequences compared to nucleotide sequences (Stackebrandt 2006). Consequently, trees based on amino acid sequences show more support for evolutionary events than nucleotide-based trees. Clustering of P. carotovorum subsp. brasiliense in distinct groups indicated that the amino acid changes were conserved and express the relative fitness of strains belonging to this taxon adapted to their specific environment or host (Davies et al. 2000).

The amount of phospholipids (PL2) was clearly higher in both strains belonging to the subspecies P. carotovorm subsp. brasiliense compared to the three strains belonging to the other two subspecies, P. carotovorum subsp. carotovorum and P. carotovorum subsp. odoriferum. Additionally, PL1 phospholipids and AL1 aminolipids were present in both strains P. carotovorum subsp. brasiliense 212T and CFBP 2046T but absent from the P. carotovorum subsp. odoriferum strain CFBP 1878T. Both P. carotovorum subsp. brasiliense strains 212T and A17 had an additional two lipoquinones, Q-7 and MK-7, while the type strains CFBP 2046T and CFBP 1878T belonging to subsp. carotovorum and subsp. odoriferum, respectively, each had only one MK-7 and Q-7, respectively. P. carotovorum subsp. carotovorum C150 has both Q-7 and MK-7. In the Sherlock microbial identification system, only the two species P. carotovorum and P. atrosepticum are present. Consequently, the pectobacterial subspecies cannot be readily differentiated on the basis of fatty acid composition. The type strains of P. carotovorum subsp. brasiliense and P. carotovorum subsp. carotovorum both have the unsaturated fatty acids C15:1 ω8c and hydroxy fatty acids C15:0 3OH and cannot be differentiated from one another, while the incidence of the summed feature seven could be discriminating for the subspecies P. carotovorum subsp. odoriferum.

Phenotypic discrimination among the three subspecies using the traditional microbiological methods is challenged by the diversity of strains within each subspecies. Strains of Pcarotovorum subsp. carotovorum and Pcarotovorum subsp. brasiliense varied in their ability to utilize a number of carbon sources. Some strains of each subspecies use a high number of carbohydrates, while others use only a few. Clade I of Pcarotovorum subsp. brasiliense in the MLSA phylogenetic reconstruction includes all the strains of the subspecies isolated from Brazil and Peru. Phenotypically, these strains conform to the discriminating criteria of Duarte et al. (2004) and so far have only been isolated from potato (El Tassa and Duarte 2004). In contrast, strains in clade II are worldwide in distribution and were isolated from various hosts including Solanum tuberosum (Ex. Canadian and Syrian strains), L., Daucus carota subsp. sativus (SCRI 132) (Nabhan et al. 2011), Capsicum anuum L. and Onithogalum spp. (Ma et al. 2007; Nabhan et al. 2011). Phenotypically, the clade II strains differ from the discriminating criteria of Duarte et al. (2004), particularly in their inability to produce reducing substances from sucrose or acid from α-methyl glucoside, characteristics that had initially caused the clade I strains to be considered ‘high-temperature’ variants of P. atrosepticum. van der Merwe et al. (2010) reported the potato blackleg disease as being caused by Pcarotovorum subsp. brasiliense in South Africa but data presented in the paper were insufficient to determine the clade affiliation of these strains.

In summary, our DDH results clearly show that Pcarotovorum subsp. brasiliense strains are taxonomically aligned with the P. carotovorum species but at varying dis-reassociation values (Table 2). The MLSA resolved two clades within Pcarotovorum subsp. brasiliense as it did for P. carotovorum subsp. carotovorum (Fig. 1) but whether each clade ought to be considered a separate subspecies requires further study. Differentiation of Pcarotovorum subsp. brasiliense as a separate subspecies provides some clarity to the heterogeneity within the P. carotovorum species.

Description of Pectobacterium carotovorum subsp. brasiliense subsp. nov

Pectobacterium carotovorum subsp. brasiliense'se. N.L. neut. adj. belonging to Brazil is a subspecies of P. carotovorum and phenotypically conforms to the species description. It is a gram-negative and pectolytic bacterium, producing characteristic pits on pectate-based selective media, such as crystal violet pectate (CVP) medium. On CVP and nutrient agar, colonies of Pcarotovorum subsp. brasiliense are indistinguishable from other Pectobacterium species and subspecies. It grows at 37°C, distinguishing it from P. atrosepticum. The initial description of the subspecies indicated that Pcarotovorum subsp. brasiliense is distinguishable from Pcarotovorum subsp. cartovorum by its ability to produce reducing substances from sucrose and acid from α-methyl glucoside and maltose. Subsequently, strains were isolated that are identified as Pcarotovorum subsp. brasiliense because they group with Pcarotovorum subsp. brasiliense in MLSA analyses but are not producing reducing substances from sucrose and acid from α-methyl glucoside. The results presented in this study show that MLSA differentiates between two clades of Pcarotovorum subsp. brasiliense, placing the strains conforming to previous results in clade I and the remainder in clade II.

All strains ferment lactose, tolerate 5% NaCl, grow at 37°C and utilize d-trehalose and d-melibiose and with few exceptions d-galacturonic acid and cannot use glucose-1-phosphate, malonic acid, dextrin and d-arabitol. Strains in clade II cannot use acetic acid and d, l-lactic acid, whereas it can utilize N-acetyl-d-glucosamine. In contrast, strains in clade I can utilize succinamic acid and lactic acid but cannot utilize N-acetyl-d-glucosamine. The type Pcarotovorum subsp. brasiliense 212T (= LMG21371T = IBSBF1692T = CFBP6617T) was previously designated. The DNA base composition of strain Pcarotovorum subsp. brasiliense 212T was determined by HPLC to be 51·7 mol% G+C content.