The use of two complementary DNA assays, AFLP and MLSA, for epidemic and phylogenetic studies of pectolytic enterobacterial strains with focus on the heterogeneous species Pectobacterium carotovorum




Amplified fragment length polymorphism (AFLP) markers and multilocus sequence analysis (MLSA) were used to analyse 63 bacterial strains, including 30 soft-rot-causing bacterial strains collected from Syrian potato fields and 33 reference strains. For the MLSA, additional sequences of 41 strains of Pectobacterium and Dickeya, available from the NCBI GenBank, were included to produce a single alignment of the 104 taxa for the seven concatenated genes (acnA, gapA, proA, icd, mtlD, mdh and pgi). The results indicate the need for a revision of the previously classified strains, as some potato-derived Pectobacterium carotovorum strains were re-identified as P. wasabiae. The strains that were classified as P. carotovorum during the analyses demonstrated high heterogeneity and grouped into five P. carotovorum highly supported clusters (PcI to PcV). The strains represented a wide range of host plants including potatoes, cabbage, avocados, arum lilies, sugar cane and more. Host specificity was detected in PcV, in which four of the six strains were isolated from monocotyledonous plants. The PcV strains formed a clearly distinct group in all the constructed phylogenetic trees. The number of strains phylogenetically classified as subspecies ‘P. c. subsp. brasiliensis’ in PcIV dramatically increased in size as a result of the characterization of new isolates or re-identification of previous P. carotovorum and P. atrosepticum strains. The P. carotovorum strains from Syria were grouped into PcI, PcII and PcIV. This grouping indicates a lack of correlation between the geographical origin and classification of these pathogens.


The pectolytic grouping of the Enterobacteriaceae family (formerly the Erwinia soft-rot group) is comprised of the genera Pectobacterium (Erwinia carotovora) and Dickeya (Erwinia chrysanthemi). The bacteria from these two genera have been classified as two separate soft-rot-causing pathogens; however, they do share some common biochemical and genetic characteristics (Dickey, 1979). The results of DNA–DNA hybridization experiments helped reclassify Pectobacterium chrysanthemi (Burkholder et al., 1953) into the genus Dickeya, where it was subsequently divided into five genetic species which, until now, have lacked diagnostic determination tests (Samson et al., 2005).

According to Gardan et al. (2003) the former species E. carotovora was incorporated into the genus Pectobacterium, and four of its established subspecies were renamed as P. carotovorum, P. atrosepticum, P. betavasculorum and P. wasabiae. None of the last three species were further subdivided into subspecies because of a low level of intraspecific genetic diversity (Avrova et al., 2002).

The species P. carotovorum is comprised of different variable strains, a small number of which are grouped under P. c. subsp. odoriferum (Gallois et al., 1992), some under ‘P. c. subsp. brasiliensis’ (Duarte et al., 2004), and yet more strains under P. c. subsp. carotovorum. Both P. c. subsp. odoriferum and ‘P. c. subsp. brasiliensis’ are described as atypical of P. atrosepticum because of their biochemical characteristics.

A potential classification problem arises from non-clustered P. carotovorum strains or strains with non-standard biochemical or genetic characteristics. Such deviations reduce the overall utility of using biochemical studies to characterize isolates of these pathogens. Moreover, there is significant genomic heterogeneity among the P. carotovorum strains classified by DNA–DNA hybridization (Gardan et al., 2003), further indicating that taxonomy of this highly divergent species is problematic.

The wide host range of P. carotovorum, which covers more than 35% of all angiosperm plant orders (Ma et al., 2007), and the ability of the P. carotovorum strains to survive in different environments (Toth et al., 2003) are the probable reasons for the high genetic diversity of the pathogen. Therefore, a clear taxonomical description of this group is vital to understanding the pathogenic differences and geographical adaptations of the species (Glasner et al., 2008).

Among the molecular methods used for classification, examination of amplified fragment length polymorphisms (AFLPs) is a technique used for genomic identification and delineation of bacterial plant and human pathogens. As AFLPs target multiple coding and non-coding loci in the genome, the technique is widely used (Portier et al., 2006). Avrova et al. (2002) successfully used AFLPs to analyse Pectobacterium strains and discovered extreme diversity within P. carotovorum. Classification studies based on multilocus sequencing analysis (MLSA) may provide phylogenetically relevant sequence information that can aid in a better understanding of speciation events in bacteria (Hanage et al., 2006). The results of previous analyses (Kim et al., 2009) indicate that MLSA is one of the most reliable methods for differentiating P. carotovorum strains.

The pectolytic bacteria are a group of economically very important plant pathogens on potatoes in Syria. Potatoes are grown in three sowing seasons in Syria and the production loss caused by pectolytic bacteria ranges from 1% to 2% up to 50%, mainly based on the high variation in irrigation conditions in different fields (Nabhan et al., 2009). In Syria, severe symptoms occur mainly in the hot spring and summer sowing seasons, where irrigation of the fields is often needed, and rarely in the cooler autumn sowing season. The pathogens P. carotovorum, P. atrosepticum and Dickeya spp. are the main causes of soft rot and blackleg disease on potato in Syria (Abu-Ghorrah et al., 2000; Nabhan et al., 2006). Evaluating 10 locally grown potato cultivars against pectolytic bacterial strain populations isolated from potato fields in Syria showed that some cultivars, such as Draga, Diamant and Anna, were significantly more resistant than others (Arenda, Lezetta and Nicola) towards these local strains.

This study aimed to elucidate the population structures of pectolytic enterobacterial strains, with an emphasis on P. carotovorum strains, and to determine the genetic diversity of strains sampled in Syria using AFLPs and MLSA.

Materials and methods

Two bacterial groups were included in this study of 63 strains (Table 1). Among them were 33 reference strains of Pectobacterium and Dickeya species and a collection of 30 pectolytic enterobacterial strains, which were isolated from infected potato plants sampled in 2002–2004 from different fields in Syria. The strains were biochemically characterized (Schaad, 1988), including characterization of their ability to utilize pectin on crystal violet pectate (CVP) medium. They were also analysed by PCR using primers Y1 and Y2 to identify P. carotovorum strains, primers ECA1f and ECA2r specific for P. atrosepticum, and ADE1 and ADE2 primers specific for Dickeya (Darrasse et al., 1994b; De Boer & Ward, 1995; Nassar et al., 1996).

Table 1.   Soft-rot enterobacterial strains included in this study
 Bacterial strainHost plantLocationSource
  1. Of the 33 reference strains, 14 were received as P. carotovorum subsp. carotovorum under the former name (Ecc); Ecc436 was re-identified as Dickeya spp.; Ecc1A-1, Ecc207 and Ecc4.2.6 were re-identified as P. wasabiae; Ecc568 and Ecc582 were re-identified as P. c. subsp. odoriferum; Ecc132 and Ecc4.3.22 were re-identified as P. c. subsp. brasiliensis.

  2. Nine strains were received as Patrosepticum under the previous name (Eca); Eca1073 strain was re-identified as P. c. subsp. brasiliensis; Eca2 and Eca3 were re-identified as P. carotovorum.

  3. SCRI: Scottish Crop Research Institute, Dundee, UK; GSPB: Göttinger Sammlung Phytopathogener Bacterien, Göttingen, Germany; JKI: Julius Kühn-Institute, Dossenheim and Quedlinburg, Germany.

  4. aReceived as P. atrosepticum.

  5. bExpected to belong to unknown species.

  6. cReceived as P. carotovorum subsp. carotovorum.

Pectobacterium carotovorum arborescens/CactusGermanyJKI
 2.2 (NCPPB275)aSolanum tuberosumUSASCRI oleracea JKI
 4.121 (NCPPB1640)bSaccharum spp./Sugar caneJamaicaSCRI
 5.109 (NCPPB929)bZantedeschia aethiopica/Arum lilySouth AfricaSCRI
 6.3 (NCPPB435)abS. tuberosum (Potato stem)ZimbabweSCRI
 7.102 (NCPPB547)bPersea americana (Avocado)IsraelSCRI
 29.A6.2, A10.1, A10.2, A16, A18, C3, C137, C140.2, C142.1b, C142.1s, C142.2 C143, C144, C150, C267, C331, C338, C364.2, C380, C412.4, M30, N78S. tuberosum (2002–2004)SyriaThis study
Pectobacterium carotovorum subsp. brasiliensis (BBA J594)S. tuberosumGermanyJKI
 31.132Daucus carota subsp. sativusJapanSCRI
 32.1073aS. tuberosumPeruSCRI
 37.A17, C18, A45, C317.1, C393.1S. tuberosum (2002–2004)SyriaThis study
Pectobacterium carotovorum subsp. odoriferum
 39.568cApium graveolens 1983SwitzerlandJKI
 40.582cCichorium endivia 1985SwitzerlandJKI
Pectobacterium atrosepticum
 41.M37S. tuberosumSyriaThis study
 42.1071 (NCPPB 549)S. tuberosumUKSCRI
 43.8S. tuberosumNetherlandSCRI
 44.185S. tuberosum JKI
 45.17A-1S. tuberosumPolandJKI tuberosum JKI tuberosum JKI
Pectobacterium wasabiae
 48.1A-1cS. tuberosumPolandJKI
 49.207 (NCPPB1274)cS. tuberosum 1962IrelandSCRI (St. 95)cS. tuberosum 1989GermanyJKI
Pectobacterium betavasculorum
 51.Pb NB 2122Beta vulgaris Prof. G. Auling
Dickeya spp. (A15)Ipomoea batataUSAJKI (NCPPB1065)Zea mays 1961EgyptJKI  JKI
 55.SR260 (R.Montgomery)  JKI
 56.4610Chrysanthemum morifolium 1956USAGSPB
 57.30177Chrysanthemum sp. 1953USAGSPB
 58.30178Phalaenopsis sp. 1953 GSPB
 59.30179Saintpaulia ionanthaGermanyGSPB
 60.3937 (CFBP3855)Saintpaulia ionanthaFranceJKI
 61.436cS. tuberosum GSPB
 63.C89.1, C89.2S. tuberosumSyriaThis study

Bacterial cultures

Cultures of strains were obtained by inoculation of single colonies and stored in 40% glycerol at −80°C and in CRYOBANK (MSAT, UK). In addition, the strains were routinely plated on King’s B agar medium.

NCBI GenBank electronic data

In addition to the 30 strains sampled in Syria and the 33 reference strains (Table 1), publically available electronic data of sequenced housekeeping genes (Ma et al., 2007) from 40 strains of Pectobacterium and Dickeya were retrieved from the NCBI GenBank (Table S1). The sequences of the acnA, gapA, proA, icd, mtlD, mdh and pgi housekeeping genes of P. carotovorum strain P.c.1 (accession number CP001657) were also obtained from the GenBank database ( Including the GenBank data, 104 taxa were subjected to MLSA.

DNA extraction

A single bacterial colony was inoculated into 20 mL liquid Luria–Bertani (LB) medium and incubated overnight in a shaking incubator set at 27°C and 120 rpm. The bacterial cells were harvested from 1 mL of the 16-h-old LB cultures at a density of about 2 × 109 CFU mL−1. The DNA was extracted using the DNeasy® Blood and Tissue Kit (QIAGEN), following the instructions as described by the manufacturer. The extracted DNA was quantified on 1% agarose gels, with λ DNA serving as a reference ladder. The gels were examined with a Gel-Pro-Analyser (INTAS) and by photometry at 260 and 280 nm. All DNA samples were stored at −20°C until use.

AFLP analyses

AFLP analyses were performed as described by Malek et al. (2000), with minor modifications. First, 200 ng DNA from each strain were digested with restriction endonucleases and ligated. All pre-amplification reactions were conducted with primers lacking any selective nucleotides. The selective amplifications were made with nine primer combinations. A combination of IRD 700 end-labelled HindIII primers with no selective nucleotide (MWG Biotech) and unlabelled MseI plus two selective nucleotides (AA, AC, AG, CC, GA, GG, TC, TG and TT) was used. The PCR products were separated on 6% polyacrylamide gels (Sequagel XR) and analysed on a Licor-DNA-Analyser, Gene ReadIR 4300 (MWG Biotech). The reproducibility of a strain’s banding profile was tested and confirmed by separate AFLP pre-amplifications and end reactions from each of two independent DNA isolations. Ten arbitrarily selected strains were used for the reproducibility tests. Independent repeat reactions were run on the gels at different time points. The numerical analyses of the binary data were conducted in the famd 123 software (Schlüter & Harris, 2006), using the Jaccard similarity coefficient for computing genetic similarities. The dendrograms were constructed as UPGMA majority-rule consensus trees after 1000 bootstrap replicates.

MLSA analyses

Eight informative housekeeping genes were used in the multilocus sequence analysis. Following the study of Ma et al. (2007), fragments of the seven metabolic genes: aconiate hydrase 1 (acnA), glyceraldehyde-3-phosphate dehydrogenase A (gapA), isocitrate dehydrogenase (icdA), malate dehydrogenase (mdh), glucose-6-phosphate isomerase (pgi), mannitol-1-phosphate 5-dehydrogenase (mtlD) and γ-glutamylphosphate reductase (proA), and the RNA polymerase subunit sigma factor 38 (rpoS) (Waleron et al., 2008) were amplified using degenerate primers. The pgi gene was amplified using primer pair 815/396 for all P. carotovorum and P. atrosepticum, and using primer pair F2/R2 for P. wasabiae.

The PCR reactions were carried out using 50 ng DNA, 1× GeneAmp PCR buffer, 1 mm MgCl2, 200 mm each dNTP (Fermentas), 10 pmol each primer and 1 U Taq DNA polymerase (2·2 × 10−5 errors per nucleotide per cycle) (Fermentas) in a total volume of 25 μL. The PCR amplification consisted of an initial denaturation at 94°C for 4 min followed by 30 cycles at 94°C for 30 s, annealing at 52°C for 30 s, extension at 72°C for 1 min, and a final extension at 72°C for 10 min in a Biometra T-gradient thermocycler. The PCR products were separated on 1% agarose gels in 1× TAE buffer at 100 V for 45 min.

The PCR products were purified from the agarose gels using a Gel Extraction Kit (Seqlab). Sequencing of the samples was performed by Seqlab using the extended Hot-Shot method [BigDye Terminator ready reaction mix v3·1 (ABI)].

Phylogenetic analyses of the housekeeping genes

Multiple alignments were generated for the individual sequences of acnA, gapA, icdA, mdh, pgi, mtlD, proA and rpoS using clustalW in mega version 4·0 (Tamura et al., 2007) with a gap penalty of 15.

Phylogenetic and molecular evolutionary analyses were conducted in mega 4·0. Three different parameters were used to compute maximum parsimony consensus trees, neighbour-joining trees and minimum evolution trees for each of the sequence data sets. Two nucleotide models of maximum composite likelihood (McL) (Tamura & Nei, 1993) and p-distances (Nei & Kumar, 2000) were implemented for the sequences of the seven genes of the 104 taxa, with bootstrapping tests of 5000 replications.

Following that, the gene sequence data sets were analysed to determine the appropriate model of evolution using FindModel ( The best resulting models were as follows: general time reversible plus gamma (GTR + G) (for gamma distribution see Tamura et al. (2004)) for the pgi and proA sequence data sets; equal-frequency Tamura–Nei plus gamma (TrNef + G) for the icdA sequence data set; equal-frequency transition model plus gamma (TIMef + G) for the mdh sequence data set; symmetrical model plus gamma (SYM + G) for the rpoS, gapA and acnA data sets; and the transition model plus gamma (TIM + G) for the mtlD data set.

The parameters for each model (Posada & Crandall, 2001) were implemented in MrBayes 3·1·2 (Huelsenbeck & Ronquist, 2001; Ronquist & Huelsenbeck, 2003) (Table 2) and used to construct maximum likelihood (ML) trees.

Table 2.   Parameters of evolutionary models computed for each sequence data set and implemented in MrBayes
Sequence data setNo. taxaNo. charAIC−lnLTs/TvG shape
  1. No. char: number of characters.

  2. lnL: maximum likelihood value.

  3. f: base frequencies (0·25).

  4. Ts/Tv estimated values of transition/transversion.

  5. G shape: gamma distribution.

Concatenated 7-gene portions104297333510·316538·92·840·46
Concatenated 8-gene portions63375345488·822613·32·680·57

For data of more than 20 000 bp, or concatenated data, the selection of the best-fit model of nucleotide substitution was conducted in mega 5 (Tamura et al., 2007), and the ML trees were constructed in phyml (Guindon & Gascuel, 2003).


AFLP analyses

AFLP reactions using nine primer combinations produced 925 clearly scorable DNA fragments. The phylogenetic tree (Fig. 1) derived from the AFLP analysis delineated 63 strains of Pectobacterium and Dickeya. The Pectobacterium species showed diverse banding profiles and clustered separately from the strains of Dickeya. Seven strains confirmed by PCR as P. atrosepticum grouped into one distinct group. Three P. carotovorum strains isolated from potatoes also separately clustered into one group. These strains were later identified as P. wasabiae by MLSA. Thirty-six of the 40 strains of P. carotovorum formed a large group that subdivided into four subgroups (PcI to PcIV), whereas one additional subgroup (PcV), together with one strain of P. betavasculorum, clustered between P. atrosepticum and P. wasabiae. The Dickeya spp. (12 strains), including two strains from Syria (C89.1 and C89.2), formed one cluster distinct from all other analysed Pectobacterium strains. This cluster was further subdivided into four subgroups.

Figure 1.

 Majority rule consensus tree of 63 strains of Pectobacterium and Dickeya based on 925 AFLP fragments. The evolutionary relationships of the 63 taxa were inferred using the UPGMA method. The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown above the branches.

MLSA analyses of the housekeeping genes

Analysis of all 63 strains indicated that the eight genes were conserved across all strains. The analysis of the eight conserved genes generated 501 (Table S3) characterized sequences. The sequences were submitted to the NCBI GenBank under accession numbers of HM156760 to HM157253 ( An MLSA ML dendrogram was computed from concatenated sequences obtained from eight genes of each of the 63 taxa based on the GTR + G + I model (Fig. S1).

To analyse the samples within a broader range of germplasm, 40 additional sequences were obtained from NCBI GenBank, as well as information from the fully sequenced genome of P. carotovorum (strain P.c.1) and, in the context of this extended data set, seven of the eight genes examined in the first MLSA analysis were reanalysed. The ML tree for this data set (Fig. 2) was constructed based on the substitution model TN93 being applied to the whole concatenated sequence. This extended set of genotypes represents a wider geographic distribution and a wider host range than the original set, and therefore may provide a better resolution of the taxonomic relationships within the subspecies of Pectobacterium. The phylogenetic relatedness found between ML trees constructed from the 63-strain/eight-gene and 104-strain/ seven-gene MLSA analyses indicated a very good correlation (Figs S1 and S2). The overall genetic diversity in both trees was about 9·7%, suggesting that the value is the average genetic diversity in the conserved genes of these pathogens.

Figure 2.

 Maximum likelihood dendrogram generated using seven concatenated housekeeping gene sequences representing the evolutionary relationships among 86 strains of Pectobacterium spp. and 18 strains of Dickeya spp. The evolutionary relationships of the 104 taxa (63 taxa from this study and 41 taxa retrieved as electronic data from NCBI GenBank) were inferred using the maximum likelihood method. The percentages of the replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown above the branches.

Although AFLP markers provide high information content and lead to highly reliable tree topologies, an assignment by AFLP to a particular species is difficult without reference strains being included in the AFLP gels. However, differentiation at the species level can be achieved by using DNA sequences of published reference strains, as exemplified by the three P. carotovorum strains (SCRI 207, JKI 1A-1 and JKI 4.2.6) which were re-identified as P. wasabiae. Therefore, using AFLP corroborates the utility of MLSA as a classification method for poorly characterized bacteria such as Pectobacterium. Another advantage of sequence-based analyses is the possibility of computing a separate dendrogram for each individual gene of interest to analyse gene flow. If the exchange of genetic material between genetically defined groups of taxa occurs, this should be reflected by divergent topologies in the single-gene trees. The ML dendrograms computed separately for the genes acnA, gapA, icdA, mdh, mtlD, pgi, rpoS and proA indicated the rare appearance of single strains within the cluster of a different species (Table S2, Figs S2–S9). This occurrence was observed for two out of the eight analysed genes in strains P. atrosepticum Pba6 and P. carotovorum SCRI121. Therefore, these strains must be regarded as more polymorphic than other strains of the same species.

A gene-dependent clustering was observed for Dickeya sp. strain 436. This strain clustered into the P. carotovorum subsp. odoriferum group if only the icd gene was considered, clustered into the P. atrosepticum group for the rpoS gene, but was correctly placed into the Dickeya group by the MLSA approach. This Dickeya strain displayed similarity to strains of Pectobacterium (Table S2). The similarity may explain the strain’s previous characterization as a Pectobacterium strain, which was changed following PCR-based diagnostics (Nassar et al., 1996). As these single-gene-dependent clusters are supported by individual bootstrap values larger than 50%, they are probably not the result of insufficient data, but more likely illustrate horizontal gene flow and subsequent recombination events occurring between strains from different species. Although different mechanisms are involved in gene transfer, conjugation is considered to be the most widely occurring mechanism for gene flow between Gram-negative bacteria.


The MLSA tree topology of each of the eight and seven concatenated genes matched the dendrogram obtained by AFLP analyses, with the one exception of the PcV cluster. These results are in agreement with other published studies using this combination of methods (Ngoc et al., 2010).

The data generated for the 104 data sets revealed an average molecular diversity of 6% among Pectobacterium strains, which formed groups clustering with the same topology as those published by Ma et al. (2007). Strains formed the following four groups: the P. atrosepticum group (10 strains), the P. wasabiae group (nine strains), the P. betavasculorum group (four strains) and a pooled P. carotovorum group of 63 strains in which all other strains were clustered. Genetic diversities of 0·01% and 1% were observed among the strains of P. betavasculorum and P. atrosepticum, respectively. A high level of homogeneity in P. atrosepticum has also been demonstrated in other studies (Darrasse et al., 1994a; De Boer & Ward, 1995; Persson & Sletten, 1995). Pectobacterium carotovorum, comprising five subclusters (PcI–PcV), is considered to be more closely related to P. atrosepticum than P. betavasculorum or P. wasabiae, with average genetic diversities from them of 7·3, 8 and 8·6%, respectively (Avrova et al., 2002).

The P. wasabiae group (Goto & Matsumoto, 1987), with a genetic diversity of 4·6%, was comprised of one strain isolated from Japanese horseradish and eight strains isolated from potatoes. Pectobacterium wasabiae was recently isolated from potatoes (Ma et al., 2007; Pitman et al., 2009). The presence of the species on potatoes does not reflect a recent extension of the host range for this taxon, because some former P. carotovorum strains isolated from potatoes were subsequently re-identified as P. wasabiae in this study.

The topology of the trees obtained from both AFLP analysis and MLSA revealed that P. carotovorum is a highly diverse plant pathogen and comprises five genetic clusters (PcI to PcV). Previous studies based on either AFLP or MLSA placed P. carotovorum strains in a maximum of three different subclusters (Avrova et al., 2002). More recent studies support the existence of additional groups of genotypes. This group expansion relies on the identification of strains from monocotyledonous plant families based on both genetic and pathogenicity data (Yishay et al., 2008).

Pectobacterium carotovorum strains isolated from Syria, with an average diversity of 3·6%, were distributed over the PcI, PcII and ‘P. c. subsp. brasiliensis’ groupings. The Syrian strains did not form a monophyletic cluster, even when the growing season, the province and the year of isolation was considered. These results are similar to results by Yap et al. (2004), who observed unexpected variability between strains obtained from the same environmental conditions. In addition, no monophyletic cluster was detected for the 45 strains isolated worldwide from potatoes. The potato strains grouped into four of the five P. carotovorum clusters with a genetic diversity of 4·0% within the strains showing potato as a common host.

The average diversity of 4·2% among all 63 strains of P. carotovorum, irrespective of host, compared with 4·0% diversity among P. carotovorum strains isolated only from potato also demonstrated that strains that can infect any other hosts can easily infect potatoes (Toth et al., 2003; Glasner et al., 2008), with the exception of P. c. subsp. odoriferum strains. Pectobacterium carotovorum combines divergent strains clustered separately in five genetic clusters. These clusters, PcI, PcII, PcIII, PcIV and PcV, are situated equally at a genetic distance of 8% from the P. betavasculorum and P. wasabiae groups and at genetic distances of 7, 8, 8, 7 and 8%, respectively, from P. atrosepticum.

The 20 strains within PcI, isolated from different geographical regions and various plant hosts (e.g. potato, marigold, cabbage, burdock and Aloe arborescens), displayed a high genetic diversity of 2·5% between strains and distances of 3·7–6·5% from the four other P. carotovorum clusters.

In PcII (13 strains), all strains were isolated from potato plants with the exception of strain JKI 4.3.8 isolated from Brassica oleracea. However, the strains in PcII were collected from different geographical regions. These strains grouped into one cluster with a genetic diversity of 1·9% between strains and a distance of 3·7–7% from other P. carotovorum clusters.

Clusters PcI and PcII, with a genetic distance of 3·7% from each other, were the most closely related P. carotovorum groups in both the MLSA and AFLP analyses. In the AFLP analyses, different banding profiles were observed for PcI and PcII. Both clusters were considered to be part of one clade by Ma et al. (2007). These two clusters are suggested to be formed by a variety of genetically different strains of the common P. carotovorum subsp. carotovorum. For better classification and deeper characterization of the strains of PcI and PcII, extensive analysis for their virulence variability is necessary.

The four P. c. subsp. odoriferum (Gallois et al., 1992) strains grouped into PcIII and showed low genetic variation (1%) between strains. The distances to the four other Pc clusters were significantly higher, with a range of 4·5–7·3%. These strains form an obvious subspecies based on their genomic differentiation. In a previous study (Ma et al., 2007), where only one P. c. subsp. odoriferum strain (SCRI482) was included, that strain clustered with other P. carotovorum subsp. carotovorum inside clade II, so was not regarded as a different subspecies. Nevertheless, when combined with three other strains from this study, that strain grouped into a separate P. c. subsp. odoriferum cluster. These results strongly confirm the necessity of using more than one strain representing genetically different taxa, especially subspecies, in a phylogenetic analysis.

The cluster PcIV, or ‘P. c. subsp. brasiliensis’ (Duarte et al., 2004), is comprised of eight strains from Brazil, America and Israel (Ma et al., 2007) as well as 10 strains from Peru, Syria, Germany and Japan. Recently, strains isolated in South Africa were identified as ‘P. c. subsp. brasiliensis’ (van der Merwe et al., 2010). The ‘P. c. subsp. brasiliensis’ strains were mainly isolated from potatoes in addition to other hosts, such as peppers, carrots and Ornithogalum. The genetic diversity within PcIV was estimated as 3·4%, and PcIV showed a distance of 4·8–7·1% to the other P. carotovorum clusters.

Two atypical P. carotovorum strains from Syria, C331 and C364.2, grouped into the clusters of either ‘P. c. subsp. brasiliensis’ or P. c. subsp. odoriferum by AFLP analysis and MLSA phylogeny, respectively. These two strains, isolated from potatoes, are able to utilize α-methyl glycoside and reduce sucrose. Both characteristics are considered stable criteria for all strains of P. c. subsp. odoriferum (Gallois et al., 1992) and for some strains of ‘P. c. subsp. brasiliensis’, especially strains identified by Duarte et al. (2004). These strains may indicate a new recombination of soft-rot isolates (Naylor et al., 2002).

PcV, including six strains, is comprised of strains from the monocotyledonous plants Saccharum arundinaceum, Zantedeschia aethiopica and Ornithogalum dubium, delineated into one monophyletic cluster along with strains from the dicotyledonous plants Solanum tuberosum and Persea americana. The monocot strains were grouped with two dicot strains at a genetic diversity of 2·6%, and a distance of 6·8–7·3% from the other P. carotovorum clusters. These distantly grouped strains indicate a distinct cluster previously unreported. This cluster presents a new phylogenetic species including P. carotovorum strain 106 (Ma et al., 2007), which was identified as an orphan taxon. A P. carotovorum taxon comprising strains isolated from monocot plants was also suggested by Yishay et al. (2008) using data from virulence tests, ITS-PCR banding patterns and 16S rRNA sequence analysis.

Strains belonging to the genus Dickeya demonstrated a clearly distinct group by both AFLP and MLSA analyses, with an average genetic distance of 18·0% from all Pectobacterium strain clusters. These results were calculated from an analysis of 104 strains. The difference in the genetic diversity of 10·8% within the genus Dickeya (18 strains) when compared to the diversity within Pectobacterium (86 strains, 6% overall genetic diversity) may stem from sampling bias caused by the high number of Pectobacterium strains included in the data set, including several highly similar strains, thus reducing the average diversity when compared to the 18 Dickeya strains. Moreover, Dickeya was formerly divided into six species by Samson et al. (2005) and a new genetic clade, tentatively called Dickeya solani (Sławiak et al., 2009). This division supports the present results which indicate the genus Dickeya to be the most diverse among the soft-rot pathogens.

An unexpected diversity of 10·3% was detected between strains of the species D. dadantii. As these strains were retrieved from GenBank under this name, it is possible that some of these strains had been misidentified. Their origins need to be confirmed to exclude such a possibility.

The new strains isolated from Syria and investigated in this study belong to five phylogenetic groups, namely: P. c. subsp. carotovorum (clusters PcI and PcII), ‘P. c. subsp. brasiliensis’, (PcIV), P. atrosepticum and Dickeya spp. The 30 strains were tested by both stem and soil inoculation (Dannon & Wydra, 2004; Costa et al., 2006) on the potato variety L390, with seven plants per strain and the experiment repeated at least twice for each strain (data not shown). The results of these tests confirmed that none of the representative Pectobacterium clusters can cause soil disease but they can induce severe stem maceration. The Dickeya spp. strains are the only soil pathogens and cause stem tissue maceration as well. Avoiding the source of the bacterial inoculation using sanitary procedures such as pathogen-free irrigation water and agriculture machinery is considered the main control procedure against stem infection of wounded plants. For soil pathogens and contamination which comes from the seed lots, there is a continuous need to select for more tolerant or resistant cultivars against locally isolated strains in each country (Pérombelon, 2002; Czajkowski et al., 2009).


This work was funded by the Islamic Development Bank. Additionally, we would especially like to thank Professor Klaus Geider (JKI, Germany), Professor Ian Toth (SCRI, UK) and Professor Edgar Maiss (IPP, Germany) for their advice and help. The majority of the reference strains were kindly provided by Professor Ian Toth, Professor K. Geider (Dossenheim) and Dr Klaus Richter (Quedlinburg, JKI).