Corps de l’article


The invasive ascomycete fungus Ophiostoma novo-ulmi Brasier is a highly aggressive vascular wilt pathogen of elms and the principal agent of Dutch elm disease today. Over the past 50 years, it has progressively replaced the less aggressive Ophiostoma ulmi (Buism.) Nannf., which was responsible for the first pandemic of Dutch elm disease across Europe, central Asia and North America in the early 1900s (Brasier 1990). Ophiostoma ulmi and O. novo-ulmi differ in many behavioural and morphological properties. For example, they differ in their growth-temperature optima of ca 22°C and 28°C respectively and in their perithecial dimensions; O. ulmi secretes little or none of the cell surface hydrophobin cerato-ulmin while O. novo-ulmi secretes high levels; and the two species are characte-ristically non-overlapping in their levels of aggressiveness on elms of moderate resistance (Brasier 1991). They also differ extensively in their molecular architecture and are phylo-genetically distinct and probably anciently divergent (Pipe et al. 1995). From their temperature relation and other properties, O. ulmi and O. novo-ulmi are believed to be adapted to tropical and temperate conditions in their res-pective centres of origin (Brasier and Mehrotra 1995).

Both species are outcrossing with two mating types, designated MAT-1 and MAT-2, controlled by alleles at a single Mat locus located on chromosome 2 of the O. novo-ulmi H327 genome (Comeau et al. 2015). They are also strongly but not totally reproductively isolated at the pre- and postzygotic levels (Brasier 1986b, 1991). Detailed analysis of natural populations in O. ulmi/ O. novo-ulmi overlap zones has revealed the occurrence of rare and transient O. ulmi x O. novo-ulmi interspecific hybrids with distinct phenotypes and reduced growth and parasitic fitness (Brasier et al. 1995, 1998; Pipe et al. 2000). An unusual, only moderately pathogenic isolate of O. novo-ulmi, AST27 (Brasier 1986a, 1987), collected in Iran at a location where O. novo-ulmi had recently replaced O. ulmi (Brasier and Afsharpour 1979), was later shown to be an introgressant harbouring some O. ulmi-specific alleles (Bates et al. 1993; Et-Touil et al. 1999). These included Pat1-m, a putative O. ulmi pathogenicity allele conferring the mode-rately aggressive phenotype in AST27 (Et-Touil et al. 1999, 2005), and the OPHIO2-int. DNA transposon (Bouvet et al. 2008). Analysis of population samples collected in Portugal and Poland during the 1980s, when O. novo-ulmi was in the process of displacing O. ulmi, revealed other interspecific hybrids, including individuals with intermediate aggressi-veness and a distinctive growth response to temperature (Brasier et al. 1998).

These hybrids and introgressants are known to have a major role in the continuing evolution of O. novo-ulmi (reviewed in Brasier et al. 2004). They also provide an opportunity to identify genes associated with fitness. Thus, previous genetic analysis of introgressant isolate AST27 allowed the identification of the Pat1 pathogenicity locus (Et-Touil et al. 1999). Matching alleles for moderate (Pat1-m) and high (Pat1-h) aggressiveness were identified in AST27 and in normal and highly aggressive O. novo-ulmi isolate H327 respectively, whereas molecular studies with restriction fragment length polymorphisms (RFLPs; Bates et al. 1993) and random amplified polymorphic DNAs (RAPDs; Et-Touil et al. 1999) confirmed that a portion of the AST27 nuclear genome comprised O. ulmi-specific sequences. Some, but not all, of the latter were found to be linked to Pat1 (Et-Touil et al. 1999).

In these previous studies, lower aggressiveness towards moderately resistant elms was the only phenotypic trait found to distinguish introgressant AST27 from other isolates of O. novo-ulmi (Brasier 1986a). However, the evidence for O. ulmi-like DNA sequences at various locations in the AST27 genome led us to investigate whether this intro-gressant isolate expressed other O. ulmi-like behavioural traits that might contribute to fitness. In the present study, mycelial growth rates at different temperatures, yeast phase development and cerato-ulmin production levels were quantified in progeny sets from controlled sexual crosses involving AST27 and in wild type isolates of O. novo-ulmi and O. ulmi. Molecular polymorphisms among the isolates were then investigated by polymerase chain reaction (PCR) using specific amplification of coding loci and random amplification of anonymous loci and possible associations between the various characters and isolates analyzed.


Isolates and culture media

The following isolates, described by Et-Touil et al. (1999), were analyzed: highly aggressive O. novo-ulmi subsp. novo-ulmi isolate H327 (Pat1-h, MAT-1); moderately aggressive O. novo-ulmi subsp. novo-ulmi introgressant isolate AST27 (Pat1-m, MAT-2); 40 F1 progeny from the cross H327 x AST27; and 50 progeny from the backcross H327 x A2P30 (Pat1-m, MAT-2). An additional set of 93 F1 progeny was recovered from the cross H327 x AST27. Ophiostoma novo-ulmi subsp. novo-ulmi isolates AST20 (MAT-1) and CKT11 (MAT-2), O. novo-ulmi subsp. americana isolates FG245 (MAT-1), W2 (MAT-2), MH75 (MAT-2) and CESS16K (MAT-2), and O. ulmi isolates Q412T (MAT-1), R21 (MAT-1) and W9 (MAT-2) were also used in some tests. Isolates were routinely grown on solid complete medium (CM) with 2.0 g L-1 ammonium sulfate as nitrogen source (Bernier and Hubbes 1990a). Genetic crosses were carried out on elm sapwood agar (Brasier 1981) supplemented with linoleic acid at 6 ml/L (Bernier and Hubbes 1990a, 1990b). The mating type of progeny strains was determined by crossing each individual with strains H327 and AST27. For long-term storage, aliquots of yeastlike cells grown in liquid CM with L-proline (1.15 g L-1) as nitrogen source (Bernier and Hubbes 1990a) were mixed with glycerol (to a final concentration of 15%) and stored at -80°C (Bernier 1993). Modified 2% Oxoid (Nepean, ON, Canada) malt extract agar (MEA) was prepared as described in Brasier (1981).

Growth measurements

Mycelial growth rate on solid medium was measured according to Brasier (1981). Each isolate was inoculated centrally on two sets of MEA plates and incubated in the dark at 21°C and 28°C (Brasier et al. 1981). After 2 d (T1) and 7 d (T2), the diameter (D) of each colony was measured. The mean radial growth rate of each isolate was calculated from three replicates using the equation described by Brasier and Webber (1987): D2-D1/2(T2-T1) = Radial growth rate per day.

Growth curves were established for yeastlike cell cultures in liquid CM using the method of Bernier and Hubbes (1990a). One-week-old cultures in liquid CM were filtered through eight layers of cheesecloth to remove clumps of cells and mycelia and centrifuged at 4100 x g for 5 min. The pellet was resuspended in sterile distilled water at 1 x 106 cells mL-1. A 500-µL aliquot was added to 50 mL of medium in a 125-mL Erlenmeyer flask and incubated on a rotary shaker (120 rev min-1) at 25°C in the dark. The cell concentration was quantified with a haemocytometer from three replicates and monitored over 8 d.

Cerato-ulmin measurement

Cerato-ulmin production levels in 1-wk-old culture filtrates of O. novo-ulmi H327, O. ulmi W9 and Q412T, introgressant AST27, F1 progeny A2P5 (Pat1-h, MAT-2) and A2P30 (Pat1-m, MAT-2), and backcross progeny A3P48 (Pat1-h, MAT-2) and A3P11 (Pat1-m, MAT-1), grown in liquid shake culture at 25°C in CM, were determined by the turbidity method described by Takai and Richards (1978) and expressed by the cerato-ulmin production index (CPI). The CPI of each isolate was measured as 100 x the optical density at 400 nm from three replicates.

DNA extraction and amplification

Total genomic DNA was extracted from yeastlike cells grown in liquid CM, according to the procedure of Zolan and Pukkila (1986). DNA was amplified by the polymerase chain reaction (PCR) using RAPD primers (Operon Technologies, Alameda, CA, USA) as described by Et-Touil et al. (1999), as well as primers designed for the specific amplification of coding sequences (Dewar and Bernier 1995; Dusabenyagasani et al. 2000). Bulked segregant analysis (Michelmore et al. 1991) was used to find RAPD markers linked to the mating-type locus. Two bulked DNA samples were prepared from 15 MAT-1 strains (A2P2, A2P3, A2P4, A2P10, A2P11, A2P16, A2P19, A2P20, A2P21, A2P22, A2P25, A2P26, A2P33, A2P35, A2P40) and 15 MAT-2 strains (A2P1, A2P5, A2P6, A2P7, A2P8, A2P9, A2P12, A2P13, A2P14, A2P15, A2P18, A2P23, A2P29, A2P31, A2P34) from the original set of 40 F1 progeny from the H327 x AST27 cross (Et-Touil et al. 1999). The bulked DNA and parent DNA were screened for differences using 80 RAPD primers as described by Et-Touil et al. (1999). RAPD amplicons were electrophoresed in 1% agarose and 0.5% Synergel (Diversified Biotechologies, Boston, MA, USA) gels in 0.09 M Tris-phosphate and 0.002 M EDTA, and visualized by ethidium bromide fluorescence. Gel electrophoresis, alone or in combination with RFLP or single strand confor-mation polymorphism (SSCP) analysis (Orita et al. 1989) was used to detect polymorphisms in PCR-amplified coding sequences located on six of the eight chromosomes (chr) found in O. novo-ulmi H327 (Comeau et al. 2015): 60S ribo-somal protein L34-B and hypothetical eukaryotic translation initiation factor 6 (chr 1); β-tubulin, mitochondrial ATPase inhibitor, 60S ribosomal protein L35 and elongation factor 1-α (chr 2); 60S ribosomal protein L30, cytochrome C oxidase chain VIIc-like protein, histidyl-tRNA mitochondrial, cerato-ulmin and nuclear ribosomal rRNA (chr 3); O-glycosyl hydrolase (chr 4); probable eukaryotic translation initiation factor 2-β, BRCT-containing protein 1 and serine proteinase-like protein (chr 6); actin (chr 8); (Genbank accession numbers AF378547 to AF378553, and AF378555 to AF37867). In addition, Ins-OPHIO2-AST27-L and R primers were used to amplify AST27-specific DNA transposon OPHIO2-int. (Bouvet et al. 2008) in the H327 x AST27 F1 progeny.

Statistical analysis of growth rate and linkage data

Mycelial growth rate data for the two sets of H327 x AST27 F1 progeny, and the H327 x A2P30 backcross were subjected to ANOVA followed by a Duncan’s multiple comparison test. The relationships between mycelial growth rate, pathoge-nicity, mating-type, and molecular markers were analyzed by carrying out two-tailed Student’s t tests. These statistical analyses were performed with SAS version 6.1 software package (SAS Institute Inc., Cary, NC, USA). Recombination among the physiological and molecular markers was studied in the first set of 40 H327 x AST27 F1 progeny, using MapMaker Macintosh V2.0 Software (Lander et al. 1987). Linkage among markers segregating in Mendelian fashion was verified with a logarithm of the odds (LOD) score of 4.0 and Theta value of 0.3. Linkage distances, estimated using the Kosambi function, were first calculated by a two-point analysis, and then by multipoint analysis.


Forty H327 x AST27 F1 progeny (F1 set 1) and fifty H327 x A2P30 backcross progeny examined for pathogenicity by Et-Touil et al. (1999) during the identification of the Pat1 locus were investigated for mating type, for other phenotypically observable traits, for segregation of putative quantitative trait loci (QTLs), and for additional associated molecular markers. A further 93 F1 progeny from the cross H327 x AST27 (F1 set 2) was also examined in many of the tests.

Mating types

The MAT-1 and MAT-2 mating types segregated in Mendelian fashion in H327 x AST27 F1 set 1 (15:25; χ2 = 2.500) but not in set 2 (34:59; χ2 = 6.720, < 0.01). The two mating types also segregated in Mendelian fashion (22:25; χ2 = 0.191) in the 47 isolates from the H327 x A2P30 backcross progeny for which mating reactions could be determined.

Mycelial growth rate at 21°C and 28°C

Growth rate tests of the two F1 and one backcross progeny sets were conducted on MEA at 21°C and at 28°C, close to the optima for O. novo-ulmi and O. ulmi, respectively (Brasier et al. 1981). The data were then analyzed for their relationship to parental growth rates, to mating type and to pathogenicity (Pat1-h or Pat1-m). The pathogenicity data were available only for F1 set 1 and for the backcross progeny (Et-Touil et al. 1999). The results are summarized in Tables 1 and 2 and Figs. 1 and 2.

At 21°C, O. novo-ulmi isolate H327 grew significantly faster (4.95 mm d-1; F = 7.70, P < 0.001) than O. novo-ulmi introgressant AST27 (4.11 mm d-1); and AST27 grew signifi-cantly faster than O. ulmi isolates Q412T and W9 (3.25 and 2.95 mm d-1, respectively) (Fig. 1A). Both H327 and AST27 therefore show the O. novo-ulmi characteristic of a faster growth rate than O. ulmi at this temperature. Linear growth rates of both F1 progeny sets were continuously distributed (Fig. 1A for set 1). For each set, the overall progeny mean did not differ significantly from the H327/AST27 midparent value. In set 1, the 21 Pat1-h and the 19 Pat1-m siblings exhibited comparable mean mycelial growth rates. The 15 MAT-1 individuals, however, grew significantly (P < 0.01) faster than their 25 MAT-2 siblings (Table 1). This behavior was also observed in the 93 F1 progeny set 2.

At 28°C, O. novo-ulmi AST27 (3.05 mm d-1) and O. ulmi Q412T (4.21 mm d-1) grew significantly faster than O. novo-ulmi H327 (2.95 mm d-1; F = 6.72, P < 0.001). Isolate AST27 also grew significantly faster than H327 at 28°C in two retests (4.98 vs 4.30 and 5.12 vs 4.17 mm d-1, respectively) i.e. it exhibited a more O. ulmi-like property. In both F1 progeny sets, the distribution of growth rates was again continuous (Fig. 1B for set 1). The overall means of the two progeny sets were again similar to their respective H327 and AST27 midparent values. With set 1, however, the Pat1-m progeny grew at a significantly higher rate than Pat1-h progeny (P < 0.001), whereas the MAT-1 and MAT-2 progeny of both F1 sets exhibited similar growth rates (Table 1).

Table 1

Relationship between mycelial growth rate, mating type and pathogenicity in O. novo-ulmi H327 x AST27 progeny sets

Relationship between mycelial growth rate, mating type and pathogenicity in O. novo-ulmi H327 x AST27 progeny sets

a Vertical bar, significant difference between the two values: *, < 0.05; **, < 0.01; ***, < 0.001.

b In two retests, no significant differences in growth rate at 28°C were observed between MAT-1 and MAT-2 backcross progeny.

-> Voir la liste des tableaux

The 50 H327 x A2P30 backcross progeny also exhibited continuous distributions for growth rates at 21°C and 28°C (Fig. 2). The 24 Pat1-m individuals and 26 Pat1-h individuals displayed a similar growth rate at both 21°C and 28°C (Table 1). The 22 MAT-1 progeny grew significantly faster, on average, than the 25 MAT-2 progeny at 21°C (P < 0.001; Table 1).

Overall, therefore, these data suggested that genetic factors controlling growth rate were segregating among the H327 x AST27 progeny and that some of these factors might be associated with the Mat1 or Pat1 loci.

Table 2

Physiological characteristics of O. novo-ulmi (H327), introgressant (AST27), progeny from H327 x AST27 crosses (A2P5, A2P30), backcross progeny from H327 x A2P30 crosses (A3P11, A3P48), and O. ulmi (W9 and Q412T)

Physiological characteristics of O. novo-ulmi (H327), introgressant (AST27), progeny from H327 x AST27 crosses (A2P5, A2P30), backcross progeny from H327 x A2P30 crosses (A3P11, A3P48), and O. ulmi (W9 and Q412T)

a The Pat1-m allele is expected to occur in O. ulmi strains.

b Within a line, means followed by a different letter differ significantly (< 0.05) according to Duncan’s multiple-range test following ANOVA.

-> Voir la liste des tableaux

Figure 1

Mycelial growth rate distribution at 21°C (A) and 28°C (B) of the 40 F1 set 1 progeny from a cross between the highly aggressive Ophiostoma novo-ulmi H327 and the moderately aggressive O. novo-ulmi introgressant AST27



Isolates W9 and Q412T belong to O. ulmi. All measurements are means of three replicates. Vertical bars represent standard error. A continuous distribution at 21°C and 28°C was also observed for a second set of 93 F1 progeny (data not shown).

-> Voir la liste des figures

Yeast growth and cerato-ulmin production

In liquid shake cultures, the yeast phases of O. novo-ulmi isolates H327 and AST27 grown in CM displayed nearly similar growth kinetics at 25°C. The only notable difference was observed after 96 h of incubation when cell concentrations of AST27 were lower than those recorded for H327 (Fig. 3). Both isolates grew more rapidly than O. ulmi Q412T and W9 after ca 80 h (Fig. 3). Similarly, H327 and AST27, two selected F1 progeny isolates (A2P5 and A2P30) and two selected backcross isolates (A3P48 and A3P11) produced comparably high levels of cerato-ulmin, whereas isolates Q412T and W9 produced very small amounts of cerato-ulmin (Table 2) typical of O. ulmi (Scala et al. 1997; Takai 1980). Since yeast growth kinetics and cerato-ulmin production were similar for parent isolates, these traits were not studied further in progeny isolates.

Molecular polymorphisms

Sixteen coding sequences which had been previously assigned to the six O. novo-ulmi chromosomes distinguishable by pulsed-field gel electrophoresis (Dewar and Bernier 1995; Dewar et al. 1997; Dusabenyagasani et al. unpubl.) were amplified in H327 and AST27 and subjected to RFLP or SSCP analysis to determine whether they corresponded to O. ulmi or O. novo-ulmi specific alleles. Results indicated that intro-gressant AST27 harboured no O. ulmi-specific alleles among the 16 coding loci analyzed. Following sequencing and assembly of the O. novo-ulmi H327 nuclear genome into eight chromosomes (Forgetta et al. 2013) and its subsequent annotation (Comeau et al. 2015), it was found that two chromosomes (chr 5 and 7) were not represented in the set of 16 coding sequences. On the other hand, the OPHIO2-int. DNA transposon (Bouvet et al. 2008) segregated in the progeny (Table 1).

A comparison of RAPD patterns of H327 and AST27 yielded 26 additional reproducible polymorphisms. These included the 10 RAPD loci linked to Pat1 reported previously (Et-Touil et al. 1999). These loci were assigned to chDNA II (Et-Touil et al. 1999) which corresponds to chromosome 1 of the O. novo-ulmi reference genome (Forgetta et al. 2013; Comeau et al. 2015). Seven RAPD loci were polymorphic between MAT-1 and MAT-2 bulked samples of F1 progeny: OPA8500, OPA111900, OPA121900, OPJ1600, OPJ61200, OPK12500, and OPK171400. One of these markers, OPA121900, was am-plified in parent strain AST27 and in O. ulmi strains but not in parent strain H327 nor in other O. novo-ulmi strains tested. Recombination analysis of the first set of 40 F1 progeny confirmed that the above markers were all linked to the Mat1 locus, located on O. novo-ulmi H327 chromo-some 2 (Comeau et al. 2015), and further indicated that an additional locus, OPJ1680, was also linked to Mat1. These eight RAPD markers and the Mat1 locus segregated independently from Pat1 and the 11 molecular markers linked to it. Among the remaining eight RAPD polymorphisms, three loci (OPC2635, OPC2760 and OPK18760) cosegregated only with each other and were assigned to linkage group III (O. novo-ulmi H327 chr 8). Locus OPC21100 was tentatively assigned to chromosome 3 (Dusabenyagasani et al. unpubl), whereas OPC191900, OPJ1620, OPJ12050 and OPK51060 could not be assigned to any linkage group or chromosome (Fig. 4).

Figure 2

Mycelial growth rate distribution at 21°C (A) and 28°C (B) of 50 progeny from a backcross between the highly aggressive Ophiostoma novo-ulmi H327 and a moderately aggressive F1 progeny (A2P30)



All measurements are means of three replicates. Vertical bars represent standard error.

-> Voir la liste des figures

Figure 3

Yeastlike cell growth kinetics of O. novo-ulmi (H327), O. novo-ulmi introgressant (AST27) and O. ulmi (Q412T, W9) in liquid complete medium

Yeastlike cell growth kinetics of O. novo-ulmi (H327), O. novo-ulmi introgressant (AST27) and O. ulmi (Q412T, W9) in liquid complete medium

Triplicate cultures were incubated on a rotary shaker (120 rev min-1) at 25°C in the dark. Standard errors are shown only when they exceed the size of the symbol.

-> Voir la liste des figures

Figure 4

Linkage analysis of coding loci (Pat1 and Mat), transposon OPHIO2-int. and 26 anonymous RAPD loci

Linkage analysis of coding loci (Pat1 and Mat), transposon OPHIO2-int. and 26 anonymous RAPD loci

The map is based on the analysis of 40 F1 set 1 progeny from the cross between Ophiostoma novo-ulmi H327 (Pat1-h, MAT-1) x introgressant AST27 (Pat1-m, MAT-2). Map distances, estimated with the Kosambi function, were calculated by multipoint analysis with a LOD score of 4.0 and theta of 0.3, using MapMaker V2.0 for Macintosh.

-> Voir la liste des figures

Relationship between RAPD markers and growth rate

Mycelial growth rates of the 40 F1 set 1 progeny data were compared to their allelic differences at the RAPD loci described above. Allelic differences at all eight RAPD loci linked to Mat1 were associated with significant (P < 0.01) differences in mycelial growth rate at 21°C (Table 3). None of the polymorphisms at the remaining RAPD loci were associated with the significant differences in mycelial growth rates observed at this temperature (Table 1). The OPHIO2-int. transposon and 10 RAPD polymorphisms linked to Pat1 were associated with significant (P < 0.01) differences in mycelial growth rate at 28°C (Table 3). None of the other RAPD loci investigated showed a significant association with mycelial growth rate at 28°C.

Table 3

Segregation of parental Pat1 or Mat alleles, molecular markers and mycelial growth rates in the set 1 F1 progeny from the cross H327 x AST27

Segregation of parental Pat1 or Mat alleles, molecular markers and mycelial growth rates in the set 1 F1 progeny from the cross H327 x AST27

a α and β represent H327-type and AST27-type alleles, respectively.

b For any given locus, values are the difference in mean mycelial growth rate (mm d-1) between progeny with the H327 allele (α) or the AST27 allele (β). *, P < 0.05; **, P < 0.01.

-> Voir la liste des tableaux


In a previous study, the unusually low aggressiveness of AST27 relative to wild-type O. novo-ulmi isolates (Brasier 1986a) was shown to be due to a single nuclear allele, Pat1-m. Since this allele was also closely linked to two of the five O. ulmi-specific amplicons found in the AST27 genome, it was considered to be introgressed from O. ulmi (Et-Touil et al. 1999). In this study we have used a combination of physiological and molecular tests to identify other traits in AST27 and its derivatives that might be both genetically tractable and associated with interspecific hybridization.

One feature observed in AST27 was that it exhibited distinctive mycelial growth rates at 21°C and at 28°C, close to the optima for growth of O. novo-ulmi and O. ulmi, respectively (Brasier et al. 1981). When incubated at either temperature, isolate AST27 exhibited a growth rate that was intermediate between the rates of O. novo-ulmi H327 and O. ulmi Q412T. Two sets of H327 x AST27 F1 progeny, and a backcross population showed a continuous distribution for growth rate at both 21°C and 28°C. This was in marked contrast to the heritability pattern of their pathogenicity phenotypes, which segregated 1:1 for highly vs moderate aggressiveness, consistent with monogenic control (Brasier 1987; Et-Touil et al. 1999). The observed differences in temperature-growth responses of H327 and AST27 at 21°C and 28°C were therefore most probably due to multiple genetic differences. The assumption that some of these loci were segregating independently from each other was supported by the fact that some of the fastest growing progeny isolates at 21°C were among the slowest growing at 28°C and vice versa (Figs. 1 and 2).

Co-analysis of additional markers with the F1 set 1 progeny further supported the hypothesis that the growth rate differences between isolates H327 and AST27 were due to allelic differences at several loci. The MAT-1 progeny grew faster than their MAT-2 siblings at 21°C and the Pat1-m progeny grew faster than the Pat1-h progeny at 28°C. Similar results were obtained when linkage of molecular markers to Pat1 or Mat was examined (Table 2). Linkage analysis of 29 genetic markers that included Pat1, Mat1, OPHIO2-int. and 26 RAPD polymorphisms (Fig. 4) clearly showed that the Pat1 and Mat1 loci belonged to different linkage groups (I and II, respectively), as expected from markers located on different chromosomes (1 and 2, respectively, according to Comeau et al. 2015).

Among the F1 and backcross progeny analyzed, however, some Pat1-h isolates grew significantly faster than both parents at 28°C. Likewise, some MAT-2 individuals were among the fastest growing isolates at 21°C. These findings suggest that the Mat1 and Pat1 loci themselves do not directly influence mycelial growth rates at 21°C and 28°C; but are rather linked to two genetic factors or putative quantitative trait loci (QTLs) controlling mycelial growth rate (mgr) that we propose to call Mgr1 and Mgr2, respectively. The precise localization of these putative QTLs is unclear owing to the low number of informative genetic polymor-phisms found so far between H327 and AST27 and the small sets of progeny analyzed. Nevertheless, it is evident from the RAPD and growth rate data that both putative mgr QTLs are located in regions of the AST27 nuclear genome that contain many O. ulmi-specific sequences. It therefore appears likely that they are derived from O. ulmi and that intro-gression of O. ulmi DNA is responsible for the distinctive growth phenotype of AST27 in addition to its reduced aggressiveness (Et-Touil et al. 1999). We found no evidence for mgr QTLs elsewhere in the Ophiostoma genome. This is not surprising since we analyzed progeny derived from only two field isolates (H327 and AST27) and sampled a very small portion of the ca. 32-megabase nuclear genome of the Dutch elm disease fungi (Forgetta et al. 2013; Khoshraftar et al; 2013). Nevertheless, our data indicate that a region of chromosome 8 in introgressant AST27 also contains O. ulmi-specific RAPD markers. We therefore expect neighbouring genes to be polymorphic between isolates H327 and AST27.

The significant difference in mycelial growth rate between Pat1-m and Pat1-h F1 progeny at 28°C was not observed in the H327 x A2P30 backcross population. This apparent discrepancy may be explained by the genotype of backcross parent isolate A2P30, which not only carried the Pat1-m and Mat1-2 alleles from AST27 but also H327-type alleles for four of the 26 RAPD loci analyzed, viz. OPJ1680, OPC21100, OPC191900 and OPK51060. H327-type coding genes linked to any of these markers may have interacted with the Mgr1 locus.

The clustering of O. ulmi DNA around the Mat1 and Pat1 loci may have adaptive significance. Little is known about the genome reorganization in interspecific hybrids among filamentous fungi, but it is likely that natural selection will have a strong influence on the process (Brasier 1995). It has been shown that as O. novo-ulmi has replaced O. ulmi across the Northern Hemisphere it has acquired the Mat1-1 allele and vegetative compatibility (Vic) alleles from O. ulmi at a high frequency (Paoletti et al. 2006). We have also found that most O. ulmi DNA detected in the O. novo-ulmi isolates examined flanked the introgressed Mat1 and Vic loci. Brasier et al (2004) proposed that the most likely mode of introgression of the Mat1 and Vic loci was via sexual hybri-dization of O. novo-ulmi with O. ulmi followed by sequential backcrossing of resulting F1s to O. novo-ulmi. O. ulmi x O. novo-ulmi F1 progeny tend to be highly unfit and transient in nature, most being of very low pathogenicity. Backcross genotypes still carrying “unuseful” O. ulmi genes, i.e. alleles conferring low fitness on O. novo-ulmi, are also probably transient, i.e. eliminated by natural selection. The end result is likely to be selective acquisition and fixation by O. novo-ulmi of “useful” O. ulmi alleles (in this case, alleles at the Mat and Vic loci that confer resistance to fungal viruses) and a clustering of residual neutral O. ulmi DNA around these loci.

Isolate AST27 exhibits higher pathogenic fitness than most O. ulmi x O. novo-ulmi F1 progeny and is also more aggressive to healthy elms than is O. ulmi. It was originally isolated from the xylem of a symptomatic elm sapling and was therefore successful in the vascular wilt phase of the pathogen. AST27 shares most of its other properties with O. novo-ulmi. It may therefore be an intermediate or late backcross product in which some phenotypically important O. ulmi genes, including O. ulmi genes for lower aggressi-veness (Pat1-m) and faster growth rate at 28°C (Mgr2), are still present. In areas with populations of moderately to highly resistant elms, it is likely that AST27 would soon have been eliminated through competition with fitter O. novo-ulmi genotypes. However, in areas where susceptible elms are frequent, an AST27-type introgressant could overcome host resistance (Et-Touil et al. 2005) and be disseminated effi-ciently by elm bark beetles given its ability to secrete high amounts of cerato-ulmin (Temple et al. 1997). Genetic ana-lysis of progeny from laboratory crosses suggests that such introgressants might contribute additional useful traits to O. novo-ulmi through sexual hybridization. For instance, H327 x AST27 F1 isolate A2P33 and backcross isolates A3P26 and A3P33 grew well at both 21°C and 28°C (Figs. 1 and 2) and were highly aggressive (Et-Touil et al. 1999). Large scale genomic analyses of O. novo-ulmi populations in historical O. ulmi/O. novo-ulmi overlap zones are needed to determine whether, apart from O. ulmiVic and Mat1 loci, O. ulmi loci governing some other adaptive traits have been fixed or simply eliminated by intense post-epidemic selection.