Introduction

Pseudomonas syringae is a heterogeneous phytopathogenic bacterial species divided into >50 pathovars (Young 1991). This pathogen is commonly encountered in fruit orchards in Wallonia in southern Belgium. P. syringae pv. syringae (Pss) has been reported in pear orchards in various regions of the world; Pss and P. syringae pv. morsprunorum (Psm) race 1 have been reported in plum orchards, and Pss, two races of Psm and P. syringae pv. avii have been reported in cherry orchards or plantations (Freigoun and Crosse 1975; Yessad et al. 1992; Ménard et al. 2003; Vicente et al. 2004; Kennelly et al. 2007; Renick et al. 2008). Among these pathovars, P. syringae pv. avii is genetically homogeneous (Ménard et al. 2003) and Psm comprises two highly homogeneous races that are clearly distinct from each other (Ménard et al. 2003; Vicente et al. 2004; Vicente and Roberts 2007). In contrast, Pss is genetically highly heterogeneous (Weingart and Völksch 1997; Little et al. 1998) and a range in aggressiveness to the reference plant, lilac, was reported among strains of this pathovar (Vicente et al. 2004).

Despite the diversity existing within P. syringae, all the fluorescent strains of P. syringae and of the closely related species Pseudomonas viridiflava, Pseudomonas ficuserectae, Pseudomonas savastanoi and Pseudomonas cannabina produce the atypical pyoverdin siderophore PaA. This atypical pyoverdin has characteristics that enable distinction to be rapidly made in identification between these species and all other fluorescent Pseudomonas species (Bultreys et al. 2001, 2003). Strains of Psm race1 and Pss can produce phytotoxins that are recognised as important components of virulence (Bender et al. 1999) and the detection of these phytotoxins can be used in identification (Young et al. 1992). Most of the Psm race 1 strains produce the phytotoxin coronatine and the detection of the cfl gene involved in coronatine production can be used in Psm race 1 identification (Bereswill et al. 1994). Also, most Pss strains produce the toxic lipodepsipeptides (TLP) syringomycins and syringopeptins, and genetical tests based on this characteristic can be used for the diagnosis of this pathovar (Quigley et al. 1994; Bultreys and Gheysen 1999). In addition, many different methods of pathovar identification have been reported, but performing pathogenicity tests is necessary for differentiating pathogenic and non-pathogenic strains (Burkowicz and Rudolph 1994; Vicente et al. 2004). Many pathogenicity tests on different organs and hosts have been described for identifying virulent P. syringae strains at the pathovar level. With regard to Pss, a standardised procedure on lilac leaves has been described (Young 1991), but lilac tests have not always been decisive in confirming Pss identification (Scortichini et al. 2003; Vicente et al. 2004).

Among the various genetical methods used for characterising P. syringae strains, the Repetitive Extragenic Palindromic (REP)-PCR, Enterobacterial Repetitive Intergenic Consensus (ERIC)-PCR, and BOX-PCR (collectively known as rep-PCR) were shown to be highly discriminating methods adapted for characterising clonal isolates. Rep-PCR and Insertion Sequences 50 (IS50)-PCR were used to classify bacterial strains between and within P. syringae pathovars (Louws et al. 1994; Weingart and Völksch 1997). REP-PCR showed that P. syringae pv. avii, Psm race 1 and Psm race 2 constitute tight genetic groups distinct from each other and from other P. syringae pathovars (Ménard et al. 2003; Vicente and Roberts 2007). In California, ERIC-PCR analyses showed that the Pss strains isolated from Prunus generated similar genetic profiles, whereas most Pss strains isolated from other hosts generated different and variable patterns. This led the authors to suggest a host specialisation of the stone fruit strains within Pss (Little et al. 1998).

The purpose of the present study was to identify and characterise a large number of P. syringae isolates isolated from diseased lesions in fruit orchards in Wallonia in order to determine the diversity within this species in Belgium. Within Pss, the combination of microbiological tests and rep-PCR and IS50-PCR results enabled a high genetic heterogeneity to be visualised, as well as the existence in this pathovar of isolate–host relationships. The information provided by a pathogenicity test on lilac leaves was also assessed. The genetic methods proved very informative in helping to identify isolates of P. syringae pv. morsprunorum race 1 and race 2 as well as unclassified P. syringae isolates that were also encountered in Belgian fruit orchards.

Materials and methods

Bacterial strains and identifications

The 400 P. syringae and P. viridiflava strains and isolates used in this study are listed in Tables 1 and 2. The 44 strains listed in Table 1 are strains from various origins obtained from culture collections. The 356 Belgian Pseudomonas isolates listed in Table 2 were isolated from diseased plants from necrotic lesions on fruits, leaves, stems, flowers or buds in 36 Belgian orchards between 1993 and 2002. The studied isolates originated mainly from pear orchards (226 isolates), but also from sweet cherry (84 isolates), sour cherry (21 isolates), and plum (24 isolates) orchards. Except for one Flemish orchard, all the orchards were in Wallonia in southern Belgium. The preliminary identifications of about 170 of the Belgian isolates have been previously reported (Bultreys and Gheysen 2003).

Table 1 Bacterial reference strains used in the study
Full size table
Table 2 Characteristics and identification of the Belgian isolates from fruit trees
Full size table

The isolates from Belgian orchards were classified by different methods over the years depending on the availability of identification tests, and based on the fact that these isolates had been isolated from cherry, plum or pear. The tests that were relevant at least at one period for classifying an isolate in P. syringae and related species were the fluorescence on King's B medium, the hypersensitive reaction (HR) in tobacco leaf (Klement et al. 1963), the oxidase activity, the Api 20 NE tests, which include the arginine dihydrolase test, and the pyoverdin siderophore visual and HPLC tests (Bultreys et al. 2001, 2003). The tests that were relevant at least at one period for classifying a strain in a P. syringae pathovar or in P. viridiflava are described below. The GATTa tests (G, gelatin liquefaction; A, β-glucosidase activity; T, tyrosinase activity; and Ta, tartrate use) combined with the lactate use test (Garrett et al. 1966; Latorre and Jones 1979) enabled identification of Pss and Psm race 1: + + − − + responses were expected for Pss, and − − + + − responses for Psm race 1. A positive response in the potato rot test enabled distinction of P. viridiflava. The detection of TLP production by a biological test and the PCR detection of the syrD gene involved in TLP secretion enabled identification of Pss (Bultreys and Gheysen 1999), whereas the PCR detection of the cfl gene involved in coronatine production enabled identification of Psm race 1 (Bereswill et al. 1994). Also, findings from this study led to the use of rep-PCR and particularly BOX-PCR (see below for protocols) to identify Psm race 2, by comparisons with the English reference strain CFBP 3800, and P. syringae pv. avii, by comparison with the French reference strains CFBP 3846 and CFBP 3848. Also, BOX-PCR was used to confirm identifications of Psm race 1 isolates determined previously using other methods.

IS50- and rep-PCR analyses

The BOX primer BOXA1R (5′-CTACGGCAAGGCGACCTGACG-3′) was used with the BOX-PCR conditions described by Louws et al. (1994). The ERIC primers ERIC-1R (5′-ATGTAAGCTCCTGGGGATTCAC-3′) and ERIC-2 (5′-AAGTAAGT GACTGGGGTGAGCG -3′), the REP primers REP1R-I (5′-IIIICGICGICATCIGGC-3′) and REP2-I (5′-ICGICTTATCIGGCCTAC-3′), and the IS50 primer (5′-CAGGACGCTACTTGTGT-3′) were used under the ERIC-PCR, REP-PCR or IS50-PCR conditions, respectively, described by Weingart and Völksch (1997). The primers were obtained from Eurogentec. Bacterial strains and isolates were streaked onto plates of nutrient agar and incubated at 28°C overnight. Approximately 106 cells mixed in 1% (v/v) Tween 20 were transferred to 25 μl of PCR mixture. The PCR mixtures contained: 50 pmol of each primer; 67 mM Tris–HCl (pH 8.8); 16 mM (NH4)2SO4; 3.5 mM MgCl2; 10 mM 2-mercaptoethanol; 400 μM of each deoxynucleoside triphosphate; 4 μg of bovine serum albumin; 5% dimethyl sulfoxide; and 0.75 units (REP-, ERIC-, and BOX-PCR) or 1.5 unit (IS50-PCR) Taq DNA polymerase (GE Healthcare). PCR was performed in a Thermal I cycler (Biorad®). The ERIC-PCR programme comprised: one cycle at 95°C for 5 min; 30 cycles at 94°C for 1 min, 52°C for 1 min, and 65°C for 8 min; and finally one cycle at 65°C for 15 min. The REP-PCR programme comprised: one cycle at 95°C for 5 min; 30 cycles at 94°C for 1 min, 40°C for 1 min, and 65°C for 8 min; and finally one cycle at 65°C for 15 min. The BOX-PCR programme comprised: one cycle at 95°C for 7 min; 30 cycles at 94°C for 1 min, 53°C for 1 min, and 65°C for 8 min; and finally one cycle at 65°C for 15 min. The IS50-PCR programme comprised: one cycle at 93°C for 3 min; 35 cycles at 94°C for 1 min, 38°C for 1 min, and 72°C for 3.5 min; and finally one cycle at 72°C for 15 min. After PCR, 18 μl of the amplification products were separated on 1% agarose gels at 3.4 V/cm for 3 h. DNA fragments were visualised under UV illumination after staining with ethidium bromide. The analysis of every strain and isolate was repeated twice. Each amplification band was treated as a unit character and was scored as present (1) or absent (0). In the TREECON software (Van de Peer and De Wachter 1994), a distance matrix was constructed using the Link coefficient, and the Unweighted Pair-Group Method (UPGMA) with arithmetic means was used for clustering. The strength of the tree topology was assessed using the bootstrap method.

Pathogenicity tests on lilac

The pathogenicity on lilac leaves was evaluated for two Belgian isolates of P. viridiflava from sour cherry and for 108 Belgian isolates of P. syringae from cherry, pear or plum: 57 TLP+ Pss, 11 TLP− Pss, 15 unclassified P. syringae, 15 Psm race 1, and ten Psm race 2. In addition, the reference strains Pss B301D and LMG5141 from pear, Pss CFBP 2118 and LMG 5494 from cherry, Pss LMG 6104 from plum and the pathotype strain Pss LMG 1247 from lilac were also tested.

The protocol of the pathogenicity test on lilac was adapted from Young (1991) and Yessad-Carreau et al. (1994). Five year-old lilac plants (Syringa vulgaris cv. And an Ludwig sp.) were grown in the greenhouse with 16 h artificial light and the temperature maintained between 20°C and 26°C. Five fully expanded leaves were cut from shoots with eight new leaves. The leaves were dipped into a solution of sodium hypochlorite (1% active hypochlorite) for 5 min, rinsed three times in sterile osmosed water and the excess of water removed with paper. A 5 mm wound approximately 5 mm away from the lamina was made on the petiole with a scalpel. A 10 μl drop of bacterial suspension (108 CFU ml−1) was then deposited on the wound. The inoculated leaves were placed on sterile paper filter over water agar (10 g agar l−1) in sterile Petri dishes. The Petri dishes were sealed with a piece of Parafilm and incubated at 20°C for 7 days under daylight conditions. Five leaves were inoculated for each isolate and a mean progression was calculated based on the results of the five repetitions. The strains and isolates were placed in six pathogenicity classes according to their aggressiveness: class 0: no necrosis; class 1: mean necrosis limited to the cut; class 2: mean necrosis from 10 to 20 mm; class 3: mean necrosis from 21 to 30 mm; class 4: mean necrosis from 31 to 40 mm; and class 5: mean necrosis from 41 to 50 mm. Following the definition of pathogenicity to lilac leaves described by Young (1991), the strains and isolates were considered pathogenic if a progressive lesion was observed (class 2 to 5). To test the reproducibility of the results, one complete repetition with two or three representatives of each class was conducted in another period using, again, 5 leaves for each isolate.

Results

Identification

Table 2 summarises all the results of the identification of the Belgian isolates. The 1993 fluorescent isolates were classified in the phytopathogenic fluorescent Pseudomonas based on their fluorescence on King's B medium and on the negative responses in the oxidase test and in the arginine dihydrolase test. Comparison of the Api 20 NE results and GATTa and lactate use tests results between control Psm r1 strains and fluorescent and non-fluorescent Belgian isolates enabled identification of Belgian Psm r1. The presence of the cfl gene was frequent among Belgian Psm r1 from sweet cherry and the four fluorescent and non-fluorescent isolates tested induced HR on tobacco. The 1996 pear Pss isolate was also identified by Api 20 NE and GATTa and lactate use test comparisons with control Pss strains. It was one of the first Pss strains used to evaluate the reliability of using pyoverdin tests to directly affiliate an isolate to P. syringae and related species, and TLP tests to identify Pss. Pathogenicity on cherry twig of several Psm race 1 isolates isolated in 1993 and pathogenicity on cherry twig and pear leaf of the Pss isolate isolated in 1996 was confirmed (Bultreys and Gheysen 1999, unpublished results).

The identifications of the 1999 isolates were based more on siderophore and phytotoxin-based tests (pyoverdin, TLP and coronatine). The isolates were classified according to their pyoverdin siderophore type or their HR reaction on tobacco, depending on their fluorescence in GASN solid/liquid medium (Bultreys and Gheysen 2000). The oxidase activity, the production of TLP and the PCR detection of the syrD and cfl genes were then determined. The isolates producing the atypical pyoverdin PaA and TLP (TLP+) were classified as TLP+ Pss. Both the fluorescent isolates producing PaA and the non-fluorescent isolates inducing HR on tobacco that possessed the cfl gene (cfl+) were classified as cfl+ Psm race 1. The isolates producing the pyoverdine PaA and inducing potato rotting were identified as P. viridiflava. The other isolates producing the pyoverdine PaA were analysed by classical procedures including Api 20 NE and GATTa + lactate use tests. Some isolates giving + + − − + responses were classified as TLP− Pss, except if they possessed the irp1 gene involved in yersiniabactin production (data not shown) because this characteristic is related to P. syringae strains belonging to the genospecies 2 (pv. phaseolicola and glycinea), 3, 7 and 8, but not to the genospecies 1 including Pss (Bultreys et al. 2006).

The BOX-PCR data provided additional information in identification of Psm race 1, Psm race 2 and P. syringae pv. avii strains and isolates because these strains showed constant specific BOX patterns (Fig. 1). Among the unclassified Belgian isolates, 12 out of the 14 isolates from sour cherry clustered together and belonged to the same BOX-P, ERIC-P, REP-P and IS-P as the British reference strain Psm race 2 CFBP 3800 (Fig. 1). These isolates were therefore included in this pathovar and race (Table 2). Also, the two other remaining unclassified P. syringae isolates from sour cherry had the same BOX-P, but slightly different ERIC-Ps, REP-Ps and IS-Ps. Interestingly, the pathotype strain Psm LMG 5075 (type 2) grouped perfectly with the Psm race 2 strain and isolates using the BOX-, ERIC-, REP- and IS50-PCR methods and was therefore considered as a Psm race 2 strain (Table 1). The same observation was made with BOX-PCR, using different procedures, for the equivalent strain CFBP 2351 from the CFBP culture collection (Bultreys and Gheysen, unpublished results). Also, BOX-PCR grouped all the cfl+ and cfl− Psm race 1 strains and isolates tested in this study in only one BOX-P (Fig. 1), which enabled the confirmation of all the previous identifications of Belgian Psm r1 isolates based on other characteristics (Table 2).

Fig. 1
figure 1

BOX-PCR patterns of Psm race 1, Psm race 2, and P. syringae pv. avii isolates and strains. Lane 1 and 13: DNA molecular weight marker XVI (Roche). Lane 2 to 5: Psm race 1 strains and isolates LMG2222 (cfl+), PmC36 (cfl+), LMG 5463 (cfl−) and PmC24 (cfl−). Lane 6 to 9: Psm race 2 strain and isolates CFBP 3800, Pm2C80, Pm2C101 and Pm2C110. Lane 10 and 11: P. syringae pv. avii strains CFBP 3846 and CFBP 3848. Lane 12: water control

Full size image

Finally, among the 356 Belgian isolates analysed in this study, 14 pear isolates were identified as TLP− Pss and 266 isolates were identified as TLP+ Pss, of which 202 had been isolated from pear, 44 from sweet cherry, 16 from plum and four from sour cherry. Thirty-nine isolates from sweet cherry were identified as cfl+ Psm race 1, and only two isolates as cfl− Psm race 1. Twelve isolates from sour cherry were identified as Psm race 2 and the two resting P. syringae isolates from sour cherry were genetically very close to this pathovar and race. The potato test identified three P. viridiflava isolates from sour cherry. Ten isolates from pear and eight from plum remained unclassified in P. syringae.

ERIC-, REP-, BOX- and IS50-PCR analyses

Among the methods investigated to evaluate variability within the 400 reference and Belgian P. syringae strains and isolates analysed, REP-PCR was the most discriminating method and generated 63 patterns. BOX-PCR was the least discriminating method and generated 43 patterns. ERIC- and IS50-PCR were intermediate with 59 and 58 patterns, respectively. The fragment lengths ranged from 150 to 6,000 bp. The highest diversity was observed among the Pss isolates (TLP+ and TLP−), with between 36 and 51 patterns, depending on the method. The Psm race 1 strains and isolates (cfl+ and cfl−) were genetically more homogeneous. The cfl+ Psm race 1 strains and isolates were grouped in only one ERIC, REP, BOX and IS50 pattern, independent of country of origin. The cfl− Psm race 1 strains and isolates generated one BOX pattern (BOX-P), two related REP patterns (REP-Ps), three related IS50 patterns (IS-Ps) and five related ERIC patterns (ERIC-Ps). For each method, these patterns differed from each other by only one or two bands. They also differed by only one or two bands from the corresponding pattern of the cfl+ Psm race 1 isolates. BOX-PCR was the least discriminating method: it grouped all the cfl+ and cfl− Psm race 1 strains and isolates tested in only one BOX-P (Fig. 1). The two strains of P. syringae pv. avii investigated gave identical results using ERIC-, REP-, BOX- and IS50-PCR; they were clearly different from all the other pathovars tested (Fig. 1).

Among the TLP+ Pss, the plum, pear and cherry strains and isolates clustered differently from the reference strains isolated from apricot, peach, cereals, lilac, lemon and orange. A high diversity was observed among the Belgian and reference TLP+ Pss strains and isolates isolated from pear, plum and cherry: 41 patterns were generated by REP-PCR, 26 by BOX-PCR, 35 by ERIC-PCR and 38 by IS50-PCR. Some patterns were frequently observed, whereas 8–13% of the patterns, depending on the method, included only one isolate. In the frequently observed patterns, the genetic groupings differed depending on the method (Table 3). For example, the ERIC-P number 5 (ERIC-P5) included 13 isolates isolated from plum and 32 isolates isolated from cherry, but these plum and cherry isolates belonged to different patterns when analysed using REP-PCR (REP-P10 and REP-P5) or IS50-PCR (IS50-P9 and IS50-P3). With BOX-PCR, they were grouped mainly in BOX-P1 with isolates from pear (Table 3). The TLP− Pss isolates generally clustered differently from TLP+ isolates, except with BOX-PCR, the least discriminating method, where 86% of TLP− Pss isolates clustered in BOX-P6, with some rare TLP+ isolates from pear (Table 3).

Table 3 Numerical distribution in rep- and IS50-patterns of Pss isolates from fruit trees
Full size table

Combined ERIC-, REP-, BOX- and IS50-PCR analyses

As the genetic groupings differed depending on the method used, a combination of the rep- and IS50-PCR results was produced. The results are presented in Figs. 2 and 3 and the resulting combined patterns were named repIS-Ps. The information given by rep- and IS-PCR was very fine because very closely related strains could be differentiated. But, distant strains did have little in common in these analyses. As a result, the dendrograms in Figs 2 and 3 give little useful relative classification information for distant strains and the bootstrap values are particularly low in the left part of these dendrograms.

Fig. 2
figure 2

Dendrogram of the genetic relatedness of the repIS-Ps of the Pss strains and isolates and distribution of the unclassified P. syringae isolates. For each line, the information is: the repIS-P number, the number of strains and isolates in the corresponding pattern (under brackets), the host, the capacity to produce TLP and the pathogenicity on lilac leaves (L) expressed as the number of strains and isolates pathogenic on lilac leaves related to the number of strains and isolates tested in the pattern. The scale indicates the degree of genetic dissimilarity between strains and isolates. The coloured patterns refer to Belgian Pss isolates: TLP+ isolates from pear (blue), TLP− isolates from pear (light blue), TLP+ isolates from cherry (red), TLP+ isolates from plum (green). The Belgian P. viridiflava isolates are coloured in purple and the Pss reference strains in brown. In repIS-P2, and -P3 the number under brackets is in brown because these patterns include one Pss reference strain. The other patterns are either unclassified P. syringae isolates or strains from specified pathovars or species. The numbers on the branches are the bootstrap values

Full size image
Fig. 3
figure 3

Dendrogram of the genetic relatedness of the repIS-Ps of the Psm race 1, Psm race 2, P. syringae pv. avii and the unclassified P. syringae isolates and strains from plum or cherry. For each line, the information is: the repIS-P number, the number of strains and isolates in the corresponding pattern (under brackets), the capacity to produce coronatine, the identification and the observed pathogenicity on lilac leaves (L) expressed as the number of strains and isolates pathogenic on lilac leaves related to the number of the strains and isolates tested in the pattern. The scale indicates the degree of genetic dissimilarity between strains and isolates. The coloured patterns include strains and isolates isolated from sweet cherry and Prunus (blue), from plum (green), from sweet cherry (red), from sour cherry (dark red) and from wild cherry (black). The numbers on the branches are the bootstrap values

Full size image

Figure 2 compares reference TLP+ strains and Belgian isolates belonging to Pss, P. syringae pv. aptata and P. syringae pv. atrofaciens, the Belgian TLP− Pss isolates, the Belgian unclassified P. syringae isolates, and the Belgian isolates and reference strains of P. viridiflava. It shows that there was high variability among the strains and isolates analysed. Two relatively well supported clusters were specific to the tested TLP+ strains of the pvs atrofaciens and aptata, respectively. Some weak genetic heterogeneity was observed among the TLP+ strains within the pvs atrofaciens and aptata, but there was no overlapping between these pathovars and the TLP+ Pss strains and isolates. The percentage of similitude between strains of pvs aptata and atrofaciens and the closest Pss strains and isolates was <30%. A great diversity was observed among the Pss strains and isolates. The high number of related patterns probably explained the low bootstrap values observed in the tree, which often excluded definition of well supported clusters. Several groups based on a genetic similarity of >50% were however apparent, as was a tendency of the isolates from the same host to group together. The Belgian isolates from fruit orchards were genetically not similar (<40% similarity) to the reference Pss strains isolated from other crops (apricot, peach, wheat, corn, millet, lilac, lemon, orange) shown in brown in Fig. 2. The three Belgian isolates of P. viridiflava from sour cherry classified in repIS-P20 were also distant from the P. viridiflava reference strains isolated from bean, pear and chicory classified in repIS-P38, 39 and 40 (Fig. 2; Table 1).

The TLP+ Pss from pear, plum and cherry were genetically highly variable and formed 59 repIS-Ps (Fig. 2). However, some patterns were clearly more frequent than others: the pear TLP+ Pss isolates and strains were classified in 34 repIS-Ps, but 69% belonged to repIS-P1 or repIS-P2; the cherry TLP+ Pss isolates and strains were classified in 15 repIS-Ps, but 63% belonged to repIS-P3 or repIS-P6; and the plum TLP+ Pss isolates and strains were classified in ten repIS-Ps, but 52% belonged to repIS-P5. Isolate–host relationships were apparent between the TLP+ Pss isolates and their hosts of isolation (Fig. 2): no Pss isolates from one host (cherry, plum or pear) had exactly the same profile as Pss isolates from another host. For example, none of the 95 isolates belonging to repIS-P1, or the 46 isolates belonging to repIS-P2, was isolated from a host other than pear. The repIS-P3 and repIS-P6 contained only 24 and eight isolates from cherry, respectively. The repIS-P5 contained only ten isolates from plum. The repIS-Ps including the Belgian isolates were not specific to Belgium. Indeed, although some investigated foreign strains had specific patterns (Fig. 2 and Table 4), the two pear strains isolated in 1959 in UK B301D and LMG 5141 belonged to the pear-specific repIS-P2, including many Belgian pear isolates, and the closely related repIS-P41, respectively. Also, the CFBP 2117 and CFBP 2118 strains isolated in 1979 from cherry in France belonged to the cherry-specific repIS-P3, including many Belgian cherry isolates, and the closely related repIS-P63, respectively.

Table 4 Numerical distribution in repIS-patterns of reference Pss strains from fruit trees
Full size table

The TLP− Pss isolates clustered differently from the TLP+ Pss isolates (Fig. 2). Out of 14 TLP− Pss isolates from pear, 12 clustered in repIS-P4. The two other isolates were genetically different and clustered together in repIS-P19.

Figure 3 compares the cfl+ and cfl− Psm race 1 strains and isolates, the Psm race 2 strains and isolates, the Belgian unclassified P. syringae isolates isolated from Prunus sp., and the P. syringae pv. avii strains. This dendrogram had better bootstrap values and it showed that three clusters were formed. Cluster 1 included the five unclassified P. syringae isolates from plum, as well as the two P. syringae pv. avii reference strains CFBP 3846 and CFBP 3848. However, the two pv. avii strains from wild cherry had completely identical patterns, different from those of the Belgian isolates from plum. Cluster 2 grouped repIS-P87 and repIS-P88. RepIS-P87 contained the British Psm race 2 CFBP 3880 strain, 12 Belgian Psm race 2 isolates and the Psm race 2 pathotype strain LMG 5075 (Fig. 3). RepIS-P88 contained the two previously evoked unclassified P. syringae isolates from sour cherry. Cluster 3 grouped the cfl+ and cfl− Psm race 1 strains and isolates. It contained six closely related repIS-Ps: repIS-P86, including 43 cfl+ Psm race 1 strains and isolates, and five other repIS-Ps, including all the cfl− Psm race 1 strains and isolates tested (Fig. 3). Other rare unclassified P. syringae isolates from Prunus sp. were distant from Pss, Psm race 1, Psm race 2 and P. syringae pv. avii (Figs. 2 and 3).

Pathogenicity tests on lilac leaves

The observed lengths of necroses on lilac leaves ranged from 0 to 46 mm. Examples of symptoms are shown in Fig. 4. A general presentation of the results is shown in Table 5 and the pathogenicity results related to the repIS-P are also shown in Figs. 2 and 3. In the observations, weakly pathogenic isolates were sometimes non-pathogenic on some leaves. However, as limited necrosis had been observed in at least one of the leaves, they were placed in class 1. The test was repeated in time with representative isolates from all pathogenicity classes, but non-pathogenic isolates were always non-pathogenic. With pathogenic isolates, the symptom severity sometimes varied from one group of tests to the other. However, the range of variation was always restricted to one pathogenicity class. For estimating variation, the isolates were classified according to the mean of the lengths of the necroses in both tests (ten inoculated leaves).

Fig. 4
figure 4

Example of symptoms on detached lilac leaves after inoculation with P. syringae isolates. a non-pathogenic isolate; b virulent isolate; c highly virulent isolate

Full size image
Table 5 Pathogenicity to lilac of P. syringae and P. viridiflava isolates and strains from fruit trees
Full size table

Among the 62 TLP+ Pss strain and isolates tested, only 40 (65%) induced progressive necroses and were pathogenic to lilac leaves. But the repartition of the pathogenic isolates to lilac was not homogeneous among the different repIS-Ps. Indeed, 20 of the 22 isolates tested (90.9%) belonging to the main patterns repIS-P1, repIS-P2, repIS-P3, repIS-P5, repIS-P6 and repIS-P7 shown in Fig. 2 were pathogenic to lilac. Surprisingly, however, there could be heterogeneity within repIS-Ps in relation to their pathogenicity on lilac. For example, one of the five isolates tested belonging to repIS-P1 was non-pathogenic to lilac, although it could not be differentiated genetically from other isolates in the same repIS-P. Many isolates belonging to the little-represented repIS-Ps were non-pathogenic to lilac. The repartition of the TLP+ Pss pathogenic to lilac leaves was analysed by the less discriminating BOX-PCR method: 100% of the pathogenic TLP+ Pss isolates were grouped in the three main patterns BOX-P1, BOX-P2 and,BOX-P3, as well as in BOX-P30, with aggressiveness ranging from class 2 to class 5 (Table 5; Fig. 5). It was interesting to note that 88.6%, 85.7% and 93.7% of the Belgian TLP+ Pss isolates from pear, cherry and plum, respectively, belonged to one of these four BOX patterns. Surprisingly, however, within the three main BOX-Ps isolates belonging to repIS-P11, 32 and 51 induced no necroses on lilac leaves (Fig. 6). Figure 6 showed that two of the three main BOX-Ps (BOX-P1 and BOX-P2) were genetically close to each other, with 15% differentiation (Fig. 6). However, BOX-P3 and BOX-P30 were genetically very different. The Pss pathotype strain LMG 1247 induced no necroses on lilac leaves under the test conditions.

Fig. 5
figure 5

Class repartition of TLP+ Pss isolates according to their aggressiveness to lilac leaves. a TLP+ Pss isolates belonging to BOX-P1, BOX-P2, BOX-P3 and BOX-P30. b TLP+ Pss isolates belonging to the other BOX-Ps. Class 0: non-pathogenic isolates; class 1: necrosis limited to the cut; class 2: necrosis from 10 to 20 mm; class 3: necrosis from 21 to 30 mm; class 4: necrosis from 31 to 40 mm; and class 5: necrosis from 41 to 50 mm

Full size image
Fig. 6
figure 6

Dendrogram of the genetic relatedness of the repIS-Ps between TLP+ and TLP− Pss isolates and strains. The genetic analysis method used was BOX-PCR. For each line, the following information is given: the repIS-P; the number of isolates and strains in the corresponding pattern (under brackets), the host and the observed pathogenicity to lilac leaves (L): presented as the number of isolates and strains pathogenic to lilac leaves related to the number of isolates and strains tested in the pattern. The colour codes are: red for TLP+ patterns pathogenic to lilac leaves; green for TLP+ pattern non-pathogenic; pink for TLP− pattern pathogenic to lilac leaves; blue for TLP− pattern non-pathogenic; and brown for TLP+ Pss reference strains. In repIS-P2, and -P3 the number under brackets is in brown because the patterns include one reference strain. The numbers on the branches are the bootstrap values. The scale at the top indicates the degree of genetic dissimilarity between isolates and strains

Full size image

With regard to the TLP− Pss isolates, three of the 11 tested were pathogenic to lilac leaves Table 5. These three isolates belonged to repIS-P4, with aggressiveness ranging from class 2 to class 3. None of the 15 cfl+ or cfl− Psm race 1 isolates tested induced necroses on lilac leaves (Table 5). However, among the ten Psm race 2 isolates tested belonging to repIS-P87 (Fig. 3), two were pathogenic to lilac leaves (Table 5) and were grouped in the pathogenicity class 2.

Two of the eight unclassified P. syringae isolates from pear tested were pathogenic to lilac leaves (Table 5). They were very different genetically from the TLP+ or the TLP− Pss isolates, but were highly aggressive (class 4). The tested unclassified P. syringae isolates from sour cherry (two isolates) and plum (five isolates) were non-pathogenic to lilac leaves. Of the two P. viridiflava isolates tested belonging to repIS-P20, one was pathogenic to lilac leaves and was classified in the pathogenicity class 2 (Table 5).

Discussion

With regard to the pathogen occurrence in Belgium, the detection of Pss isolates in pear, sweet cherry, plum and sour cherry orchards in the country agreed with the findings of other studies (Yessad et al. 1992; Burkowicz and Rudolph 1994; Vicente et al. 2004).

In Belgium, the Psm race 1 isolates have been found in sweet cherry and sour cherry (Bultreys and Gheysen 2003, 2004; Bultreys et al. 2007). In other countries, Psm race 1 has also been reported in plum orchards (Garrett et al. 1966; Burkowicz and Rudolph 1994; Vicente et al. 2004). Most but not all the Belgian Psm race 1 isolates possessed the cfl gene involved in coronatine production.

Psm race 2 was found only occasionally in Belgium in sweet cherry (Bultreys and Gheysen 2003, 2004; Bultreys et al. 2007). These results contrast with the UK, where many strains were isolated from cv. Roundel and to a lesser extend from cv. Napoleon (Freigoun and Crosse 1975; Vicente et al. 2004). In contrast, 12 Belgian sour cherry isolates of the 21 tested were identified as Psm race 2. Psm race 2 was described on sweet cherry (Prunus avium; Freigoun and Crosse 1975), and it was also detected on wild cherry (Prunus avium; Vicente et al. 2004). In this study and previous reports (Bultreys and Gheysen 2004; Bultreys et al. 2007), it appears as a frequent pathogen of sour cherry (Prunus cerasus). In another study, Psm was reported as the cause of bacterial canker of sour cherry in Michigan, USA, but the study and the identification methods (GATTa tests) were orientated towards Psm race 1 only (Latorre and Jones 1979).

No P. syringae pv. avii isolate was detected in Belgium in this study, but wild cherry plantations were not investigated. However, P. syringae isolates other than Pss, Psm race 1, Psm race 2 and P. syringae pv. avii were found in pear, plum and sour cherry orchards in Belgium. The study of the ability of these unclassified P. syringae isolates to induce disease is being undertaken.

Pseudomonas syringae pv. syringae

Many isolates were named Pss in the present study based on their presence in diseased tissues in pear, cherry and plum orchards, on the possession of the atypic pyoverdin Pa A and on the production of TLPs. However, TLPs are also produced by the pvs aptata and atrofaciens in P. syringae (Bultreys and Gheysen 1999). Both pathovars were shown to be pathogenic to fruit trees in laboratory assays (Quigley et al. 1994). However, there should be no confusion with these two pathovars because the TLP+ P. syringae isolates isolated from Belgian fruit orchards had different rep-Ps and IS50-Ps than the tested P. syringae pv. aptata and P. syringae pv. atrofaciens control strains. In addition, the respective strains of pvs aptata and atrofaciens grouped together in clearly distinct genetic clusters. Whereas the detection of Pss by TLP-related tests proved to be very powerful, 5% of the isolates finally classified in Pss in this study did not produce TLP. These isolates were identified using the GATTa and lactic acid use tests. It has already been shown by Vicente and Roberts (2007) that some Psm race 2 strains could give the same GATTa tests results (+ + − −) as Pss strains. In addition, the results from the GATTa tests for the other pathovars of P. syringae (>50 known pathovars) are largely unknown and therefore it is possible that other P. syringae strains could give the same GATTa results as Pss strains. All the Belgian TLP− Pss isolates were isolated from pear. Most clustered together in a BOX-P comprising poorly represented TLP+ isolates. Interestingly, some TLP− isolates in this BOX-P were pathogenic to lilac, but not the TLP+ isolates. Scortichini et al. (2003) have already shown that some TLP+ and TLP− Pss strains could share the same BOX-P. In the present study, however, the combined repIS-Ps differentiated all the TLP+ Pss isolates from the TLP− Pss isolates. In addition, the two repIS-Ps containing the TLP− Pss isolates were genetically distant from the main repIS-Ps formed by the TLP+ Pss isolates.

The analysis of the repIS-Ps indicated a high genetic heterogeneity among the TLP+ Pss isolates from cherry, pear and plum in Belgium. Such heterogeneities had also been observed in other countries (Little et al. 1998; Weingart and Völksch 1997). This analysis, however, showed that specific repIS-Ps contained numerous genetically similar isolates from the same host (repIS-P1 and repIS-P2 from pear; repIS-P3 and repIS-P6 from cherry; repIS-P5 from plum). The pathogenicity of the isolates belonging to these repIS-Ps is clear since 17 out of the 18 strains tested were pathogenic to lilac. A global pathogenicity study on multiple hosts is in process and these isolates are also pathogenic to their respective isolation hosts (Gilbert and Bultreys, unpublished results). Also, the isolates belonging to repIS-P3 and repIS-P2 were genetically identical or very similar to reference strains from France (CFBP 2117 and CFBP 2118) and UK (B301D; Table 4) received in 1997 from J. P. Prunier and L. Gardan and from D. C. Gross as highly virulent strains on cherry and pear, respectively. In contrast, other repIS-Ps contained only one isolate. The existence of both isolates belonging to dominant repIS-Ps and to less frequent repIS-Ps on a same host suggests a better ecological adaptation of the isolates belonging to the dominant repIS-Ps to the host, or to diseased tissues of the host (the isolation places), or both.

The existence of phytopathogenic isolates belonging to different dominant repIS-Ps on different hosts (pear, cherry and plum) is very interesting. One possible explanation could be that these clonal lineages reflect the effective dissemination of the pathogen with the planting material, or the occurrence of an effective dissemination of bacteria among trees of the same species because of pruning tools, for instance. However, the Belgian isolates isolated in 1999 in Table 2 were all isolated from the orchards of the Walloon Agricultural Research Centre in Gembloux and the pear orchard is directly near the cherry variety collection. Close proximity between cherry and pear orchards was also observed in other locations in Wallonia. It is difficult to imagine that exchange of strains would never have occurred under these conditions. Also, we tested in this study two reference Pss strains isolated in 1959 from pear in the UK and both were similar or identical to repIS-P2 containing 45 Belgian isolates from pear isolated in 2001 and 2002 in various pear orchards (Tables 2 and 4). The UK strain belonging to repIS-P2 is the well known phytopathogenic strain B301D, which has been the most studied Pss strain regarding TLP production (Bender et al. 1999). Moreover, we also tested in this study two phytopathogenic reference Pss strains isolated in 1979 from cherry in France and both were similar or identical to repIS-P3 containing 23 Belgian isolates from cherry (Tables 2 and 4). This is intriguing because it is very difficult to imagine that no exchange of strains would have occurred over such long periods. All this suggests that clonal populations of Pss could be specifically adapted to certain cultures (pear, cherry, and plum) in Belgium. Such specialisation was suggested by Little et al. (1998) with regard to strains isolated from Prunus in California, compared with strains from other hosts. Nevertheless, in this study a distinction could be made between Belgian Pss isolates isolated from pear, cherry and plum.

Interestingly, several clonal populations were detected on a same host, indicating, as discussed by Sarkar et al. (2006), that there could be different ways by which P. syringae can adapt to the same host. Interestingly, the TLP+ Pss strains from South Africa and Switzerland were genetically different. This could suggest a distinct evolution of Pss in Belgium, France and the UK compared with other regions of the world. Little et al. (1998) also suggested a distinct evolution between North American and western European Pss strains.

The pathogenicity tests on lilac leaves gave interesting results. Although most TLP+ Pss isolates from the main repIS-Ps induced progressive necrotic lesions on lilac, some isolates from these patterns were non-pathogenic to lilac. The study therefore indicated that isolates with identical genetic profiles using rep- and IS50-PCR can differ in their pathogenicity to lilac. Scheck et al. (1997) had already shown that TLP+ Pss could be non-pathogenic to lilac. In addition, some English Pss strains from the same rep-PCR-based genetic groups varied in their pathogenicity to lilac (Vicente and Roberts 2007). These data suggest that negative responses in a lilac pathogenicity test should be interpreted with caution. More surprisingly, two Psm race 2 isolates, two Belgian unclassified P. syringae isolates and one P. viridiflava isolate were pathogenic to lilac leaves in this study. These isolates were TLP-, belonged to repIS-Ps clearly distinct from the Pss patterns, and could not be identified as Pss by physiological tests. It appears, therefore, that a positive result in a lilac pathogenicity assay should also be viewed with caution. Scortichini et al. (2003) also showed that lilac leaf and petiole inoculation assays were important, but not decisive, in confirming Pss identification.

In this study, four BOX patterns contained 100% of the TLP+ Pss isolates pathogenic to lilac and 88.4% of the Belgian TLP+ Pss isolates belonged to these BOX-Ps. Two of these patterns were close to each other, but the two others were genetically very different. Besides, 11.6% of the Belgian TLP+ P. syringae isolates from fruit orchards were non-pathogenic to lilac, were not found frequently in orchards and, as in the case of reference Pss strains from other hosts, were dispersed in BOX-Ps that were clearly different from the four BOX-Ps pathogenic to lilac. As in the case of TLP− Pss isolates, it is questionable whether they should be named Pss. However, the exclusion of these isolates from Pss should be done with caution. Pss is known to be genetically heterogeneous (Weingart and Völksch 1997; Little et al. 1998), to attack many different hosts (Young 1991) and to be dispersed in various ways (Morris et al. 2007), so it is possible that Pss strains not specifically adapted to attack pear, cherry or plum, but possibly pathogenic on other hosts could occasionally be encountered as epiphyte strains in Belgian orchards due to aerial, dust or rain propagation. In addition, it cannot be assumed that the isolates non-pathogenic to lilac will not be pathogenic to the host from which they have been isolated. In a currently global pathogenicity study currently underway we verified that 88.2% of the TLP+ isolates that were non-pathogenic in our lilac pathogenicity test were pathogenic in at least one other pathogenicity test on another host, and that 96.4% of the tested Belgian Pss isolates producing TLP were pathogenic in at least one test (Gilbert and Bultreys, unpublished results).

Young et al. (1992) considered, in a review on taxonomy of plant pathogenic bacteria, that the production of syringomycin by Pss formed part of the circumscription of the pv. syringae and that this character as well as other determinative tests can be used in identification. They considered that avirulent strains could even be placed in a pathovar if an appropriate identification test had been used. Indeed, pathogenicity tests should not be seen as the only method of allocating stains to a pathovar, but as the basis of circumscription of a pathovar (Young et al. 1992). For all these reasons, we believe that it is presently more appropriate to keep the P. syringae isolates non-pathogenic to lilac but producing TLP or giving the appropriate GATTa responses in the pv. syringae, rather than to name these isolates P. syringae. It is, however, clear that the pv. syringae is heterogeneous in different characteristics and that additional classification work would be useful. Taken together, the results of this study show that the use of different methods to identify Pss can lead to different conclusions and that it is still not clear what strains should be included in this pathovar.

Pseudomonas syringae pv. morsprunorum

REP, BOX, ERIC and IS50 analyses generated distinct profiles for Belgian isolates of the two races of Psm, confirming earlier observations in the UK and France (Ménard et al. 2003; Vicente and Roberts 2007). The BOX analysis was a good method for identifying Psm race 1 and race 2 because clearly related profiles were obtained for all members of each race. In addition, no Belgian Psm race 2 isolate was a coronatine producer, whereas the Psm race 1 isolates were mainly coronatine producers. Also, all the Psm race 2 isolates possessed the yersiniabactin gene irp1, but not the Psm race 1 isolates (Bultreys et al. 2006; Bultreys and Gheysen, unpublished results). Vicente et al. (2004) reported differences in biochemical, physiological and serological tests between Psm race 1 and race 2. Psm race 1 and race 2 belong to two distinct genomospecies when analysed by DNA–DNA hybridisation (Ménard et al. 2003). Freigoun and Crosse (1975) showed that the two races represented two distinct physiological and pathological groups. In Belgium, Psm race 2 strains were isolated mainly from sour cherry (P. cerasus) and Psm race 1 strains mainly from sweet cherry (P. avium; Bultreys and Gheysen 2004). In the UK both races were reported on sweet cherry and wild cherry (Freigoun and Crosse 1975; Vicente et al. 2004), but there is no information about the occurrence of Psm race 2 on sour cherry. All these elements suggest that these two races are clearly distinct pathogens that attack the same hosts, maybe at different frequencies, and keeping them in a same pathovar probably needs to be further evaluated.

Interestingly, it was noticed in this study that the pathotype strain Psm LMG 5075 (t2) had the same genetic profile as Psm race 2 strains and is therefore a Psm race 2 rather than a Psm race 1 strain. This was also noticed with the equivalent strain CFBP 2351, which had previously been shown to belong to the genomospecies 3, as Psm race 2 strains, rather than 2, as the Psm race 1 strains (Gardan et al. 1999; Ménard et al. 2003). According to the LMG culture collection information, this strain was isolated by H. Wormald from plum and is probably a representative from his original study in UK on P. morsprunorum (Wormald 1932). In his work, Wormald (1932) noticed strains that were gelatine hydrolysis-positive and others that were negative. This was noted by Freigoun and Crosse (1975) as a distinctive character between the two races of Psm in the UK. The disease symptoms described by Wormald are clearly related to the symptoms generally attributed to race 1 of the pathogen (Garrett et al. 1966). One explanation could be that Wormald (1932) encountered both types of strains, as discussed by Freigoun and Crosse (1975), but that a Psm race 2 strain was finally chosen as the pathotype strain of P. morsprunorum, although it was not the best representative of the most damaging strains causing the disease he described.

In conclusion, the present study shows the high diversity existing among P. syringae isolates from Walloon fruit orchards, and that combining rep and IS50 analyses is a good way of assessing this diversity. It shows isolate–host relationships that could possibly reflect specialisation of Pss isolates on their host, and indicates that, beside Pss and both races of Psm, unclassified P. syringae strains are present in Belgian pear, sour cherry and plum orchards. BOX-PCR proved useful for identifying Psm race 1 and race 2 and indicated that four BOX patterns contained 100% of the TLP + Pss isolates pathogenic on lilac. The study also indicates that some Pss isolates with common genetic profiles can vary in pathogenicity on lilac and that isolates that are clearly not Pss isolates can be pathogenic on lilac leaves.