Cryphonectria parasitica vegetative compatibility type analysis of populations in south-western France and northern Spain

Authors


*E-mail: robin@bordeaux.inra.fr

Abstract

A comprehensive study of the population biology of Cryphonectria parasitica, the causal agent of chestnut blight, is required to understand the spread of the epidemic in Europe and its natural regulation by the Cryphonectria hypovirus-1 (CHV-1). With this objective in mind, the diversity in vegetative compatibility (vc) types and mating types of C. parasitica populations was assessed in 43 chestnut sites in western France and northern Spain. Isolates were sampled in three regions along the Pyrénées mountains (Hautes Pyrénées, Pyrénées Atlantiques and Navarra) and in two regions north of the Pyrénées (Landes and Dordogne). There were 61 vc types observed in 682 isolates sampled. Ninety-five isolates (14%) could not be assigned to any one of the previously known European vc types. The finding of 47 incompatible vc types confirmed that the genetics of vegetative compatibility could not be totally accounted for by six diallelic vic genes. The two idiomorphs of the mating type gene were detected in all regions, indicating that sexual reproduction could occur within and between vc types. In all regions except Dordogne, C. parasitica populations were dominated either by the vc type EU-66 or EU-72. Neither vc type has been reported elsewhere in Europe, which suggests that two different introductions of C. parasitica have occurred in the study area. In Dordogne, populations were dominated by EU-33 and EU-2, and to a lesser extent EU-66 and EU-72. The low diversity in vc types for most of the C. parasitica populations provides good opportunities for natural regulation in forest coppices and for biocontrol with CHV-1 in orchards.

Introduction

Native plant pathogenic fungi which co-evolve with their hosts can cause little damage to the populations of plants they infect. However, the same pathogens may have a serious impact on newly introduced hosts or on native species where the pathogen is introduced (Parker & Gilbert, 2004; Desprez-Loustau et al., 2007). Cryphonectria parasitica, the causal agent of chestnut blight, is an example of such a biological invasion. It was accidentally introduced in North America, where it was first reported in 1904 in New York City on Castanea crenata and C. pumila trees imported from Japan (Anagnostakis, 1987). It was later introduced to Europe where it was first reported in Italy in 1938 on European chestnut (C. sativa) (Biraghi, 1946). By that date, the epidemic was already well established in this region, suggesting that the introduction had occurred much earlier (Grente, 1981). In France the report of the disease was officially announced in 1956 in Ardèche (south-eastern France), and C. parasitica was reported as early as 1947 in the Spanish Basque country, close to the French border (Grente, 1981; Heiniger & Rigling, 1994; Aguin et al., 2005). In North America, C. parasitica led to considerable destruction of the American chestnut (C. dentata) stands. However, in Europe, where the fungus spread in almost all regions where C. sativa was cultivated for fruit or as a forest tree, its impact on natural ecosystems proved to be low thanks to natural regulation of the disease. Spontaneous healings of chestnut blight cankers were observed and were attributed to the hypovirulence of C. parasitica (Heiniger & Rigling, 1994). Hypovirulence can be described as a disease of the fungus, caused by a double-stranded RNA hypovirus, which results in the attenuation of fungal pathogenesis (Nuss, 1992; Hillman et al., 2000).

This virus-fungus interaction provided the rationale for the development of a biological control strategy, deployed first in France during the 1970s to decrease chestnut blight severity (Grente & Berthelay-Sauret, 1978; Heiniger & Rigling, 1994; Milgroom & Cortesi, 2004). Hypoviruses of Cryphonectria are transmitted between fungal strains via hyphal anastomosis. The probability of hypovirus transmission between two fungal isolates in vitro depends on the number and nature of vegetative incompatibility (vic) genes shared by these isolates (Liu & Milgroom, 1996; Cortesi et al., 2001). The virus transmission occurs at 100% frequency between isolates sharing the same alleles at the six vic loci, i.e. between isolates belonging to the same vegetative compatibility (vc) type. By contrast, heteroallelism at vic loci strongly decreases the rate of transmission (Cortesi et al., 2001; Papazova-Anakieva et al., 2008). Consequently, diversity of vc types is believed to be a major factor for restricting the spread of the hypoviruses in C. parasitica populations and affecting the success of chestnut blight biological control (Heiniger & Rigling, 1994; Liu & Milgroom, 1996). However, this hypothesis should be revisited given the results of a genetic analysis of the viruses of two fungal populations suggesting that their movement between vc types within a population is not as restricted as previously thought (Carbone et al., 2004).

Although biological control was first proposed and performed in Europe in the 1960s, only during the last ten years had a clearer picture of diversity and differentiation of European C. parasitica populations emerged from several studies using vc types as diversity markers (Bissegger et al., 1997; Cortesi & Milgroom, 1998; Robin et al., 2000; Robin & Heiniger, 2001; Trestic et al., 2001; Sotirovski et al., 2004; Perlerou & Diamandis, 2006; Krstin et al., 2008). A common characteristic for all these European C. parasitica populations, especially from the Mediterranean and Balkan regions, is their low diversity in vc types compared to North American populations. This provided a plausible explanation for the failure of biological control attempts in North America and the successful spread of Cryphonectria hypovirus-1 in Europe (Anagnostakis et al., 1986; Heiniger & Rigling, 1994; Robin & Heiniger, 2001). Moreover, these studies also showed the strong geographic differentiation of European C. parasitica populations and led to the hypothesis of multiple introductions of this pathogen in Europe. Populations in Italy, eastern France, Switzerland, Austria, north-eastern Spain and Germany, which are dominated by vc types EU-1, EU-2 and EU-5, would be the result of a first introduction event (Cortesi et al., 1998; Robin et al., 2000; Robin & Heiniger, 2001). In south-western Europe (western France, western Spain and Portugal) other dominant vc types (EU-1, EU-11, EU-33, EU-66, EU-72) have been identified (Robin et al., 2000; Aguin et al., 2005; Braganca et al., 2007; Zamora et al., 2008). In the eastern Mediterranean and Balkan countries, more recently invaded by C. parasitica (Bosnia Herzegovina, Greece, Macedonia, Slovakia), populations are dominated by the EU-12 vc type (Trestic et al., 2001; Sotirovski et al., 2004; Juhasova et al., 2005; Adamcikova et al., 2006; Perlerou & Diamandis, 2006).

Understanding how C. parasitica has invaded the European continent is of crucial interest for protecting chestnut stands against this invasive pathogen. The objective of this study was to assess the diversity and distribution of vc types and mating types in C. parasitica populations from western France and north central Spain, at a local and regional scale. The pattern of distribution of vc types of C. parasitica in south-western Europe could provide support for the hypothesis of different introductions in Europe. An accurate knowledge of vc type and mating type diversity is also required to develop chestnut blight biological control programmes in these regions and to make predictions about the natural regulation of C. parasitica populations by CHV-1.

Material and methods

Study sites and sampling design

In France, C. parasitica was sampled from 19 populations within chestnut forest sites (approximately 1 ha), distributed in four regions. Two regions were located along the Pyrénées mountains (seven sites in the region Hautes Pyrénées and six in Pyrénées Atlantiques, Fig. 1, Table 1). The other two regions (the Landes and the Dordogne, with three chestnut sites in both regions), were located north of the Pyrénées. Minimal and maximal distances between sites within regions were 6 and 99 km, respectively. The number of cankers sampled per site varied from 6 (site 6) to 46 (site 2), with an average of 24 isolates per population. In a fifth region located in north central Spain (Navarra), the sampling scheme was slightly different. A 4 × 4 km grid was superimposed on the vegetation map of Navarra. This grid was obtained by densifying the l6 × 16 km Level 1 monitoring grid established by the International Co-operative Programme on Assessment and Monitoring of Air Pollution Effects on Forests (ICP-Forests) operating under the United Nations Economic Commission for Europe (UNECE). Each node with presence of chestnut (24 in total) was considered a population. Eleven populations were located in orchards (five were old orchards still in production and seven were abandoned), ten within high forests and three within coppice forests. The number of cankers sampled per site was 10, with an average of nine successful isolates per population (ranging from two in site 43 to 18 in site 20).

Figure 1.

Diversity in vegetative compatibility (vc) types of Cryphonectria parasitica in five geographical regions. Sobs: number of observed vc types; SICvct: number of vc types incompatible with European (EU) vc type testers EU-1 to -74; pie charts represent the frequency of vc types EU-66, EU-72, of other European vc types and of IC vc types in five regions. Populations studied are represented by black dots for populations with a sampling size > 14, and white dots for the others.

Table 1.  Diversity of Cryphonectria parasitica populations in vegetative compatibility types and mating types
RegionPopulationCodeNaVegetative compatibility typesMating typesg
Richness and evennessPercentage of isolates inf:MAT1-1MAT1-2MAT1-1/2
SobsbS15cSICvctdE5eEU-66EU-72EU vctIC vct
  • a

    Number of isolates.

  • b

    Sobs: observed number of vegetative compatibility (vc) types.

  • c

    S15: richness of vc types estimated by rarefaction analysis with 15 isolate samples (only for populations with 15 or more isolates).

  • d

    SICvct: number of IC vc types (incompatible with European vc type testers EU-1 to -74).

  • e

    E5: evenness index (see Material and methods).

  • f

    EU vct: European vc types, other than EU-66 and EU-72.

  • g

    A subsample of isolates in all regions were assayed for mating types: numbers of isolates observed for each mating type and MAT heterokaryons are given for each population.

DordogneDoissat 136 85·140·7519367117234
Mazeyrolles 246117·430·96171352179113
Moncléra 321 75·350·455 712423 
LandesBelin-Beliet 441 63·740·477610 155212
Luc 524 65·140·651754 297141
Mimbast 66 3 1 2050 30   
Hautes PyrénéesArné 724 21·610·46096 41153
Bordes 843 74·150·45274 247111
Frechendets 919 32·820·54 84 169 1
Monpezat1020 54·530·591065 25152
Oroix1123 85·860·46465 3034 
Pontacq1217 43·630·48 82 1821 
Uzos1319 21·810·48 95 5 1 
Pyrénées AtlantiquesAldudes1431 63·950·5668  3211125
Banca1520 21·710·4195  51  
Barcus1619 21·810·48 95 5 1 
Bunus1724 53·930·692958 13962
Sare1823 65·140·67529 39853
Lantabat197 2   6040  221
NavarraLabaien2018 22·000·698218  12 
Lintzoain2115 22·000·6287 13  3 
Valcarlos2212 2   8218  12 
Donamaria2311 3 1 87 13    
Zilbeti2411 2   67 33 15 
Aitziano2510 1   36 2736   
Basaburua2610 1   91 9  2 
Faceria2710 2   100    1 
Leitza2810 2   100      
Lesaka2910 2   80 20 21 
Sunbilla3010 2   6040     
Urrotz3110 2 1 9010     
Arantza329 3   50 50    
Arbea339 1   90  10   
Artesiaga349 2 1 78 22 12 
Bco. Aritzakun359 1   100    1 
Larraun369 2   67  33 1 
Almandotz378 1   100      
Araitz388 1   6733     
Doneztebe397 1   100   12 
Erratzu407 2   100    1 
Eugi417 1   100      
Igantzi423 3 1 0 6633   
Arizkun432 1   100      

At all sites in both countries, the minimal distance between two sampled trees was 5 m, and only one canker was sampled per tree. All types of cankers were sampled, including healing or active, but active cankers were preferred in order to have the highest probability of isolating C. parasitica. Four bark samples were removed in each canker with a cork borer and stored at 5°C for C. parasitica isolation.

C. parasitica isolation and vc type determination

Isolation was performed following Robin et al. (2000) and only one isolate per canker was randomly selected. Isolates were grown on PDA (39 g L−1, potato dextrose agar, Difco Laboratories). Determination of vc type was carried out as described by Robin et al. (2000). European vc type testers from EU-1 to EU-74 were used (Cortesi & Milgroom, 1998; Cortesi et al., 1998; Robin et al., 2000). In 27 populations (populations 4, 17, 18 and 20–43) the diversity of vc types was determined by first pairing all isolates from each population in all combinations and secondly by pairing one isolate per vc type with the different European testers. For the other populations, all isolates were first paired against the vc types dominating in populations 4, 17 and 18 (EU-72, EU-66, EU-33, EU-2). Within each population, the isolates which were incompatible with the dominant testers were paired in all possible combinations, and one isolate per vc type was then paired to all the other European testers. Pairings were repeated several times to confirm compatibility.

Mating type determination

All together, 247 C. parasitica isolates from France and 33 from Spain were assayed for mating type using the PCR method (McGuire et al., 2004). Primers M1-GS3-rev (5′-CAGATGTCAACGGCCTTCAGGCCAGGA-3′) and M1GS1n (5′-TATGACAACGACGTCGAAGAATCAGAGTG-3′) were used to amplify the MAT1-1 idiomorph and primers gsl-d-l (5′-CTCCCGATGGATTGGGGAAGATAATGGGC-3′) and M2-GS3 (5′-TTCAACCTGTCCAAGACTGTAGCCTTCG-3′) to amplify the MAT1-2 idiomorph. Methods for DNA extraction and for PCR followed those previously reported by Breuillin et al. (2006) and Marra & Milgroom (2001), respectively.

Data analysis

The analysis was performed at two levels. At the population level, the richness component of vc type diversity was analysed, assessed by the number of observed vc types, Sobs. However, because sampling size of populations varied from 2 to 61, a computer-sampling algorithm of rarefaction was used (EcoSim7 program, Gotelli & Entsminger, 2001). Fifteen isolates were randomly drawn from each population having 15 or more isolates (Table 1). The process was repeated 10 000 times to calculate the mean of vc type richness (S15) and analysis of variance was performed to compare the estimated vc type richness in each region. The evenness component, E5, was estimated for all populations represented by 15 or more isolates using the Shannon index (HS) and Stoddart and Taylor's G statistic (Grunwald et al., 2003):

E5 = (G − 1)/(exp(HS) − 1), HS = −∑ pi Log(pi) and G = 1/∑pi2

where pi is the frequency of the ith vc type. At the regional level, all sites and all isolates were included in the analysis and a χ2 test was performed to compare vc type and mating type frequencies among regions. Differences in the estimated richness of vc types and the evenness index among regions were tested with one-way anova with region as a fixed factor and populations as repetitions. All statistical analyses (analysis of variance, coefficient of correlation, χ2 test) were performed with SAS/Stat Software (SAS Institute).

Results

Diversity in vc types and mating types of C. parasitica populations

In all populations, except those in Dordogne, more than 50% of the isolates belonged to one vc type, which was either EU-66 or EU-72 (Table 1). The remaining isolates belonged to other European vc types or to vc types which were incompatible with any one of the known European vc types (called IC types). In Dordogne, vc types EU-66 and EU-72 were represented by less than 19% of the isolates in the three populations studied. The dominant vc types (dominant defined as a vc type observed in at least one third of the isolates) were EU-33 (populations 1 and 3) and EU-2 (population 1). In population 2, six vc types (EU-2, EU-14, EU-33, EU-66, EU-72 and an IC vc type) were each observed in a range of 13 to 17% of the isolates. The richness S15 was estimated for the 19 populations with a sampling size of 15 or larger (Table 1). This parameter varied from 1·6 (population 7) to 7·4 (population 2), with a mean value of 3·7 (standard deviation 1·7). It was significantly correlated with the sample size (RPearson= 0·548, P = 0·0151 with df = 17). There was no significant difference among regions (F = 3·02, P = 0·0545).

The evenness index, E5, of these 19 populations varied from 0·41 (population 15 in Pyrénées Atlantiques) to 0·96 (population 2 in Dordogne), with a mean value of 0·57 (standard deviation 0·14). This index was not correlated to sample size (RPearson = 0·336, P = 0·16 with df = 17) and there were no significant differences of evenness among regions (F = 1·91, P = 0·16).

When at least 10 isolates per population were assayed for mating types (12 sites), both MAT idiomorphs were detected within populations (Table 1). Moreover, in these populations, the PCR assay revealed the presence of isolates in which both idiomorphs were amplified (Table 1). Single spore isolates obtained from three of these suspected mating type heterokaryons were homokaryotic (results not shown). The mating type ratios were tested for deviation from 1/1 in the 12 populations. The ratio was significantly different from 1/1 in two populations (4 and 9), the idiomorph MAT1-2 being in excess. When two to 10 isolates were assayed per population (14 sites), both MAT idiomorphs were detected in 10 populations (Table 1).

Distribution of vc types and mating types at the regional level

When all populations were pooled, the 682 C. parasitica isolates studied could be distributed among 61 vc types. Fourteen vc types were compatible with European testers. The vc types EU-66 and EU-72 comprised 43·3 and 29·8% of all isolates, respectively. Five EU vc types (EU-2, EU-33, EU-31, EU-45 and EU-69) comprised from 0·3 to 5·1% of the isolates, whereas seven EU vc types (EU-5, EU-13, EU-14, EU-22, EU-29, EU-36 and EU-42) were single-isolate vc types (i.e. represented by a single isolate). The remaining 95 isolates were assigned to IC vc types. Thirteen IC vc types were ‘multi-isolate’ types, i.e. were represented by several isolates. Six of these were detected in several populations with a frequency varying from 0·3% (two vc types, one in populations 4 and 11 and the other in populations 10 and 14) to 3·7% (one vc type in populations 2, 4, 5, 6, 10 14 and 18), and the other three in only one population with a 0·3% frequency. The remaining 34 IC types were single-isolate vc types.

The diversity of vc types identified in each region ranged from 8 to 21 and the evenness from 0·327 to 0·782 (Fig. 1, Table 2). The frequency of the vc types EU-72, EU-66, EU-33 and EU-2, of the ‘other’ European vc types (pooled into a single category) and of the IC types (pooled into a second single category) were significantly different among regions (P < 0·0001). EU-66 was the dominant vc type in Navarra (81% of the isolates, Fig. 1). In Hautes Pyrénées, 79% of the isolates belonged to vc type EU-72. Cryphonectria parasitica populations in the Pyrénées Atlantiques and in the Landes were characterized by the same evenness index (Table 2) and similar percentages of isolates belonging to the vc type EU-66 (59 and 51%), EU-72 (30 and 28%) and to IC types (19 and 21%).

Table 2.  Diversity in vegetative compatibility (vc) and mating types of Cryphonectria parasitica in the five study regions
 DordogneLandesHautes PyrénéesPyrénées AtlantiquesNavarra
  • a

    Number of isolates.

  • b

    Observed number of vegetative compatibility types.

  • c

    Evenness index (see Material and methods).

  • d

    A subsample of isolates (N) in all regions were assayed for mating types: numbers of isolates observed for each mating type and MAT heterokaryons are given.

  • e

    χ2: test of the mating-type ratio for deviation from a ratio of 1/1 (NS: the H0 hypothesis is not rejected,

  • *

    : the hypothesis is rejected at a significance level of P ≤ 0·05).

Vc types
 Na10371165125224
 Sobsb 21 9 20 16  8
 E5c  0·782 0·636  0·327  0·630  0·422
Mating typesd
 N 6250 67 68 33
 MAT1-1 1812 33 31  7
 MAT1-2 37 5 27 26 26
 MAT1-1/2  7 3  7 11 
 χ2e 2·69 5·29 24·36 15·84  5·94
 NS*NSNS*

The Dordogne was the region exhibiting the highest diversity of vc types (Table 2) with 61% of isolates belonging to EU vc types other than EU-66 and EU-72. Among these vc types the most frequent were EU-33 (34%), EU-2 (18%) and EU-14 (6% of isolates).

Among the 280 isolates assayed for mating types, 101 isolates were MAT1-1 and 151 were MAT1-2. The mating type ratio was significantly different from 1/1 in the Landes and in Navarra. Twenty-eight isolates were suspected MAT heterokaryons (Table 2) and they belonged to eight vc types (10 isolates in EU-66 and 12 in EU-72). In each region, both mating types were detected in the dominant vc types.

Discussion

A most striking result of this study is the high number of vc types (47) which were incompatible with all previously reported EU testers. Thirteen of them have been detected in different sites or in different trees, indicating that this incompatibility phenotype was stable and likely to be the result of genetic determinism and not of a cultural artefact. However, isolation of single-spore strains for the single-isolate IC types would be necessary before updating the list of European EU testers. The occurrence in France of 12 vc types (EU-65 to EU-74) which did not belong to any of the described vic genotypes (Robin et al., 2000) and the high diversity of vc types in Asia (Wang et al., 1991; Liu & Milgroom, 2007) have already substantiated that the genetics of vegetative compatibility could not be totally accounted for by six diallelic vic genes. In China and Japan, 131 and 71 vc types were reported, respectively. More interestingly in Japan, 68 vc types had unknown genotypes indicating that at least two additional vic genes must be polymorphic to result in the observed diversity in vc types (Liu & Milgroom, 2007). The current results not only confirm these previous findings but also emphasize the qualitative and quantitative importance of these ‘new’ vc types. These stayed unnoticed because vc type diversity of C. parasitica populations in Western Europe has received little attention until now, in spite of the report of chestnut blight in northern Spain and France early on in the European epidemic (Heiniger & Rigling, 1994) and in spite of the economic and social importance of chestnut in these regions.

The second notable result of this study is the high frequency of two vc types, which define distinct populations in the study area. Near the Pyrénées mountains the vc types EU-66 (in Navarra and Pyrénées Atlantiques) and EU-72 (in Hautes Pyrénées) were dominant whereas in Dordogne and Landes other dominant vc types were also occurring. The dominance of vc types EU-66 and EU-72, previously only reported in Dordogne (Robin et al. 2000), indicates that the vic gene(s) only recently identified in the European population are occurring at a high frequency in this area. From this pattern, a likely scenario for the invasion of the study area by C. parasitica arises. Two different introductions might have occurred: a strain belonging to EU-66 could have been introduced somewhere between Spain (Navarra) and France (Pyrénées Atlantiques) and an EU-72 strain introduced further inland in Hautes Pyrénées. These strains have not only established within their region of introduction but have dispersed further north in Landes and the Dordogne. Under this scenario, the introduction of C. parasitica in Europe would have occurred earlier than previously suggested and directly from Japan or China. This hypothesis is supported by the C. parasitica-life cycle and recent chestnut history. The saprophytic, endophytic and parasitic stages of C. parasitica allow for long-distance dispersal, from one continent to another. This long-distance dispersal is mainly human mediated since the pathogen can be transported and stay unnoticed on several types of plant material (Bissegger & Sieber, 1994; Prospero et al., 2006). Although the first official report of C. parasitica in Europe was in 1938, Darpoux (1949) suggested that introduction of the pathogen in Europe could date back to the First World War with infected wood. Infected Asian plants imported to Italy in the 1920s might also be responsible for the epidemic (Heiniger & Rigling, 1994). In the northern coast of Spain near Navarra in 1949, 90% of the C. crenata trees in a 20-year-old plantation were infected (Darpoux, 1949). It is well documented that there were several imports of chestnut germplasm from China and Japan to France to provide alternative chestnut trees, since indigenous trees were devastated by ink disease. In particular, numerous nuts of C. crenata and C. mollissima were introduced in south-western France (Pyrénées Atlantiques) in the beginning of the 20th century (Darpoux, 1949). Since nuts are possible vectors of the pathogen (Collins, 1913), the introduction of C. parasitica on this material cannot be ruled out. For this reason, these imports were stopped by 1925. Both C. crenata and C. mollissima are highly resistant to chestnut blight, and in this area where ink disease was prevailing on C. sativa trees, the presence of C. parasitica could have gone unnoticed for some time.

Cryphonectria parasitica populations which established following these introductions in the Pyrénées and in Landes regions were characterized by a low evenness of vc types, similar to the evenness of populations from Macedonia (Sotirovski et al., 2004) but had a higher richness due to the presence of several single isolate vc types. The genetic diversity of these populations and, therefore, their clonality cannot be directly inferred from their vc type diversity (Liu & Milgroom, 1996). However, it can be assumed that the dominant vc types correspond to different clonal lineages. Indeed, comparisons of microsatellite analyses and vc type analyses have shown that vc type diversity is a good predictor of the overall genetic diversity of C. parasitica populations, even if microsatellite markers had a higher discrimination power (Breuillin et al., 2006). Clonality can result from a founder effect in newly established populations exhibiting a low genetic diversity and only one mating type. Such an effect was observed in Switzerland and in the Balkans (Hoegger et al., 2000; Sotirovski et al., 2004). However, in the present study area C. parasitica populations are probably much older, the two mating types were well represented and the mating type ratio was significantly different from 1/1 in only two regions. Crossing between isolates from different vc types is thus theoretically possible and should increase the diversity in vc types of these populations. However, this increase might be low if parental vc types differed by only one or two vic genes. Moreover, crosses between dominant and rare vc types might occur at low frequency whereas crosses within dominant vc types could contribute to reinforce the dominance of these lineages. Presence of mating type heterokaryons, as already reported in American populations (McGuire et al., 2004) and autofecundation could also result in the amplification of one lineage within populations. Another explanation for the dominance of some vc types could be that recombinant vc types are counter selected compared to the dominant lineages and could not successfully establish. Although not yet demonstrated, coevolution of C. parasitica with its host and/or its hypovirus could have resulted in significant local adaptation of some dominant lineages. Genetic variation in chestnut blight resistance has been reported in C. sativa (Bazzigher & Miller, 1991) and previous results have revealed significant interactions between CHV-1 and fungal isolates (Peever et al., 1998).

Another factor which can reduce sexual reproduction in the current study populations is the CHV-1, which causes female sterility and a decrease in the number of conidia produced which serve as male gametes in C. parasitica (Nuss, 1992). This effect could be significant as the dominance of a few vc types could have favoured the dissemination of CHV-1 in populations. Hypovirulent strains were isolated in the Pyrénées in 1964 (Grente, 1981) and CHV-1 has been detected in several populations of the study area (results not shown). To date the average percentage of infected trees per chestnut site is 67% in Hautes Pyrénées and 51% in Pyrénées Atlantiques (de Villebonne, 1998) but the severity of the disease is low with no dieback or mortality. This could be due to the successful establishment of the hypovirus. This natural attenuation of disease severity in the forest coppices from the study regions and the vc type structure of C. parasitica populations are good arguments to deploy hypovirulent isolates in orchards or high value forest stands.

The diversity in vc types of Dordogne populations was notable both in terms of richness and evenness, and is similar to those of three American populations (Sotirovski et al., 2004). The assemblage of the Pyrenean types EU-66 and EU-72, the vc type EU-2 which is one of the dominant vc types from eastern France, north-eastern Spain, Italy and Switzerland populations (Cortesi et al., 1998; Robin et al., 2000; Robin & Heiniger, 2001) and the vc type EU-33 suggests that different introduced populations mixed in this area. In Dordogne, chestnut is also cultivated for fruit. It is likely that transport of plants or grafted varieties have resulted in introductions from other regions. It was not possible to determine if this mixture of isolates has produced new vc types by recombination. Complementary studies with microsatellite markers are required to fully understand the genetic makeup of C. parasitica in this region and its invasion process in Europe.

Acknowledgements

This work was supported in part by grant ITT-CTP2000-4 from CTP (Pyrénées Work Community). We would like to thank S. Schafleitner, O. Fabreguettes and M. Gómez for assistance with C. parasitica sampling and typing, and C. Dutech for helpful discussions. We also thank C. R. Fischer for the English revision of the manuscript.

Ancillary