Rapid emergence of hybrids between the two subspecies of Ophiostoma novo-ulmi with a high level of pathogenic fitness


  • This paper is dedicated to the memory of the late Adrian Van Magdeburg of De Dorschkamp, Wageningen who collected the disease samples from South Limburg, Netherlands

*E-mail: clive.brasier@forestry.gsi.gov.uk


During the 1970s Europe was invaded by two subspecies of the Dutch elm disease pathogen Ophiostoma novo-ulmi: subsp. americana from the west and subsp. novo-ulmi from the east. As a result their geographic ranges began to overlap in several areas. Only a weak prezygotic barrier to hybridization exists between the subspecies and in 1980 two hybrids were detected in the Netherlands. A subset of 107 O. novo-ulmi isolates collected in a subspecies overlap zone in Limburg, Netherlands in 1983 was characterized for three phenotypic markers and seven RAPD PCR markers. By phenotype, 33% were shown to be hybrid whereas by RAPD markers 69% were shown to be hybrid. Some isolates shown to be hybrid by phenotype were not revealed to be hybrid by PCR and vice versa. Combining the phenotype and RAPD data the estimated hybrid frequency was ∼78%. The mean growth rate of Limburg hybrid isolates was significantly faster than that of the Limburg subsp. novo-ulmi isolates but not significantly different from Limburg subsp. americana isolates. The Limburg hybrid isolates were just as pathogenic as the parent subspecies on both clonal Ulmus procera and on U. × Commelin. A subset of 100 isolates collected in another subspecies overlap zone at Orvieto, Italy in 1986 was also assessed with RAPD markers and 72% were shown to be hybrids. When 20 isolates of a ‘pure’ subsp. novo-ulmi population in the Baltic Ports area of Poland collected in 1980 were assessed by RAPD markers three isolates exhibited early introgression of subsp. americana DNA. This study therefore demonstrates very rapid emergence of O. novo-ulmi subspecies hybrids and introgressants in Europe in the early 1980s. In terms of two major fitness characters, growth rate and pathogenicity, these early hybrids were as fit as their parent subspecies. It is likely that complex hybrid swarms are now expanding across the continent.


The first pandemic of Dutch elm disease, caused by the ascomycete Ophiostoma ulmi, began in northern France around 1910 and spread rapidly to North America and across Europe to Central Asia (Gibbs, 1978). Since the mid 1900s, O. ulmi has been largely replaced across the same regions by a much more aggressive pathogen, O. novo-ulmi, the cause of the current second Dutch elm disease pandemic (Brasier, 1990, 2000a; Brasier & Buck, 2002). Ophiostoma ulmi and O. novo-ulmi are anciently divergent species (Pipe et al., 2000) and are believed to have originated in different biogeographic regions of eastern Asia. However, despite global surveys (e.g. Brasier, 1990; Brasier & Mehrotra, 1995) their origins remain unknown.

Ophiostoma novo-ulmi is polytypic, spreading in the form of two subspecies, subsp. novo-ulmi and subsp. americana, previously referred to as the Eurasian and North American races, respectively (Brasier, 1979; Brasier & Kirk, 2001). Subsp. novo-ulmi probably first arrived in the Black Sea – Ukraine area in the 1940s and migrated westwards across Europe, reaching parts of western Europe by the 1970s (Fig. 1). Subsp. americana is believed to have arrived in the southern Great Lakes area of North America in the 1940s and spread across the USA and Canada. It was introduced into the UK in the late 1960s on diseased elm logs from Ontario and spread to neighbouring parts of western Europe in the 1970s and 80s (Brasier & Gibbs, 1973; Brasier, 2000a; Brasier & Kirk, 2000).

Figure 1.

 Summary of the known distribution of Ophiostoma novo-ulmi subsp. americana and subsp. novo-ulmi in Europe 1990, based on >2500 samples collected by the authors. Open circles, subsp. americana; closed circles, subsp. novo-ulmi. Subspecies overlap zones included parts of Ireland, Netherlands, Belgium, Norway, Sweden, Denmark and Germany. LI, Limburg, Netherlands; OR, Orvieto, Italy; SZ, Szczechin (Baltic Ports) Poland. Modified from Brasier & Kirk (2001) omitting central Asia.

The above sequences of invasion and migration have brought these three previously geographically isolated Dutch elm disease pathogens into sudden geographic overlap and physical proximity. This has led to a potential for competitive interactions, especially during their long saprotrophic phase in diseased elm bark, when different taxa or genotypes may be introduced into the same tree by the vector scolytid beetles (Brasier, 1986a; Webber & Brasier, 1984; Webber et al., 1988). As the three taxa are obligatorily outcrossing, this has also resulted in a potential for novel evolutionary development via inter-taxon gene transfer.

The interaction between O. ulmi as a ‘resident’ species and O. novo-ulmi as an ‘invader’ has led to the rapid decline and virtual extinction of O. ulmi throughout most of its known range and its replacement by O. novo-ulmi (Brasier, 1986a, 2000a). It also led to a remarkable series of micro-evolutionary events. Although the two species exhibit a strong prezygotic mating barrier and low post-zygotic hybrid vigour, rare O.ulmi × O. novo-ulmi hybrids formed in the overlap zones, but were highly unfit and transient (Brasier et al., 1998). Deleterious fungal viruses also spread among O. novo-ulmi epidemic front populations, the viruses themselves probably originating from the resident O. ulmi population (Brasier, 1988, 2000b; Buck et al., 2002; Brasier et al., 2004). However, O. novo-ulmi also acquired sexual compatibility type (MAT-1) and vegetative compatibility (vic) loci from O. ulmi, probably via backcrosses between O. novo-ulmi and the transient hybrids (Paoletti et al., 2006). This resulted in greatly enhanced sexual recombination potential and in a rapid increase in diversity of vegetative compatibility phenotypes in the invasive O. novo-ulmi populations, which in turn probably suppressed the spread of the viruses (Brasier, 1988, 2000b; Brasier et al., 2004). Other ‘less useful’O. ulmi loci were also acquired by O. novo-ulmi in the process, such as O. ulmi pathogenicity and ceratoulmin (cu) genes (Et Touil et al., 1999; Pipe et al., 2001), but were probably quickly eliminated by natural selection.

Regarding the interaction between the two subspecies of O. novo-ulmi, extensive sample surveys across Europe and central Asia during the 1970s and 80s revealed a largely westerly distribution of O. novo-ulmi subsp. americana and an easterly distribution for subsp. novo-ulmi (Brasier & Kirk, 2001; Fig. 1), i.e. Europe was effectively subject to a pincer movement by the subspecies. The surveys also revealed a number of areas where the two subspecies were beginning to overlap. In some early overlap zones, such as Italy, Belgium, Netherlands and southern Scandinavia the two subspecies probably came into contact through confluent migration. Other overlap zones, such as south-west Ireland and southern Norway, probably represent sudden jumps by subsp. novo-ulmi into subsp. americana‘territory’ due to importation of diseased elm logs.

The two O. novo-ulmi subspecies differ in a range of phenotypic characters including colony development, perithecial size and shape, mean growth rate, pathogenicity and ceratoulmin production (Brasier, 1986b; Brasier et al., 1990; Brasier & Kirk, 2001) and can be distinguished by a wide range of neutral molecular markers (e.g. Bates et al., 1993a,b; Hoegger et al., 1996; Jeng et al., 1988; Pipe et al., 1995; Konrad et al., 2002; Santini et al., 2005; Paoletti et al., 2005). However, in contrast to O. ulmi and O. novo-ulmi, only a weak, unidirectional prezygotic mating barrier operates between the O. novo-ulmi subspecies, although this fertility interaction is distinctive and important for discriminating them in culture (Brasier, 1979, 1984). In addition, F1 progenies of laboratory generated subsp. novo-ulmi × subsp. americana crosses have shown no measureable reduction in growth or pathogenic fitness (Brasier, 1986a). Ophiostoma novo-ulmi isolates with recombinant colony types and fertility responses, i.e. with the phenotypic properties of subspecies hybrids, were identified from the Netherlands in 1980 (Brasier, 1986a,c). From this evidence, and from the absence of strong pre- and post-zygotic barriers to hybridization, it was suggested that relatively unrestricted hybridization was likely between the two O. novo-ulmi subspecies in nature (Brasier, 1986a,c, 2000a).

Several recent studies have confirmed that hybrids are emerging. Konrad et al. (2002) identified seven isolates from Austria as hybrids on the basis of their ceratoulmin (cu) and colony type (col1) locus polymorphisms. Brasier et al. (2004) reported hybrids in different parts of Europe on the basis of RAPD polymorphisms. Santini et al. (2005) identified several Italian isolates as subspecies hybrids from a combination of their PCR profile with the M13 primer and their fertility response. Solla et al. (2008) identified two isolates from Spain as subspecies hybrids based on RAPD profiles and fertility response. However, the population dynamics of the subspecies interaction in the field has yet to be investigated in detail. The behavioural properties of any emerging hybrids are of particular interest, since they may represent the phenotype(s) of O. novo-ulmi that will compete with and possibly replace the two subspecies in future (Brasier, 1986a,c; Brasier et al., 2004).

In view of the early indications of hybridization between the subspecies, dedicated population samples of O. novo-ulmi were collected from emerging subspecies overlap zones in the Netherlands and Italy between 1980 and 1986. The isolates were initially compared phenotypically to the typical subspecies types. Subsequently they were examined for their RAPD profiles when this technology became available. A sample from a possible subspecies introgression zone in north east Poland in 1980 was also characterized. The results of these observations are described here. They provide evidence of a rapid emergence of subsp. novo-ulmi × subsp. americana hybrids and of high growth and pathogenic fitness among the hybrids.

Materials and methods

Isolates studied, isolation methods, media and storage of cultures

The geographical origins of isolates of O. ulmi and of O. novo-ulmi subsp. novo-ulmi and americana used as controls or testers in this study are shown in Table 1. The locations of population samples of O. novo-ulmi from the Netherlands, Italy and Poland are described under Results. Each isolate was derived from a 1–2 cm diameter diseased elm twig with live bark collected from an individual diseased tree. Slivers of xylem showing dark streaks characteristic of infection were removed with a scalpel, plated on a selective medium containing penicillin, cyclohexamide and streptomycin (Brasier, 1981) and incubated at 20–22°C.

Table 1.   Isolates of Ophiostoma ulmi and the O. novo-ulmi subspecies used as testers and controls
Taxon and culture no.CountryLocationSampled by/dateSexual compatibility type (MAT)Used in experimenta
  1. aR, subspecies in RAPD analysis; PC, control isolate in pathogenicity tests; Fr, recipient tester in subspecies fertility response tests; Fd, donor tester in subspecies fertility response tests; SX, parent of inter-subspecies cross.

  2. bIsolate AST27 carries an introgressed O.ulmi pathogenicity gene (see Et Touil et al., 1999).

O. ulmi
R21RomaniaBozoviciC.M. Brasier1980MAT-1PC
YU99SerbiaAndranelovacC.M. Brasier1980MAT-2PC
O. novo-ulmi subsp. novo-ulmi
H346TurkeyBoluS. Sumer1979MAT-2R
H648UzbekistanTashkentH. Kryvkova1981MAT-2R
H371NetherlandsRhenenH.M. Heybroek1980MAT-2R
CA4UzbekistanTashkentC.M. Brasier1979MAT-2R
CA7UzbekistanTashkentC.M. Brasier1979MAT-2R
AST20IranAssalem ForestC.M. Brasier1977MAT-1R, Fr
H2038HungaryBudapestC.M. Brasier1989MAT-2R
P11PolandWarsawC.M. Brasier1980MAT-2R
R58RomaniaRacaciuniC.M. Brasier1980MAT-1R
V3RussiaVolgogradC.M. Brasier1979MAT-2R
P135PolandMiedzyzdoojeC.M. Brasier1980MAT-2R
TR64TurkeyEspiye, SamsunC.M. Brasier1980MAT-2R
TR98TurkeyDinar, IspartaC.M. Brasier1980MAT-2R
AST 27bIranAssalem ForestC.M. Brasier1977MAT-2Fd
H224DenmarkOdenseH. A. Jorgensen1978MAT-2Fd
H241IranAssalem ForestF. Afsharpour1978MAT-1Fr
H327SlovakiaBrezno-NizkeJ. Jamnicky1979MAT-1Fr, SX, PC
CKT11IranChachkam ForestC.M. Brasier1977MAT-2Fd
P114PolandTroszynC.M. Brasier1980MAT–2SX, PC,
O. novo-ulmi subsp. americana
H6NetherlandsSloten, FieslandH.M. Heybroek1972MAT-2R
H312FranceGirondeJ. Pinon1979MAT-1R
H161CanadaPhilip, NSJ.N. Gibbs1977MAT-2R
H937USAMillinocket, MED.A. Houston1986MAT-2R
W2UKTewkesbury, GloucestershireJ.N. Gibbs1973MAT-2R
ES78SpainCasade Campo, MadridC.M. Brasier1984MAT-2R
ES51SpainPeñarandaC.M. Brasier1984MAT-2R
ES120SpainGuadalajaraC.M. Brasier1984MAT-2R
0101UKOrsett, EssexC.M. Brasier1983MAT-2R
US290USAApple Creek, OHC.M. Brasier1996MAT-1R
H967CanadaSt Augustin, QEG.B. Oullette1986MAT-1R
MM2/1 UKGloucestershireM. Smith1974MAT-1Fr
RDT38UKReading, BerkshireC.M. Brasier1975MAT-2Fd
H363NICaledon, County, ArmaghD. Seaby1980MAT-2Fd
W4UKBasildon, EssexJ.N. Gibbs1973MAT-2Fd
C74UKChichester, SussexC.M. Brasier1982MAT-2SX, PC
S173UKSouthampton, HampshireC.M. Brasier1982MAT-1SX, PC

Elm sapwood agar (ESA), 2% Oxoid malt extract agar (OMEA) and Tchernoff’s liquid culture medium were prepared as described in Brasier (1981). Short-term stock cultures were maintained on OMEA at 20º and subcultured at 2 week intervals. Long-term stock cultures were maintained on OMEA slopes at −20º; and as a glycerol and spore mixture under liquid nitrogen.

Single-spore cultures were prepared by blending a small piece (1 mm3) of a colony growing on OMEA in 10 mL of sterile water, spreading 0·1 mL of the resulting spore suspension on an OMEA plate, incubating 18–24 h at 25º and transferring resulting germlings to a fresh OMEA plate using a dissecting microscope and a fine tungsten wire needle.

Growth rate, colony morphology and pathogenicity tests, and statistical analysis

Growth rate and colony morphology tests with two replicate plates per isolate were carried out on OMEA using the method described previously (Brasier, 1981). Subsp. novo-ulmi and americana exhibit different colony patterns on OMEA, and a distinctive and unpredictable wild type-non wild type colony dimorphism, termed the uniform powdery mutant or ‘up-mut’ factor, occurs in subsp. novo-ulmi but not in subsp. americana. In addition subsp. novo-ulmi colonies are on average slower growing than subsp. americana colonies (Brasier, 1986b). In the present study four growth rate / colony morphology tests were carried out to allow maximum expression of the up-mut growth dimorphism among the isolates.

In some but not all tests, subsp. novo-ulmi colonies are on average less pathogenic (Brasier, 1986b). Pathogenicity tests were carried out on 6-year-old and 3-year-old clonal English elm (Ulmus procera) and 3-year-old clonal Commelin elm (U. × ‘Commelin’) grown at 1 m spacing. The isolates to be tested were derived from single spores, allotted a code number, inoculated to a 10 mL universal bottle containing 5 mL of Tchernoff’s liquid culture medium, and shaken at room temperature for 3 days to induce the development of the ‘yeastlike’ stage of the fungus (see Brasier, 1981). Two drops of the resulting spore suspension (1 × 105 spores mL−1) were inoculated into the current annual ring of each tree at a point half-way (for the more resistant U. × ‘Commelin’) or one fifth-way (for the more susceptible U. procera) down the crown using a scalpel and a hypodermic syringe. Each experiment was randomized overall. Three replicate trees per isolate were used for U. procera and four replicate trees for U. × ‘Commelin’. Disease (as percentage defoliation of the crown) was assessed at 4- or 5-week intervals by three independent assessors. On Commelin elm, replicates of an isolate giving <6% defoliation at 12 weeks when other replicates were >40% were considered to be failed inoculations.

Because the experiments involved unequal groups, pathogenicity and growth rate data were analysed using the Restricted Maximum Likelihood (REML) model (Payne, 2007) with groups as fixed effect and isolates as a random effect. All data were checked for approximate normality using a normal probability plot.

Sexual compatibility type

Ophiostoma novo-ulmi is obligatorily outcrossing, with a single-locus, two-allele sexual compatibility system that operates independently of the vegetative incompatibility system (Brasier, 1984, 1986a). The sexual compatibility type of an isolate, MAT-1 or MAT-2 (previously known as the A and B-types, respectively), was determined by the presence or absence of perithecia when conidia were applied to a 1 cm2 patch on a mature colony of MAT-1 tester isolate MM2/1 grown on ESA (see Brasier, 1981).

Subspecies fertility barrier

Sexually compatible subsp. americana recipient (♀) × subsp. novo-ulmi donor (♂) pairings are fully fecund, producing abundant perithecia. Subsp. novo-ulmi (♀) × subsp. americana (♂) pairings have greatly reduced fecundity, forming on average only ∼10% as many perithecia as in subsp. novo-ulmi (♀) × novo-ulmi (♂) controls (Brasier, 1979, 1981, 1986b). Tests were carried out to identify the fertility response phenotypes of ‘unknown’ isolates from a potential hybrid zone in the Netherlands.

To test the fertility response of ‘unknown’MAT-2 isolates, standard MAT-1 tester isolates MM2/1 of subsp. americana and AST20, H241 and H327 of subsp. novo-ulmi (Table 1) were grown individually on ESA plates as recipients (♀). The resulting colonies were then fertilized in two 1 cm2 patches by donor conidia (♂) of the ‘unknown’ isolates, and by MAT-2 subsp. americana and subsp. novo-ulmi control isolates RDT-38 and CKT-11, using the patch fertilization method (Brasier, 1981). The total number of mature perithecia formed on each patch was counted and the mean perithecia per isolate on the four recipients calculated. ‘Unknown’ isolates producing 0·5–1·5 times the perithecial frequency in the control subsp. americana patches were scored as subsp. americana phenotype. Isolates producing 4–10 times the control frequency were scored as subsp. novo-ulmi phenotype.

To test the fertility response of ‘unknown’MAT-1 isolates, each unknown isolate was grown individually on two ESA plates as a recipient (♀), together with subsp. americana isolate MM2/1 and subsp. novo-ulmi isolates AST20, H241 and H327 as controls. The recipients were then fertilised in two 1 cm2 patches with donor conidia of MAT-2 tester isolates RDT38, H363 and W4 of subsp. americana and AST27, H224 and CKT11 of subsp. novo-ulmi. The mean number of mature perithecia produced by each recipient isolate with the six donors was then calculated. ‘Unknown’ isolates producing 0·5–2 times the perithecial frequency of the control subsp. americana isolate were scored as subsp. americana phenotype. Isolates producing 4–10 times the control frequency were scored as subsp. novo-ulmi phenotype.

Sexual crosses and generation of single ascospore progeny

Two sets of subsp. novo-ulmi × americana single ascospore F1 progenies were generated to use as controls in pathogenicity tests. These comprised isolates S173 (subsp. americana MAT-1) ×P114 (subsp. novo-ulmi MAT-2) and C74 (subsp. americana MAT-2) ×H327 (subsp. novo-ulmi MAT-1) (Table 1). The subsp. americana isolates were grown on ESA as recipients (♀) and fertilized by conidia of the subsp. novo-ulmi donor, using the patch fertilization method. Random single ascospore progeny were obtained as described previously (Brasier & Gibbs, 1976).

RAPD analysis of genomic DNA

Mycelium was harvested from 10-day-old colonies grown on MEA overlain with cellophane (Rogers et al., 1986). Genomic DNA was extracted using freeze dried ground mycelium. A primary extraction buffer containing 10 mm EDTA and 0·5% SDS was followed by a minimum of four phenol/chloroform/isoamyl alcohol [25:24:1] and one chloroform/isoamyl alcohol [24:1] extractions. After precipitation the DNA was subjected to an additional extraction with ribonuclease (Rnase One, Promega) to remove any extraneous RNA, and then dried down under vacuum and stored at −80º.

A Perkin-Elmer GeneAmp 9700 thermal cycler was used to amplify the DNA in 25 μL reaction volumes comprising 10 mm Tris-HCl [pH 9·0 at 25º], 50 mm KCl, 0·1% Triton X-100, 2 mm MgCl2, 0·2 mm each of dATP, dCTP, dGTP, dTTP, 5 pmoles Operon primer and 50 ng DNA. The template DNA was denatured at 95º for 4 min, then held at 72º while 2·5 units of Taq DNA polymerase (Promega) were added. The amplification conditions were: denaturing at 94º for 15 s, annealing at 36º for 1 min, followed by extension at 72º for 1 min. This sequence was repeated for 40 cycles after which samples were held at 72º for a further 9 min. Amplified products were loaded on horizontal 1·2% agarose gels (Molecular biology grade agarose, Sigma) prepared in 1× TAE electrophoresis buffer (prepared from ultrapure grade Promega 40× TAE buffer). Gels were run at 90 V for 3 h and stained with 0·5 μg mL−1 ethidium bromide and visualized under shortwave UV light. A 1 kb ladder (Promega) was used as the size marker.

The sequences of Operon oligonucleotide primers used in the study were as follows: OPA2 TGCCGAGCTG; OPA4 AATCGGGCTG; OPA8 GTGACGTAGG; OPB17 AGGGAACGAG; OPK16 GAGCGTCGAA; OPK19 CACAGGCGGA; OPK20 GTGTCGCGAG.


Characterization of a subspecies overlap population in Limburg, Netherlands

In 1980, 171 disease samples were collected across the 12 provinces of the Netherlands. Sixteen of the resulting isolates (∼9%) were of O. ulmi, representing the then rapidly declining O. ulmi population. The O. novo-ulmi isolates were examined for their colony type and the presence or absence of the up-mut colony dimorphism. Fourteen isolates that could not be clearly discriminated on colony type were also screened for their fertility response phenotype. On this basis subsp. americana was found to be predominant across the Netherlands (139 isolates), but subsp. novo-ulmi (14 isolates) was also present in the south and east of the country (Fig. 1). However, two isolates, H450 and H473 (from Arcen in Limburg and Elsen in Overijssel, respectively), exhibited an indeterminate colony type, the up-mut dimorphism of subsp. novo-ulmi and the fertility response of subsp. americana. These were considered to be subspecies hybrids, the first such hybrids to be identified.

Of particular interest in this Netherlands survey were the results for Limburg Province, a southern peninsula of the Netherlands lying between Belgium and Germany. Here, among 22 samples collected, fifteen (68%) were assigned to subsp. americana, seven (32%) to subsp. novo-ulmi and one (P450) was a putative hybrid. Limburg was therefore considered a promising location for investigating emergence of subspecies hybrids.

In 1983 a further 108 samples were collected in South Limburg, between Heerlen in the north and Aachen and Maastricht to the east and west. One isolate was O. ulmi. The remaining 107 isolates were O. novo-ulmi all of which were assessed for their colony type and for the up mut dimorphism in repeat tests. Eighty-six of the isolates were then tested for their fertility response, including all the isolates of uncertain colony type, all the probable subsp. americana isolates and most of the probable subsp. novo-ulmi isolates. Based on all three characters, 16 isolates (∼15%) were assigned to subsp. americana and 25 isolates (23%) to subsp. novo-ulmi. Another 31 isolates (29%) exhibited both the up-mut dimorphism of subsp. novo-ulmi and a subsp. americana fertility response and were designated as hybrids. A further 35 isolates gave equivocal results. All these isolates exhibited a subsp. novo-ulmi colony pattern and often up-mut dimorphism but their fertility response was intermediate between the subsp. novo-ulmi and americana type. Since they could not be assigned either to a subspecies or to the hybrid group they were designated ‘status undetermined’ and considered as possible hybrids.

Comparative growth rates of the hybrid and subspecies groups

The growth rates of isolates assigned to subsp. americana, subsp. novo-ulmi and hybrid groups in the above 1983 Limburg sample were compared in two consecutive tests (Growth tests 1 and 2). Two further tests (Growth tests 3 and 4) were also carried out routinely in preparation for pathogenicity tests on the isolates. The results of all four tests are summarized in Table 2 and Figs. 2 and 3.

Table 2.   Comparative growth rates (mm per day) at 20°C of Limburg 1983 Ophiostoma novo-ulmi subsp. americana, O. novo-ulmi subsp. novo-ulmi and hybrid groups
 Limburg isolatesControl isolates
subsp. novo-ulmisubsp. americanahybridsubsp. novo-ulmisubsp. americana
  1. aSee text for statistical significance levels.

  2. bnt, not tested.

Growth test 1
 No. of isolates1313141412
 Isolate range2·58–3·552·80–4·332·98–3·902·05–3·633·28–4·10
 Mean and SEa2·98 ± 0·093·41 ± 0·093·46 ± 0·092·92 ± 0·093·54 ± 0·09
Growth test 2
 No. of isolates231529ntbnt
 Isolate range2·00–3·582·53–4·432·60–3·80  
 Mean and SEa2·74 ± 0·093·25 ± 0·113·35 ± 0·08  
Growth test 3
 No. of isolates201618ntnt
 Mean and SEa2·96 ± 0·063·58 ± 0·063·45 ± 0·05   
Growth test 4
 No. of isolates141615ntnt
 Mean and SEa2·96 ± 0·053·33 ± 0·063·32 ± 0·05  
Figure 2.

 Growth rate 1. Distributions of Ophiostoma novo-ulmi subsp. americana and O. novo-ulmi subsp. novo-ulmi control groups and of the Limburg subsp. americana, subsp. novo-ulmi and hybrid groups at 20°C. SSNU, subsp. novo-ulmi; SSAM, subsp. americana. Heavy bar is the group mean.

Figure 3.

 Growth rate 2. Distributions of the Limburg subsp. americana, subsp. novo-ulmi and hybrid groups at 20°C. SSNU, subsp. novo-ulmi; SSAM, subsp. americana. Heavy bar is the group mean.

In Growth test 1, subsets of the Limburg subsp. americana, subsp. novo-ulmi and hybrid categories were selected at random and their growth rates compared with those of 12 Spanish subsp. americana control isolates and 14 Polish and Romanian subsp. novo-ulmi control isolates (from collections of >800 Spanish, 150 Polish and 150 Romanian isolates). As expected, because of instability generated by the up-mut factor, the mean growth rate of the novo-ulmi control group at 2·92 mm per day was significantly slower than that of the americana control group at 3·54 mm per day (< 0·001; Table 2, Fig. 2). However, the mean growth rate of the Limburg novo-ulmi group, at 2·98 mm per day, did not differ significantly from that of the novo-ulmi control group, i.e. this group conformed to the novo-ulmi‘type’. Likewise the mean of the Limburg americana group at 3·48 mm per day did not differ significantly from that of the americana control group. However, the mean growth rate of the Limburg hybrid group, at 3·46 mm per day was significantly faster than that of the Limburg novo-ulmi group (< 0·001), but not significantly different from that of the Limburg americana group. The behaviour of the Limburg hybrid group was therefore subsp. americana-like.

In Growth test 2 the growth rates of the Limburg subsp. americana, subsp. novo-ulmi and hybrid groups were re-examined using all 70 of the isolates assigned to these groups (Table 2, Fig. 3). Again, the mean growth rate of the Limburg hybrid group did not differ significantly from that of the Limburg americana group, but was significantly faster than the Limburg novo-ulmi group (< 0·001). Growth tests 3 and 4 also gave similar results (Table 2). The results for all four growth tests were therefore in close agreement.

Comparative pathogenicity of the hybrid and subspecies groups

The random subset of isolates used in Growth test 3 (less two subsp. americana isolates) was inoculated into 3-year-old clonal U. procera (moderately susceptible) and 5-year-old clonal U. × Commelin (moderately resistant) in June 1985, together with two control O. ulmi isolates. In addition, in the test conducted on U. procera, two sets of 14 subsp. americana × novo-ulmi F1 hybrid isolates generated from the laboratory crosses S173× P114 and C74× H327 were included for comparison.

The results on U. procera at 12 weeks after inoculation are shown in Table 3. Two O. ulmi control isolates caused 24·6% mean defoliation, two subsp. americana control isolates 86·7% mean defoliation and two subsp. novo-ulmi control isolates 85·3% mean defoliation (not shown). The Limburg americana, novo-ulmi and hybrid groups were all highly aggressive, each causing ca. 85% mean defoliation, and the means were not significantly different. The laboratory generated subsp. americana × novo-ulmi F1 hybrids caused 83 and 84% mean defoliation and were not significantly different from the three Limburg groups. Also, in neither progeny set did the mean of the F1 differ significantly from the parent mid-point value, i.e. there was no evidence of a negative fitness cost among the laboratory generated hybrids.

Table 3.   Pathogenicity of Limburg 1983 Ophiostoma novo-ulmi subsp. americana, subsp. novo-ulmi and hybrid groups and of in vitro generated subsp. americana × subsp. novo-ulmi F1 progenies
 Limburg isolatesIn vitro generated inter subspecies
F1 progenies
subsp. americanasubsp. novo-ulmihybridS173× P114aC74× H327b
  1. aParents, % defoliation: S173 84·8%, P114 82·8%, parent mid-point 83·8%. Progeny mean vs. parent mid-point not significant.

  2. bParents, % defoliation: C74 88·9%, H327 87·8%, parent mid-point 88·4%. Progeny mean vs. parent mid-point not significant.

  3. cSee text for statistical significance levels.

  4. dnt, not tested.

Ulmus procera (12 weeks)
 No. of isolates1414141414
Pathogenicity (% defoliation)
 Isolate range76·8 – 91·271·9 – 91·879·3 – 91·366·9 – 92·871·5 – 92·3
 Mean and SEc85·3 ± 1·484·9 ± 1·284·3 ± 1·083·4 ± 2·184·2 ± 1·6
Ulmus × Commelin (16 weeks)
 No. of isolates141414ntdnt
Pathogenicity (% defoliation)
 Isolate range43·1 – 73·633·0 – 67·941·1 – 73·8  
 Mean and SEc58·6 ± 2·853·1 ± 2·858·9 ± 2·8  

The results on the more resistant U. × Commelin after 16 weeks are shown in Table 3 and Fig. 4. Two O. ulmi control isolates caused 0·1% mean defoliation (data not shown). The mean pathogenicities of the Limburg americana and hybrid groups were 58·6% and 58·9%, respectively. Although the Limburg novo-ulmi group, at 53·1% mean defoliation, was slightly less pathogenic than the americana and hybrid groups, in the REML test there were no significant differences between groups.

Figure 4.

 Pathogenicity distributions of the Limburg subsp. americana, subsp. novo-ulmi and hybrid groups on the moderately resistant U. × Commelin after 16 weeks. SSNU, subsp. novo-ulmi; SSAM, subsp. americana. Heavy bar is the group mean.

RAPD analysis of the isolates

When PCR methodology became available a retrospective RAPD analysis was carried out on the 1983 Limburg sample. Seven oligonucleotide primers were identified that would discriminate between O. ulmi and O. novo-ulmi, and between a set of 11 subsp. americana and 13 subsp. novo-ulmi control isolates from a wide range of geographic locations (Table 1) on the basis of the presence and absence of major bands. These control isolates were selected from countries outside the known subspecies overlap zones, except for two isolates (H6 and H371) which were the first subsp. americana and subsp. novo-ulmi isolates to be found in the Netherlands (disease samples supplied by H. M. Heybroek, in 1972 and 1980, respectively). Control isolates of a subspecies selected from the same country were of a different vegetative compatibility type.

Thirty-six isolates from the 1983 Limburg sample were selected at random (every third isolate) and subject to RAPD analysis alongside the subsp. americana and novo-ulmi controls. The results are shown in Table 4. An interpretation of the RAPD pattern as americana-like, novo-ulmi-like or hybrid is shown in Table 5. For sixteen of the 36 Limburg sample isolates one polymerization attempt failed; for four isolates two polymerization attempts failed, and for one isolate three polymerization attempts failed. Since these failures were considered unlikely to significantly alter the results the interpretation was based on the pattern of the remaining primers. On this basis, 25 of the isolates, or 69%, were shown to be hybrid; five (14%) produced a subsp. americana-like profile and six (17%) a subsp. novo-ulmi-like profile.

Table 4.   Distribution of RAPDs markers among Ophiostoma novo-ulmi subsp. americana and O. novo-ulmi subsp. novo-ulmi control isolates and a random sample of the 1983 Limburg (Netherlands) population
IsolatesPrimer and fragment size (bp)
  1. aOpen circles, no fragment; closed circles, fragment present; nd, no data (primer failure).

Subsp. americana controls
Subsp. novo-ulmi controls
Limburg population
Table 5.   Interpretation of the evolutionary status of 36 Limburg province Ophiostoma novo-ulmi isolates based on RAPD and phenotypic characters
Isolate no.Interpretationa
RAPDPhenotypeRAPD and phenotype combined
  1. assa, subspecies americana-like; ssnu, subspecies novo-ulmi-like; H, recombinant or hybrid pattern; ?, status undetermined by phenotype; * polymerization failed with one or more oligonucleotides.


A comparison between the results of the RAPD analysis and the groupings of the isolates based on phenotypic characters is also shown in Table 5, together with an overall designation for each isolate on the basis of both RAPDs and phenotype. The proportion of isolates shown to be hybrids on the basis of the seven RAPD markers (69%) is considerably higher than the proportion revealed by the three phenotypic characters (33%). Indeed it is notable that all nine isolates that were designated ‘status undetermined’ on the basis of phenotype were shown to be hybrid on the basis of their RAPD profile. In addition seven isolates assigned to subsp. americana or novo-ulmi on the basis of phenotypic characters were shown to be hybrids by the RAPD analysis. Conversely, three isolates (H732, H786 and H798) assigned to subsp. americana on the basis of their RAPD profile with all seven primers appeared to be hybrids on the basis of their phenotypic characters.

Combining both the phenotypic and RAPD data (Table 5), it can be estimated that at least 28 isolates, or 78%, of the 1983 South Limburg population were hybrid. Among the remainder, five (14%) conformed closely to subsp. novo-ulmi and three (8%) to the subsp. americana.

RAPD characterization of a subspecies overlap population in Orvieto, Italy

The first subsp. novo-ulmi isolates to be identified (then termed the EAN race) were from samples collected for the senior author by H.S. McNabb in the Po Valley of Italy in 1973 (cf. Brasier, 1979). A further 200 samples collected by the senior author throughout most of Italy including Sicily in 1979 showed that, in addition to O. ulmi, both O. novo-ulmi subsp. americana and O. novo-ulmi subsp. novo-ulmi were already relatively widespread in the country (Brasier & Kirk, 2001; C.M. Brasier, unpublished observations). At some sample locations both subspecies were present (Fig. 1). Historically, therefore, parts of Italy were very early, and possibly the earliest, subspecies overlap zones in Europe. Within the 1979 survey a subset of 15 isolates from Orvieto, central Italy (Fig. 1), yielded seven subsp. americana and seven subsp. novo-ulmi isolates, together with a single O. ulmi isolate. Potentially, Orvieto was therefore another good location for investigating the possible emergence of subspecies hybrids.

In 1986 the Orvieto population was resampled and one hundred isolates collected. Eighteen isolates of O. novo-ulmi were selected at random and compared with three subsp. americana and three subsp. novo-ulmi control isolates for their RAPD profiles using the seven oligonucleotide primers (Table 6). Again there were polymerization failures mainly among the Orvieto sample isolates. Thirteen of the isolates, or ∼72% of the sample, were shown to be hybrids. Three isolates (17%) exhibited a subsp. americana-like RAPD pattern and two (11%) a subsp. novo-ulmi-like pattern. At least one of the isolates with a subsp. americana-like pattern (I229) was also a hybrid because, in separate phenotype tests, it produced the subsp. novo-ulmi fertility response and exhibited the novo-ulmi up-mut dimorphism.

Table 6.   Distribution of RAPD markers among a sample of Ophiostoma novo-ulmi isolates from a potential subsp. americana × subsp. novo-ulmi hybrid zone at Orvieto Italy, 1987
IsolatesPrimer and fragment size (bp)
RAPD interpretationa
  1. assa, subspecies americana pattern; ssnu, subspecies novo-ulmi pattern; H, recombinant or hybrid pattern; * polymerization failed with one or more oligonucleotides..

  2. bOpen circles, no fragment; closed circles, fragment present; nd, no data.

Subsp. americana controls (3 isolates)b 
Subsp. novo-ulmi controls (3 isolates) 

RAPD characterization of a subspecies novo-ulmi population in western Poland

One hundred and fifty disease samples were collected by the senior author across Poland in 1980. On the basis of their phenotypic characters, the O. novo-ulmi population comprised exclusively subsp. novo-ulmi (Brasier & Kirk, 2001). A few surviving pockets of O. ulmi occurred alongside O. novo-ulmi in the extreme east of the country.

The coastal area around Szczechin in north-west Poland is geographically close to areas of Denmark and Sweden that, by 1980, already had mixed populations of O. novo-ulmi subsp. americana and novo-ulmi (Fig. 1). A subset of 20 subsp. novo-ulmi isolates from this coastal area, encompassed by the triangle of Swinoujscie, Miedzyzdooje and Szczecin, was analysed for RAPD profiles alongside three subsp. americana and three novo-ulmi controls (Table 7). Seventeen of the isolates (85%) exhibited a typical subsp. novo-ulmi profile. However, one isolate, P143, exhibited subsp. americana markers with three of the seven oligonucleotides. Two other isolates, P141 and P146, exhibited a subsp. americana marker with one of the oligonucleotides. These three isolates therefore appeared to represent early introgression of subsp. americana DNA into the local subsp. novo-ulmi population.

Table 7.   Distribution of RAPD markers among a sample of Ophiostoma novo-ulmi subsp. novo-ulmi isolates from the Baltic Ports region of Poland in 1980
IsolatesPrimer and fragment size (bp)
RAPD interpretationa
  1. assnu, subspecies novo-ulmi; i, introgressant carrying subspecies americana DNA; H, hybrid or introgressant isolate.

  2. bOpen circles, no fragment; closed circles, fragment present; nd, no data.

Subsp. americana controls (3 isolates)b 
Subsp. novo-ulmi controls (3 isolates) 
Baltic Ports isolates
Budapest isolate P2046H

The holotype material used in the formal description of O. novo-ulmi subsp. novo-ulmi comes from the Szczecin area. It is comprised of the paired sexually compatible isolates P127 from Swinoujscie and P155 from Szczecin (IMI 313091; Brasier & Kirk, 2001). These two isolates were therefore included in the above RAPD analysis (Table 7) and both were found to exhibit the typical subsp. novo-ulmi pattern.

In an O. novo-ulmi population sampled by the senior author in Budapest, Hungary in 1989, 20 resulting O. novo-ulmi isolates were examined for phenotypic characters and 19 were assigned to subsp. novo-ulmi. A single isolate, H2046, exhibited a subsp. americana colony type and a subsp. novo-ulmi fertility response. When analysed for its RAPD profile, it exhibited six subsp. americana bands and a single subsp. novo-ulmi band (Table 7). H2046, therefore, appeared to be another example of introgression of subsp. americana DNA in a predominantly subsp. novo-ulmi population.


During the 1970s the two subsp. of O. novo-ulmi spread rapidly into western Europe, americana from the west and novo-ulmi from the east, bringing them into direct contact (Brasier & Kirk, 2001; Fig. 1), probably for the first time in their evolutionary history. The present study shows that, in two places in Europe where the subspecies overlapped relatively early, Limburg, Netherlands and Orvieto, Italy, subsp. americana × novo-ulmi hybrids already comprised 70–80% of the population by the early 1980s. Moreover, with only 10 distinguishing RAPD and phenotypic markers examined in total, it is likely that some of the isolates from these sites assigned here to subsp. americana or subsp. novo-ulmi were carrying undetected amounts of DNA of the ‘other’ subspecies. The present study also shows that introgression of some subsp. americana DNA into the subsp. novo-ulmi population of north west Poland was occurring by 1980. This suggests a low level introduction of americana × novo-ulmi hybrids or subsp. americana genotypes into the area prior to 1980, perhaps from the nearby mixed subspecies populations in Denmark or southern Sweden. In addition subsp. americana DNA had penetrated as far east as Budapest by 1989. Overall, it is clear therefore that the process of inter subspecies hybridization and introgression began very rapidly.

Ground surveys revealed other locations in central and western Europe where the two subspecies were overlapping including Belgium, Germany, Ireland, Denmark, Sweden and Switzerland (Fig. 1). The evidence from the present study, together with evidence for emerging subspecies hybrids in Austria (Konrad et al., 2002), suggests that hybrid swarms are likely to have developed at all such locations. Also, that introgression is likely to be occurring at many adjacent locations as a result of sporadic incursion of hybrid genotypes or genotypes of the ‘other’ subspecies. Through intercrosses between the hybrids and backcrosses between the hybrids and the parent subspecies, multiple expanding hybrid zones are likely to be developing across many parts of western and central Europe, leading to the disappearance of the ‘pure’ subspecies. However, unless an introduction of subsp. novo-ulmi occurs, a large population of ‘pure’ subsp. americana is likely to survive across North America through geographic isolation. Likewise, a large population of subsp. novo-ulmi may survive for a considerable time between eastern Europe and central Asia. Nonetheless, because this is a continuous land bridge, gradual eastward migration of hybrid genotypes into this region seems inevitable. Sudden jumps through human movement of diseased elm material are also likely.

During its spread into Europe in the 1970s, O. novo-ulmi (both subspecies) exhibited a remarkable evolutionary change from clonal to heterogeneous population structure at epidemic fronts as a result of its interaction with O. ulmi (Brasier, 1988; Brasier et al., 2004). Ophiostoma novo-ulmi is now going through another extraordinary change via the interaction of the two subspecies. That both events have occurred within only a few decades, between ca. 1940–1980, is a further indication of the rapid evolutionary developments that can occur in invasive pathogens as a consequence of episodic selection and horizontal gene flow (Brasier, 1995, 2000b, 2001). In the case of the O. novo-ulmi – O. ulmi interaction the resulting hybrids were rare and transient, the latter due to their very low fitness (Kile & Brasier, 1990; Brasier et al., 1998). Nonetheless, fixation of ‘useful’O. ulmi vic and Mat loci into the O. novo-ulmi genome occurred (Paoletti et al., 2006). In the case of the subspecies interaction however, hybrids are arising freely. This is probably for three main reasons. First, although the prezygotic barrier between the two subspecies results in a marked reduction in perithecial frequency, this is only when subsp. novo-ulmi is the female parent. Secondly, normal perithecial development and normal ascospore production occurs regardless of which subspecies is the female parent. Thirdly, the present study shows little or no fitness cost to the early hybrids, at least for growth rate and pathogenic fitness. Thus the Limburg isolates designated as hybrids on phenotypic characters were just as fast growing and as pathogenic as the control subsp. americana isolates; and were actually faster growing and at least as pathogenic as the control subsp. novo-ulmi isolates.

It could be argued that the Limburg hybrid isolates, having been isolated directly from the pathogenic phase O. novo-ulmi in elm xylem (Brasier, 1986a), consisted of genotypes that had already survived a degree of natural selection, i.e. were already selected for their pathogenic fitness from within the hybrid gene pool. However, random F1 hybrids generated via subsp. americana × novo-ulmi crosses in the laboratory were very similar to the Limburg hybrids in both growth rate and pathogenicity. Also, the growth rate and pathogenicity means of the progeny sets did not differ significantly from their parent mid-point values, i.e. there was no evidence of a negative interaction for either fitness character. This indicates that in nature hybrid progeny are likely to be just as fitted for growth and pathogenesis as their parent subspecies.

The hybrids may at some point need to be accorded taxonomic status in their own right, to distinguish them from the parent subspecies. In this regard, it is of interest to note that both of the isolates which comprised the holotype for subsp. novo-ulmi, P127 and P155 (Brasier & Kirk, 2001), came from the population sampled in the Polish Baltic Ports area. Both isolates exhibited a typical subsp. novo-ulmi RAPD pattern. In a separate analysis, Konrad et al. (2002) have shown these isolates also exhibit cu and col 1 locus polymorphisms characteristic of subsp. novo-ulmi. Nonetheless, in the present study two other isolates sampled from the same population showed evidence of introgression, exhibiting some subsp. americana RAPD loci. Another isolate, P150 from Szczecin (Table 7) was also examined by Konrad et al. (2002) and shown to exhibit an americana-like polymorphism at the col 1 locus. This indicates that it too has some americana DNA. There remains a possibility, therefore, that the two subsp. novo-ulmi holotype isolates, though phenotypically ‘true’ to subsp. novo-ulmi, contain small amounts of neutral americana DNA. It has already been demonstrated that most O. novo-ulmi isolates in Europe carry small amounts of O. ulmi DNA (introgressed vic and Mat loci), almost by definition (Paoletti et al., 2006). Together, these examples demonstrate a previously unappreciated risk when choosing holotype material in an invasive fungus, that of recent horizontal gene transfer. Without extensive genome sequencing it could be difficult to define what is truly ‘representative’ holotype material. Indeed, with O. novo-ulmi at least, a degree of introgression of ‘alien DNA’ would appear to be the norm.

Observations of hybrid swarms are very rare in fungal biology (Brasier, 2000b; Burnett, 2003). The current emergence of hybrids between the O. novo-ulmi subspecies therefore provides a rare opportunity to study the development of a fungal hybridization process; and, moreover, to study the future behaviour of Dutch elm disease in Europe. Ophiostoma novo-ulmi is effectively re-inventing itself. Since there are multiple phenotypic differences between the parent subspecies, most of which are of presumed adaptive significance, an obvious question is what phenotypes of O. novo-ulmi will survive in the hybrid zones in future. Will they resemble one subspecies more closely than the other? Will different phenotypes arise in different overlap locations or will a common successful phenotype emerge more generally? Will the hybrids be of largely neutral fitness relative to the parent subspecies, arising in proportion to the frequency of sexual recombination events alone?

Characters of particular interest for further study, in addition to growth rate and pathogenicity, include perithecial form, which is very different in the two subspecies; the frequency of the up-mut colony dimorphism, which was originally confined to subsp. novo-ulmi; the level of ceratoulmin ‘toxin’ production, which is lower in subsp. novo-ulmi; and the status of the subspecies fertility barrier, which may have a simple two locus control (Brasier, 1984; Brasier & Kirk, 2001). The samples from the Limburg and Orvieto areas described here were collected some 22–25 years ago. Many cycles of sexual recombination and natural selection will have occurred since, providing considerable opportunity for further change. In September 2008 the Limburg and Ovieto sites were resampled with a view to examining the current properties of the hybrid populations.


We are grateful to the late Adrian von Magdeburg and Hans Heybroek for collection of disease samples in the Netherlands and Andrew Jeeves for technical assistance. We also thank Geoff Morgan for assistance with statistical analysis and Konrad Heino and Thomas Kirisits for providing unpublished information.